∗

News Momentum

Hao Jiang†, Sophia Zhengzi Li‡, and Hao Wang§

This Draft: February 2018

Abstract
We decompose daily stock returns into news- and non-news-driven components, using a
comprehensive sample of intraday firm-level news arrivals matched with high-frequency
movements of their stock prices. We find that, consistent with prior literature, non-
news returns precede a reversal. For news-driven returns, however, we find strong
evidence of return continuation without subsequent reversals. A strategy of news mo-
mentum that buys stocks with high news returns and sells stocks with low news returns
generates an annualized return of 40.08% in the following week, with a four-factor al-
pha of 40.44%, controlling for the market, size, value, and momentum. The strategy’s
profitability is driven by positive serial correlations in individual stock returns, and is
particularly pronounced for overnight and weekend news and among small firms with
low analyst coverage, high volatility, and low liquidity. These results suggest that in-
vestor under-reaction to news, coupled with limits to arbitrage, drives news momentum.

JEL classification: G02; G10; G14

Keywords: News; Momentum; Reversal; Underreaction; Attention; Limits to Arbi-

trage.

∗
We are grateful to Ren-Raw Chen, Charles Hadlock, and Stefan Nagel, along with seminar participants
at Michigan State University, Rutgers University, and University of Wisconsin-Madison, and conference
participants at the Conference on Financial Economics and Accounting for their helpful comments.
†
Eli Broad College of Business, Michigan State University, East Lansing, MI 48824; E-mail:
[email protected].
‡
Rutgers Business School, Piscataway, NJ 08854; E-mail: [email protected].
§
Prime Quantitative Research LLC, Piscataway, NJ 08854; E-mail: [email protected].
1 Introduction

The extensive literature on return predictability has established an interesting array of facts
regarding the dynamics of individual stock returns. In particular, whereas short-horizon
stock returns within the past month and long-horizon returns in the past 3–5 years exhibit
reversals, returns in the period of 3–12 months show a pattern of continuation in the sub-
sequent 3–12 months. This finding on the stock price momentum has received widespread
attention, and generated substantial controversy among financial economists regarding its
implications for market efficiency (see Jegadeesh and Titman (2011) for a recent survey).
Underlying this controversy is the joint-hypothesis problem highlighted by Fama (1970),
which states that tests of market efficiency are inherently tied to tests of specific asset pricing
models. It is therefore difficult to draw a clear inference from apparently anomalous price
behavior regarding market efficiency. A powerful solution to this problem is to focus on the
behavior of stock prices in a short time window, during which expected returns on individual
stocks are small, so that the results are not particularly sensitive to the choice of specific
asset pricing models (Fama, 1998).
In this paper, we exploit this insight to contribute to the literature on return predictabil-
ity. Specifically, we combine a comprehensive sample of time-stamped firm-level public news
announcements with high-frequency (e.g., within a 15-minute time interval) price movements
of individual stocks, to identify the very short-term response of individual stocks to firm-
specific information events. In this way, we decompose daily stock returns into news-driven
and non-news-driven components, revisiting the issue of short-term return predictability.
Our results indicate that whereas non-news-driven return precedes a reversal, news-driven
return tends to exhibit a strong pattern of continuation. For instance, from 2000 to 2012, a
strategy of news momentum that buys stocks with high news returns and sells stocks with
low news returns in the previous day with a one-week holding period generates an average
annualized return of 40.08%, with a four-factor alpha of 40.44%, controlling for the market,

1
size, value, and momentum.1
As implied by the similar magnitudes of average returns and four-factor alpha, the news
momentum strategy has small exposures to pervasive factors such as market, size, and value
factors: The absolute value of the factor loadings equals to or below 0.06. News momentum
exhibits a moderate correlation with the Jegadeesh and Titman (1993) medium-term price
momentum: The loading on the momentum factor is 0.06 in the four-factor model and the
univariate correlation is 0.11. Unlike the price momentum strategy, which is characterized
by high crash risk (a negative skewness of -0.79 and maximum drawdown of 63.66%), news
momentum has a positive skewness of 0.51 and maximum drawdown of only 15.92%. Using
the Fama-MacBeth (1973) cross-sectional regressions, we find similar evidence of news return
continuation and non-news return reversals.
It should be noted that our research design focuses on the post-news-arrival return pat-
terns, in contrast to studies that emphasize the anticipation of important economic news as
a source of risk, driving stock prices. In particular, that line of research, such as Savor and
Wilson (2013, 2016) and Lucca and Moench (2015), documents a large unconditional return
premium on days with important news arrivals, which could reflect the compensation for
bearing risk associated with the economic news. Our approach instead examines the differ-
ence in returns among firms with news stories following the initial market reaction, which is
less likely to reflect risk premiums for information uncertainty.
To better understand the source of the news momentum, we follow Lo and MacKinlay
(1990), decomposing the expected news momentum profit into three components: the average
autocovariance of individual stock returns, the average cross-autocovariance across stocks,
and the cross-sectional variance in expected stock returns. The first component captures the
serial correlation in individual stock returns; a positive value would imply positive average
returns to a strategy that buys winners and sells losers conditional on news arrivals. The
second component reflects the lead-lag effects across stocks; a positive value would imply
1
We also use a multifactor model that includes the Fama and French (2015) five-factor model, the
momentum factor, and the short-term return reversal factor. The annualized alpha of our news momentum
strategy is 40.03%.

2
negative average returns to our news momentum strategy. The third component measures
the dispersion in expected returns across stocks; if firms with positive news on average have
higher expected returns, the news momentum strategy may be profitable due to the difference
in expected individual stock returns. Our decomposition results indicate that the profit of
the news momentum strategy comes almost entirely from the positive autocovariance of
individual stocks. One limitation of this return decomposition strategy is its requirement
that firms must be in existence over our full sample period, which could introduce a certain
degree of survivorship bias. To mitigate this bias, we also explore an industry-level news
momentum strategy, which generates a similar decomposition result.
What leads to the strong positive serial correlation in individual stock returns after the
news arrival? The behavioral finance literature offers two possible interpretations: underre-
action and delayed overreaction. When we extend the post-event holding horizon of the news
momentum strategy, we find that the news momentum effect appears to persist, even one
year after news arrival, without reversals. The evidence therefore tilts toward stock market
underreaction as a possible driving force of news momentum.
To shed further light on news momentum, we examine the heterogeneity of its effects
across firms. In particular, we find that the news momentum effect tends to be stronger a-
mong firms that are smaller, receive less analyst coverage, exhibit more volatile stock prices,
and have less liquidity. These results are consistent with attention-based interpretations,
according to which less visible firms tend to receive less attention from the investment com-
munity, which results in underreaction to public news. These results are also consistent with
the argument of Shleifer and Vishny (1997) that mispricing is likely to be more pronounced
among stocks with higher limits to arbitrage.
Digging into the pattern of return continuation following news arrival, we examine the
high-frequency response of stock prices to overnight news, which constitutes more than half
of our sample. To measure the initial response of stock markets to news that arrives anytime
after 4:00 p.m. of day t − 1 and before 9:45 a.m. of day t, we compute overnight returns as

3
the percentage change in prices from market close as of 4:00 p.m. of day t − 1 to 9:45 a.m.
of day t. Then we examine the adjustment of stock prices in every half-hour interval from
10:00 a.m. of day t until market close of day t + 4, by tracking the cumulative returns of
the news momentum strategy over these five trading days. The results indicate a gradual
drift in stock prices following the initial response to overnight news. Specifically, the news
momentum return on average drifts up to 25 basis points (bps) at 3:30 p.m. and then shows
a slight dip to 17 bps at 4:00 p.m. of day t. When the market opens on day t + 1, the
momentum profit jumps to 59 bps. The post-news announcement return drift occurs in a
repeated pattern over the subsequent days, and the news momentum return reaches 104 bps
at the market close of day t + 4.2 This detailed account of the stock market response to
overnight news provides further support for investor underreaction to public news.
We perform several robustness tests. First, instead of using transaction prices from the
Trade and Quote (TAQ) database to compute intraday returns, we consider daily stock
returns from the Center for Research in Security Prices (CRSP) database to compute news-
driven returns. This method loses the advantage of finely decomposing daily stock returns
and obtains a noisy measure of news-driven returns. We find that a similar news momentum
strategy generates an annualized average return of 23.88% and a four-factor alpha of 23.76%.
This result illustrates both the gain in precision of measuring news returns using high-
frequency data (approximately 68% increase in news momentum profits and 94% increase
in Sharpe ratio) and the robustness of the news momentum effect. Second, we consider
mid-quote prices instead of transaction prices to compute returns, based on which a news
momentum strategy generates similar and slightly stronger performance. Third, instead of
computing news returns and holding-period returns using close-to-close transaction prices, we
consider returns based on open-to-open prices and observe an even stronger news momentum
effect. Fourth, we compute firm-specific news-driven returns by subtracting from the overall
2
This effect is unlikely to be driven by the periodicity pattern reported by Heston et al. (2010); whereas
the overnight non-news returns exhibit a pattern of continuation in the subsequent overnight period, they
are followed by an even stronger reversal during the next day, with the net effect of a reversal. Section 5
compares in detail the news momentum and the periodicity pattern of Heston et al. (2010).

4
stock price change during the short news arrival period, the common component of price
changes due to movements in the aggregate stock market. The news momentum effect
remains large and statistically significant with this change in experimental design.
Fifth, we use the Daniel et al. (1997) characteristic adjustment to compute the abnormal
return of the news momentum strategy and obtain similar results to the four-factor alpha.
Sixth, we drop earnings announcements from our news universe to eliminate the effect of
the post-earnings announcement drift. With this filtering, our news momentum strategy
remains highly profitable. Seventh, we filter out extreme daily price movements (with ab-
solute values exceeding 10%) as Savor (2012) investigates, and the performance of our news
momentum strategy is still largely intact. Eighth, to consider whether the news momentum
effect results mainly come from news clustering (good news tends to follow good news), we
exclude the news-driven returns from the holding period returns to the news momentum
strategy. The news momentum effect remains large and statistically significant, suggesting
that news clustering is unlikely to drive our results. Ninth, we consider how news momentum
return varies over each day of the week and document stronger market reaction to Friday
and weekend news. Finally, we consider Chan (2003)’s research design to classify news and
non-news returns using monthly stock returns. We find that news return based on that
approach does not predict future stock returns in our sample period from 2000 to 2012.
Our study is related to the growing literature that explores the effect of investor atten-
tion in financial markets. For instance, Hirshleifer et al. (2009) and DellaVigna and Pollet
(2009) explore how investor inattention may exacerbate post-earnings-announcement drift.
Cohen and Frazzini (2008) study how insufficient attention to a firm’s major customers may
lead to return predictability along the value chain. Our evidence of widespread post-news-
announcement drift, and the particularly strong news momentum following overnight and
weekend news, suggests that investor attention may be an important factor contributing to
the pervasive phenomenon of news momentum.
Our paper is also related to but different from a large literature that uses linguistic anal-

5
yses of media articles to extract sentiment and predict stock returns. For instance, Tetlock
et al. (2008) use the fraction of negative words in news stories to predict future earnings
surprises and stock returns. Tetlock (2011) employs linguistic analyses to identify stale news
and reports evidence of overreaction to stale news (initial momentum and subsequent rever-
sal). Our paper uses the stock market reaction to identify good and bad news, with a focus
on a high-frequency return decomposition, to understand the nature of short-term return
predictability.
In relation to the voluminous literature on return predictability, we focus on the role
of public news announcements. Chan (2003) shares a similar spirit, with methodological
differences. In particular, Chan (2003) examines stock return patterns following the month
with headline news. He finds evidence of post-news price drift for stocks with headlines
and reversal for stocks without identifiable news. Our paper differs in both methodology
and empirical results. First, our methodological contribution is to use tick-by-tick data to
more accurately capture market reaction to news, thereby increasing the statistical power
in identifying investor underreaction to public news. As an illustration, we replicate Chan
(2003)’s study in our new sample period. We find that the news return based on his approach
has no power to predict subsequent stock returns, but the news return out of our method has
strong return forecasting power. Second, in terms of results, we show that there is continued
market reaction to firm news after the initial reaction, with much of the further reaction
concentrated in the first week following the news. Chan (2003) shows, however, that during
the month immediately following the month with headline news, there is no further stock
return continuation; rather, the return continuation starts to build up in the subsequent
eleven months. These large differences suggest that the underlying phenomenon we capture
may be quite different.
We organize the rest of this article as follows. Section 2 introduces the data and variable
construction. Section 3 shows evidence of news momentum. Section 4 studies the source of
news momentum. Section 5 details the analysis of market reaction to overnight news, and

6
Section 6 presents the robustness tests. Section 7 concludes.

2 Sample Construction and Variable Definition

Our sample consists of all the firms listed on the New York Stock Exchange (NYSE), National
Association of Securities Dealers Automated Quotations (NASDAQ), and American Stock
Exchange (AMEX) with at least one news story covered by the Dow Jones News Wire. Our
intraday price and quote data come from the TAQ database; high-frequency firm news data
are from RavenPack; dividends, share splits and other stock market data are from CRSP;
accounting data are from Compustat; and analyst forecasts are from I/B/E/S. Our sample
period is from March 2000 to October 2012. Following prior literature, we use common
stocks with share code of 10 or 11.
To prepare the intraday return data, we gather minute-by-minute observations of intraday
prices by applying the cleaning rules of Barndorff-Nielsen et al. (2009) and Bollerslev et al.
(2016) to the TAQ database. Using these intraday prices, we then compute intraday returns
as every 15-minute return between 9:45 a.m. and 4:00 p.m. and the overnight return as the
return between 4:00 p.m. on the previous trading day and 9:45 a.m. on the current trading
day.3 Because the TAQ transaction prices are raw prices without adjustments for share
splits, we use the daily “cumulative factor to adjust price” and “dividend cash amount”
variables in the CRSP database to adjust for split and dividend.
The RavenPack news database provides a comprehensive sample of firm-specific news
stories from the Dow Jones News Wire (see, e.g., Jiang and Sun (2015), and Kelley and
Tetlock (2017) for recent studies using this data set). To capture a news story specifically
about a given firm, we use the “relevance score” that RavenPack provides, which ranges from
0 to 100, capturing how closely the underlying news applies to a particular company, with a
score of 0 (100) meaning that the entity is passively (predominantly) mentioned. We require
3
We use the price at 9:45 a.m. for overnight returns to ensure that most stocks have traded at least once
after the market open, following Patton and Verardo (2012) and Bollerslev et al. (2016). As a robustness
test, we use 9:30 a.m. price to compute the overnight return and find qualitatively similar results.

7
news stories in our sample to have a relevance score of 100. To include only fundamental news,
we select acquisitions-mergers, analyst-ratings, assets, bankruptcy, credit, credit-ratings,
dividends, earnings, equity-actions, labor-issues, product-services, and revenues from a total
of 29 news groups. We exclude repeated news by setting the “event novelty score” (ENS)
provided by RavenPack to be 100, which captures only the fresh news about a company.
Applying these filters introduces no look-ahead bias because RavenPack assesses all news
articles within milliseconds of receipt and immediately sends the resulting data to users. All
information is thus available at the time of news release.
To capture the high-frequency market reaction to firm-level news, we combine the intra-
day return data partitioned at 15-minute intervals and firm-specific news event data time-
stamped at the second level. To avoid extremely illiquid stocks, we eliminate stocks that are
priced below $1 at the end of the portfolio formation period. Our final sample includes a
total of 5,480 firms that have at least one news story over the period of 3,189 days between
March 2000 and October 2012. A typical day has an average of 3,781 firms covered by news
stories.
Our main innovation is to decompose stock returns into news-driven and non-news-driven
returns based on high-frequency market reaction. Specifically, we classify a stock’s return
according to whether firm-level news is released during the return measurement period. For
news occurring within regular trading hours, the news return is simply the 15-minute return
over the same period that the news occurs. For news occurring during the weekend, holiday,
or overnight, the news return is the nearest subsequent overnight return to reflect that the
first reaction to such news stories is incorporated into the stock’s price only for the first trade
of the following trading day. For example, the return for news events during the weekend
is the return over the period of 4:00 p.m. of the surrounding Friday and 9:45 a.m. of the
surrounding Monday. After classification, we aggregate all news and non-news returns within
each day starting from 4:00 p.m. on day t − 1 to 4:00 p.m. on day t to form daily news and
non-news returns. This essentially decomposes the overall daily return into two orthogonal

8
components. More formally, suppose there are M overnight plus intraday returns per day.
j
For example, in the case of 15-minute returns, M = 26. Let ri,t be the jth overnight or
intraday simple return for firm i on day t, where j = 1, 2, . . . , M . We compute the daily
news and non-news returns for firm i and day t as follows:

M
Y M
Y
j j
Ri,t,news = (1 + ri,t,news ) − 1, Ri,t,non-news = (1 + ri,t,non-news ) − 1, (1)
j=1 j=1

j j j
where ri,t,news = ri,t if there is a news story in the interval j and 0 otherwise, and ri,t,non-news =
j
ri,t if there is no news story in the interval j and 0 otherwise. Clearly, the daily overall return
is the product of news and non-news returns, namely,

Ri,t,overall = (1 + Ri,t,news ) × (1 + Ri,t,non-news ) − 1.

We construct a set of control variables according to standard definitions in the literature.

Market value of equity (Size) is the product of the closing price and the number of shares
outstanding, updated daily from CRSP. Book-to-market ratio (BM ) in June of year t is
computed as the ratio of the book value of common equity in fiscal year t − 1 to the market
value of equity in December of year t − 1 and is updated every year. Volatility is defined as
the realized variance (RV ), which is the sum of squared overnight and 15-minute intraday
returns within each trading day. Turnover is the total number of shares traded in the prior
five-day period divided by the average number of shares outstanding during the same period.
Illiquidity (ILLQ) is the Amihud measure of illiquidity (Amihud, 2002), which is the average
daily ratio of the absolute stock return to the dollar trading volume over the five-day period
preceding each day. Momentum (M om) is the cumulative returns from day t − 252 to day
t − 21 for a given day t and is updated daily. Analyst coverage (Analyst) is the monthly
number of sell-side analysts forecasting annual firm earnings.
Table 1 provides descriptive statistics for these variables. In Panel A, the mean, standard
deviation, and five quantiles are first computed cross-sectionally and then averaged over time.

9
Since our interest is in market reaction to firm-level news, we require at least one news story
for a given firm in a given day to be included in the computation of daily news return Rnews .
For a firm-day pair without relevant news stories, the entire daily return is non-news return
Rnon−news . The results indicate that the average news return is 0.18% per day as compared
with the 0.06% per day for the average non-news return. The cross-sectional dispersion of
news returns measured by their cross-sectional standard deviation is 4.33%, which is larger
than the dispersion of 3.31% for the non-news returns. Panel B shows the average cross-
sectional correlations among our key variables. It indicates a moderate correlation between
news and non-news returns. Their correlations with firm characteristics are generally low.
The correlation structure among firm characteristics is consistent with previous literature.
For instance, we find that smaller firms tend to have higher volatilities, less liquidity, and
lower analyst coverage.

3 News Momentum

3.1 Univariate Sorts: A Strategy of Buying Winners and Selling

Losers

We start by testing the profitability of a news momentum strategy designed to exploit short-
term market reactions to firm-level news. Figure 1 shows the timeline for our strategy. At
the 4:00 p.m. market close of each trading day t, we sort stocks into decile portfolios based
on their news returns on day t, Ri,t,news . We compute the equal-weighted returns for each
decile portfolio and a self-financing strategy that buys stocks in the top decile with high
news returns and sells stocks in the bottom decile with low news returns with a one-week
holding period until market close of day t + 5. To increase the power of our tests, we follow
Jegadeesh and Titman (1993) by using portfolios with overlapping holding periods. That is,
we revise the weights on one-fifth of the securities in our news momentum strategy on any
given day and carry over the rest from the previous day, resulting in non-overlapping series

10
of portfolio returns throughout calendar days.
Panel A of Table 2 summarizes the portfolio returns, which are converted to monthly
returns by multiplying daily returns by 21 for the ease of interpretation. The row labeled
“Return” reports average realized returns of each equal-weighted decile portfolio. It shows
a monotonically increasing relation between news returns and future stock returns. The
average monthly return increases from −0.78% for the loser portfolio in decile 1 to 2.55% for
the winner portfolio in decile 10, yielding a return of 3.34% per month with a t-statistic of
11.72. Stated in annual terms, the news momentum strategy generates a return of 40.08%
per year and an annualized Sharpe ratio of around 3.29.
To determine whether the return of our news momentum strategy results from their
exposures to other return factors, especially the price momentum factor, we use the popular
Fama-French-Carhart four (FFC4) factors (Fama and French, 1993; Carhart, 1997) to control
for the risk exposures of news momentum. Specifically, we regress excess returns of each
decile portfolio along with the long-short news momentum strategy against the FFC4 factors
and compute the regression intercepts, which are named as FFC4 alphas. The row labeled
“FFC4” in Panel A of Table 2 shows a similarly strong positive relation between news
returns and abnormal future returns in terms of FFC4 alphas. The FFC4 alpha of the news
momentum strategy is 3.37% per month and remains highly significant with a t-statistic of
11.76.
The similar magnitudes between raw and abnormal returns of the news momentum strat-
egy are explained by the low exposures of the strategy to the four factors, as shown in Panel
B of Table 2, which indicates that the news momentum strategy has statistically insignificant
loadings on the market, size, and value factors. The only statistically significant exposure
of the news momentum strategy is to Jegadeesh and Titman (1993)’s medium-term price
momentum factor, with positive sign but a weak magnitude of only 0.06.
A number of studies have emphasized the “crash” in returns of the price momentum
strategies, e.g., in early 2009 (Barroso and Santa-Clara, 2015; Daniel and Moskowitz, 2016).

11
How does our news momentum strategy perform through time? Figure 2 tracks the perfor-
mance of the news momentum strategy over our sample period. Specifically, we compound
the daily returns of the news momentum portfolios over time and measure the cumulative
profit Wt on day t as follows:

Wt = Wt−1 × (1 + Rwinner,t − Rloser,t + Rrf,t ), t = 1, 2, . . . ,

where Rwinner,t , Rloser,t and Rrf,t are returns of the winner portfolio in decile 10, returns of the
loser portfolio in decile 1, and the risk-free rate on day t, respectively. In Figure 2, the y-axis
presents the dollar value given W0 = $1 initial investment at the start of March 2000. Note
that the news momentum strategy generates superior performance throughout our sample
period without experiencing major drawdowns. The maximum drawdown is approximately
15.92%, which took place at the start of the sample during a short period between March 8,
2000, and May 04, 2000. Furthermore, our news momentum strategy appears to avoid the
severe crash that nearly wiped out the capital of traders on the traditional medium-term
price momentum in early 2009.4
How long does the news momentum effect persist? We answer this question by computing
the cumulative news momentum profits following an event study approach. For each portfolio
formation day t, we form decile portfolios based on day t’s news returns Ri,t,news at the end
of day t, and then compute the cumulative returns from day t + 1 to day t + k for each decile
portfolio. The spread between the cumulative returns in deciles 10 and 1 then forms a time
series of cumulative profits of winner-minus-loser portfolios for the event day k. To draw
inference about these cumulative profits, we aggregate them through time to compute their
average and the associated confidence intervals for a given event time k. By construction,
there are k − 1 days of overlap between any two consecutive observations of the spread series,
so we use Newey-West (1987) robust standard errors with lag k − 1. For comparison, we
4
For instance, Daniel and Moskowitz (2016) report that the price momentum strategy lost 42.28% and
45.52% in March and April of 2009, respectively.

12
perform a similar exercise by forming portfolios based on past non-news returns.
The upper and lower solid curves in Figure 3 plot the average cumulative profits for
strategies that buy winners and sell losers using news and non-news returns, respectively,
against the event day k for up to 252 days after portfolio formation. The results are striking.
The news momentum strategy continues to generate higher returns for up to 252 days after
portfolio formation, which is consistent with delayed investor reaction to initial news and a
gradual adjustment in prices. In contrast, non-news return experiences a subsequent reversal,
which leads the strategy of buying winners and selling losers to generate negative returns.
This reversal takes place gradually and remains statistically significant for approximately
75 days after portfolio formation. In our sample period (2000–2012), we do not observe the
shift in sign for the short-term reversal to medium-term momentum as Gutierrez and Kelley
(2008) observe, most likely due to the momentum crashes that mitigate the medium-term
price momentum effect over our sample period.

3.2 Fama-MacBeth (1973) Regressions

In this subsection, we examine the news momentum using the method of Fama and MacBeth
(1973). Specifically, for each day t, we perform the following cross-sectional regressions:

p
X
Ri,t+1:t+5,overall = γ0,t + γnews,t Ri,t,news + γnon-news,t Ri,t,non-news + γj,t Zj,i,t + i,t , (2)
j=1

where Ri,t+1:t+5,overall is the cumulative overall return from day t + 1 to day t + 5, and the
news return Ri,t,news , the non-news return Ri,t,non-news , and the control variables Zj,i,t are all
measured at the end of each day t for firm i. For each day t, we obtain the slope coefficients
from these cross-sectional regressions. We compute the time-series average of each slope
coefficient to test if the predicting variables are statistically significant in forecasting the five-
day-ahead returns. Our control variables include firm size, the book-to-market ratio, stock
returns from day t−252 to day t−21 as a proxy for stock price momentum, the daily realized

13
variance, and the prior day illiquidity measure of Amihud (2002). Because the dependent
variable has overlapping returns of four days, we use the Newey-West (1987) procedure with
four lags to adjust for serial correlation in the time series of the slope coefficients.
Table 3 reports estimated regression coefficients and t-statistics under several model
specifications. Consistent with portfolio analyses, the results show that news returns have
positive predictive power for the five-day-ahead overall returns. The average slope coefficient
for news returns in Regression (I) indicates that 4.25% of the prior day’s news return carries
over into the following week’s overall return. The results remain intact after controlling for
other predicting variables. The magnitude of news momentum is in the range of 4.25% with
a t-statistic of 8.28 in Regression (I) and 6.41% with a t-statistic of 11.45 in Regression
(V). To get a sense of economic significance, note that Table 1 shows the average cross-
sectional standard deviation of daily news return is 4.33%. Therefore, a two-standard-
deviation increase in news returns predicts a rise of approximately 28.87% (2 × 4.33% ×
0.0641 × 52) per annum in future returns. In contrast, the non-news returns tend to reverse
in the subsequent week. Among the control variables, book-to-market ratio has a positive and
statistically significant slope coefficient, and firm size is negatively related to future returns,
both of which are consistent with previous literature. In summary, the Fama-MacBeth
regressions lend further support to the strong news momentum effect in stock returns.

3.3 An Investment Perspective

In this subsection, we examine the incremental investment value of the news momentum
strategy. We consider a mean-variance investor whose investment opportunities include five
popular trading strategies: holding the market portfolio (MKT), a small firm strategy (SMB)
that buys small firms and shorts big firms, a value strategy (HML) that buys value stocks
with high book-to-market ratios and shorts growth stocks with low book-to-market ratios, a
momentum strategy (UMD) that buys past winners and sells past losers during the past 2
to 12 months, and a short-term return reversal strategy (REV) that buys stocks that have

14
gone down and shorts stocks that have gone up during the prior month.5
For each of the five strategies together with our news momentum strategy, we compare
the mean, standard deviation, Sharpe ratio, skewness, kurtosis, and maximum drawdown
based on the daily returns. Panel A in Table 4 reports these descriptive statistics. The
results illustrate the appealing feature of the news momentum strategy: It has the lowest
risk as measured by standard deviation, skewness, kurtosis, and maximum drawdown, but
has the highest average return. Taking both risk and return into account, a monthly Sharpe
ratio of 0.95 suggests the news momentum far outperforms other strategies, among which the
highest Sharpe ratio is approximately 0.27 for the short-term return reversal strategy. Panel
B in Table 4 displays the time-series correlation matrix among the six strategies. The news
momentum has a moderate positive correlation with the price momentum strategy (0.11)
and negative correlations with the value and short-term return reversal strategies (-0.07 and
-0.07). The low correlations between news momentum and other trading strategies imply
large potential gains for a mean-variance investor.
To illustrate these gains, we construct the mean-variance frontiers implied by a set of
investment opportunities for the five traditional trading strategies and another set including
the news momentum strategy. We then construct optimal portfolios with the highest Sharpe
ratio from these two investment opportunity sets. Panel C of Table 4 reports the performance
of these two tangency portfolios. The results show substantial gains of tilting the portfolio
toward the news momentum strategy: The monthly Sharpe ratio increases from 0.40 to 1.07,
with declines in higher-moment risk.

3.4 Exploration of Transaction Cost

From a practical point of view, it is useful to consider whether the strategy of news mo-
mentum remains profitable after transaction costs. To shed light on this question, we follow
Chordia et al. (2000), using the proportional effective spread (PES) as one measure of trading
5
We obtain the daily returns of these five strategies from Kenneth French’s website.

15
cost:

2|P ricei,t − 0.5 × (Bidi,t + Aski,t )|

PESi,t = , (3)
P ricei,t

where P ricei,t , Bidi,t , and Aski,t are the last transaction, bid and ask prices of stock i on
day t. By definition, PES measures the scaled difference between execution price and the
midpoint of NBBO (National Best Bid and Offer). We multiply the absolute difference by
two to measure the round-trip trading cost. In the literature, effective spread is a widely
used measure to estimate the transaction cost (e.g., Hasbrouck 2009; Novy-Marx and Velikov
2016).6
To evaluate the trading cost of the news momentum strategy, we compare the gross profit
with the average effective spreads associated with the strategy. Because the news momentum
strategy in Section 3.1 has a five-day holding period, we assume a 20% daily turnover for each
side of the long short portfolio, using one-fifth of PES as a proxy of the trading cost for each
of the decile portfolios formed in Section 3.1. The row labeled “PES based on 20% turnover”
in Panel A of Table 5 shows the average one-fifth of total PES of each decile portfolio in
bps. Interestingly, these decile PESs exhibit a U-shaped pattern, with the highest average
PES of 10.5 × 5 = 52.5 and 11.3 × 5 = 56.5 bps in deciles 1 and 10 and the lowest PES
of approximately 4.6 × 5 = 23 bps in deciles 5 and 6. This pattern indicates the higher
bid-ask spread associated with extreme news-driven returns. A possible explanation is that
market markers increase the bid-ask spread to compensate for the greater adverse selection
surrounding the arrival of important firm news.
To gauge the profitability of the strategy, we obtain the net return for the news momen-
tum strategy by subtracting the trading costs from the row “Gross Return,” which shows
the same daily decile portfolio returns in basis points as in Table 2. The results show that
6
The exact definition of effective spread appears to vary across applications. For example, Chordia et al.
(2000) uses the definition as in Eq. (3); Lesmond et al. (2004) replaces the transaction price with mid-quote
price in the denominator of Eq. (3); and Hasbrouck (2009) uses the difference between the log transaction
price and the log mid-quote price. We find that these variations make little difference to our results.

16
the return to the equal-weighted news momentum strategy does not survive the erosion of
transaction costs. For the long-short news momentum portfolio, the daily average transac-
tion cost associated with the long portfolio in decile 10 and short portfolio in decile 1 adds
up to 21.7 bps, which is slightly larger than the average daily return of 15.9 bps to the spread
portfolio. This result may not be too surprising in that the equal-weighted portfolio is tilted
toward small stocks that are expensive to trade, and is therefore known to be less profitable
in practice (Novy-Marx and Velikov, 2016).
The value-weighted news momentum strategy, however, remains profitable after transac-
tion cost. Panel B of Table 5 shows the value-weighted portfolio returns and their average
trading costs. While the trading costs exhibit a U-shaped pattern similar to those of the
equal-weighted portfolios, their magnitudes are less than one-half of those of the equal-
weighted portfolios. Deciles 1 and 10 have an average trading cost of 4.0 and 4.3 bps.
Therefore, the trading cost for the long and short portfolio totals only 4.0 + 4.3 = 8.3 bps.
Since the value-weighted news momentum strategy has an average daily return of 11.7 bps,
it generates a net profit of (11.7 − 8.3)/10000 × 252 = 8.6% per year. These results suggest
that the news momentum strategy has the potential to be profitable even after accounting
for transaction costs.

4 Source of News Momentum Profits

4.1 Decomposing News Momentum Profits

The high return to the news momentum strategy can arise from several sources. To better
understand its nature, we follow Lo and MacKinlay (1990), decomposing the expected news
momentum profit into three components: the average autocovariance of individual stock
returns, the average cross-autocovariance across stocks, and the cross-sectional variance in
expected stock returns (see also Lehmann (1990); Lewellen (2002); Nagel (2012)). The first
component captures the serial correlation in individual stock returns: A positive value is

17
consistent with the market underreaction hypothesis for news momentum (because we have
found evidence against the hypothesis of delayed overreaction). The second component re-
flects the lead-lag effects across stocks: If the average cross-autocovariance among stocks is
positive (e.g., returns of large stocks lead those of small stocks due to their higher liquidity),
it would reduce the return to the news momentum strategy. The last component reflects the
cross-sectional dispersion in expected stock returns: If news momentum strategy systemat-
ically picks up more risky stocks with higher expected returns, a high average return could
thereby emerge. Clearly, these different components associate with very different interpre-
tations. Examining which source drives the return to the news momentum strategy thus
illuminates the nature of the news momentum effect.
Following Lo and MacKinlay (1990), we consider a news momentum strategy with the
following portfolio weights:

1
wi,t = (Ri,t,news − Rm,t,news ),
N

where Rm,t,news = ( N
P
i=1 Ri,t,news )/N is the average news-driven return on day t. The port-

folio return on day t + 1 equals:

N N
X 1 X
πt+1 = wi,t Ri,t+1,overall = (Ri,t,news − Rm,t,news )Ri,t+1,overall .
i=1
N i=1

We can then show that the expected news momentum profit equals the sum of three com-
ponents:

N −1 1 0
E(πt+1 ) = tr(Γ) − [1 Γ1 − tr(Γ)] + Cov(µnews , µoverall ), (4)
N2 N2

where Γ = Cov(Rt,news , Rt+1,overall ) is the covariance matrix between news-driven return

Rt,news ≡ (R1,t,news , R2,t,news , . . . , RN,t,news )0 on day t and the overall return Rt+1,overall ≡
(R1,t+1,overall , R2,t+1,overall , . . . , RN,t+1,overall )0 on day t+1, and Cov(µnews , µoverall ) is the cross-
sectional covariance between average news returns and overall returns.

18
 Eq. (4) shows that there are three possible sources of the news momentum profit. The
N −1
first term, N2
tr(Γ), is the average autocovariance of individual stocks. It is positive when in-
dividual stocks with high past news-driven returns tend to have high overall returns in the fu-
ture. The second term, − N12 [10 Γ1−tr(Γ)], is the negative of the average cross-autocovariance.
It is positive when there is on average a negative cross-autocovariance (e.g., good news for
one company leads bad news for another company). The third term, Cov(µnews , µoverall ),
is the cross-sectional covariance of average news returns and average total returns, which
captures the dispersion in expected returns associated with news returns. It is positive when
firms with high news returns tend to have high expected returns.
Our empirical implementation of this decomposition follows Lehmann (1990) and Nagel
(2012), using scaled portfolio weights to ensure that the portfolio is $1 long and $1 short,
with the magnitude of profits more interpretable:

1
wi,t = (Ri,t,news − Rm,t,news ),
Ct

where Ct = ( N
P
i=1 |Ri,t,news − Rm,t,news |)/2 is the normalizing constant.

The first row in Panel A of Table 6 shows the decomposition results, which are consistent
with the underreaction hypothesis. The total return to the news momentum strategy is
3.79% per month, almost all of which comes from the first, autocovariance component. The
total return and the autocovariance component are also highly significant with Newey-West
t-statistics of 6.03 and 5.29, respectively. In contrast, the second (cross-autocovariance) and
third (dispersion in expected returns) components are close to zero. When we compute the
four-factor alpha for the total news momentum return and the three components in the
first row of Panel B, we obtain similar findings. The result that the positive autocovari-
ance in individual stock returns drives news momentum supports the hypothesis of market
underreaction.
One limitation of the decomposition using individual stocks is that it requires complete
observations of stocks over the entire sample period. The resulting restriction is that we

19
have only 970 stocks for this analysis. To improve the power of our test and mitigate the
concern of potential survivorship bias, we also consider an industry-level news momentum
strategy. In particular, we construct news-driven and overall returns of industry portfolios
by first classifying stocks into the Fama and French (1997) seventeen sectors.7 Each day,
we calculate the industry news-driven return as the average news-driven returns of all firms
within an industry, and the industry overall return as the average return of all firms within
the same industry.
The industry news momentum strategy also indicates underreaction as the driving factor
for the news momentum. As the second row labeled “Industry Portfolio” in Panel A of
Table 6 shows, the industry news momentum earns a total monthly return of 1.09%, to
which the first component, autocovariance, contributes a positive monthly return of 2.07%.
In contrast, the second component, which captures the lead-lag effect across industries,
contributes a return of −0.98% per month, and the third component, dispersion in expected
industry returns, contributes a return of 0.01% per month. Panel B of Table 6 shows the
results based on the four-factor alpha, which generate a similar pattern.

4.2 News Momentum and Firm Characteristics

In this subsection, we examine the cross-sectional determinants of the news momentum effect
to shed further light on its nature. Specifically, we study whether the performance of the
news momentum strategy concentrates among stocks with certain characteristics, including
firm size, analyst coverage, volatility, illiquidity, and past returns.
Several strands of literature in behavioral finance motivated our choice of these variables.
The literature on limited investor attention naturally points to firm size and analyst coverage
as proxies for investor attention: Small firms with lower analyst coverage tend to receive
less attention from investors. Theoretical works on overconfidence such as Daniel et al.
(1998) argue that overconfident investors tend to underreact to public news but overreact to
7
We obtain qualitatively similar results using the Fama and French 10 and 12 industry classifications,
and the 20 industry portfolios used by Moskowitz and Grinblatt (1999).

20
their private information signals. Smaller firms tend to have fewer information disclosures,
firms with lower analyst coverage tend to lack timely financial analyses, and firms with
more volatile stock prices tend to have more asymmetric information; therefore, through
this theoretical lens, we would expect these companies to have more private information,
which results in more underreaction to public news and thus stronger news momentum.
From the point of view of limits to arbitrage (Shleifer and Vishny, 1997), illiquid firms with
high volatilities are more costly and risky to trade, which could deter arbitrageurs from
betting against perceived mispricing. We would expect those firms to exhibit stronger news
momentum. Finally, it would be of interest to explore potential interaction between news
momentum and the Jegadeesh and Titman (1993) price momentum.
We perform independent double sorts to examine these conjectures.8 At the end of each
day t, we independently sort all stocks into three portfolios along one dimension based on
a particular stock characteristic and three portfolios along another dimension based on the
news return on day t. We compute equal-weighted returns on the nine portfolios. Similar
to the univariate sorts in Subsection 3.1, these portfolios have five-day overlapping holding
periods, with one-fifth of the portfolio rebalanced each day. Because volatility and analyst
coverage strongly correlate with size, as shown by the large correlation coefficients of 0.5 or
higher in Table 1, we compute size-adjusted volatility and analyst coverage to alleviate the
confounding effect of firm size.
Table 7 presents the results from the double sorts. Panel A shows that, consistent with
our conjecture, the news momentum effect is stronger among small stocks. In particular, the
four-factor alphas of winner-minus-loser strategies are 2.19%, 1.45%, and 0.51% per month
for small-, medium-, and large-sized firms with the corresponding t-statistics of 9.05, 7.28,
and 2.89, respectively. The difference in news momentum profits between small and large
firms is 1.68% per month, which is statistically significant with a t-statistic of 5.60. Similarly,
we find evidence in Panel B that the news momentum effect tends to be stronger among
8
We also performed sequential sorts that first sort stocks according to a particular stock characteristic
into terciles and then sort stocks within each tercile into terciles based on the daily news returns. The results
are qualitatively similar to independent sorts.

21
stocks with low analyst coverage. For instance, stocks in the bottom tercile with low analyst
coverage (adjusted by size) have a news momentum profit that is approximately 60% higher
than firms in the top tercile with high analyst coverage. This difference in the four-factor
alpha is 0.67% per month, with a t-statistic of 2.50. These results support the notion that
stronger underreaction to public news, due to either overconfidence or investor inattention,
may be driving the stronger news momentum effect among smaller firms with less analyst
coverage.
Panel C of Table 7 shows size-adjusted volatilities. The results indicate a stronger news
momentum effect among more volatile stocks. For example, the four-factor alpha for the
winner-minus-loser strategy is 1.84% per month in the most volatile tercile, but only 1.01%
in the least volatile tercile. The difference in the four-factor alpha between the two groups of
stocks is 0.83% per month, with a t-statistic of 2.88. This result is consistent with investor
overconfidence as well as higher trading costs deterring the effectiveness of arbitrage against
investor underreaction.
In Panel D, we use the Amihud illiquidity measure to explicitly gauge how the news
momentum effect interacts with trading costs. Indeed, we find particularly large news mo-
mentum profits for stocks most costly to trade. Among the tercile of stocks with the highest
illiquidity measure, the four-factor alpha of the news momentum strategy is 1.95% per month,
with a t-statistic of 9.67, whereas among the tercile with the lowest illiquidity, the four-factor
alpha of a similar news momentum strategy is only 0.27% per month, with a t-statistic of
1.54. This return spread of 1.68% per month has a t-statistic of 6.65.
Before concluding this subsection, we examine the interaction between the news mo-
mentum effect and the Jegadeesh and Titman (1993) price momentum. Panel E shows the
results. In our sample period of 2000–2012, the price momentum effect is statistically in-
significant, but the news momentum effect is strong and significant across the three terciles
of stocks based on past returns. Interestingly, we find that the news momentum effect is
stronger among loser stocks in the past year.

22
5 Overnight News

In this section, we perform an in-depth analysis of overnight news, which has received rel-
atively little attention in the literature. Using an intraday event study approach, we find
compelling evidence for delayed reaction to overnight news, which constitutes more than
half of our sample. It lends further support to underreaction as the main driver of news
momentum.
Specifically, to investigate potential delayed reaction to overnight news, at 10:00 a.m. on
each trading day t, we sort the news returns computed from the close of day t − 1 to 9:45
a.m. of day t into deciles and then hold the winner-minus-loser portfolio for the subsequent
five trading days.9 We calculate returns of this strategy over the 30-minute interval when
the market is open and the interval of overnight return when the market is closed. For
comparison, we perform a similar exercise using the overnight non-news returns.
Panel A in Figure 4 plots the time-series averages and 95% confidence intervals against
the event time for the two strategies and highlights the difference in post-formation return
patterns. Immediately after the portfolio construction at 10:00 a.m., the overnight news
momentum gradually increases to 25 bps at 3:30 p.m. and then slightly drops to 17 bps at
the 4:00 p.m. close. After the following overnight period, the overnight news momentum
sharply rises to 59 bps at 9:45 a.m. after market opens and keeps rising for every overnight
period on the subsequent event days. Interestingly, except for event days 1 and 2, the
overnight news momentum produces small spreads during the open-to-close periods.10 The
average five-day return in investing in the overnight news has a large spread of 104 bps,
substantially larger than the average five-day cumulative return of 88 bps shown in Figure
3, where the impact of the overnight news on the immediate open-to-close period is ignored.
9
We skip the return between 9:45 a.m. and 10:00 a.m. to reduce the contamination induced by mi-
crostructure effects such as bid-ask bounce.
10
Lou et al. (2017) finds that returns of the price momentum strategy based on past 12-month returns
tend to accrue overnight. Although a substantial part of our overnight news momentum profits also manifest
during the overnight holding period, the intraday component of the overnight news momentum strategy is
also statistically and economically significant. For example, Figure 4 suggests that, of the 104 bps of five-day
cumulative returns, 17 and 12 bps accrue during the first and second intraday periods, respectively.

23
 The overnight non-news returns, in contrast, immediately induce a gradual reversal until
the 4 p.m. close and then exhibit a momentum pattern during the subsequent overnight
periods that offsets some of the intraday reversal. This pattern repeats itself on every
subsequent event day. When aggregated, the intraday reversal part dominates the overnight
momentum, and the average five-day return in investing in the overnight non-news reversal
is 52 bps.
Despite the overall patterns of overnight news momentum and non-news reversal, Panel
A of Figure 4 also shows that both types of overnight returns continue during the overnight
period in the following days, consistent with the intraday periodicity pattern of Heston et al.
(2010).11 This overnight return periodicity suggests that controlling for it might help tease
out the news momentum effect more clearly. To do so, we employ a matched sample proce-
dure as follows. At 10:00 a.m. on every trading day and for every overnight news return, we
find one non-news return with the smallest return differences. By using this matched sam-
ple, we can compare the predictability of the news- and non-news-driven overnight returns
of similar magnitude, reducing the confounding effect of the return periodicity. We sort
these matched non-news returns for constructing the winner-minus-loser portfolio and then
subtract the cumulative return of the winner-minus-loser portfolio based on sorting matched
non-news returns from the cumulative return of the winner-minus-loser portfolio based on
sorting news returns over the same holding period. These differences constitute the abnor-
mal overnight news momentum adjusted by the return periodicity. For a given event time,
we compute the average of the time series of these abnormal returns and its 95% confidence
interval in Panel B of Figure 4. Note that the abnormal overnight news momentum based
on the adjustment of the matched portfolio remains strong. Overall, the average five-day
abnormal news momentum profit is 221 bps, highly statistically significant as indicated by
the tight confidence intervals. The abnormal profit is especially prominent during the intra-
day period when the news-driven returns generate momentum while the matched non-news
11
Heston et al. (2010) excludes overnight close-to-open price movements from their analysis and documents
a striking pattern of return continuation at half-hour intervals that are exact multiples of a trading day.

24
driven returns generate reversal. The abnormal profit, however, dips during the overnight
periods starting at 4:00 p.m. of each day, suggesting that the return periodicity of Heston
et al. (2010) absorbs the overnight news momentum profits accrued during the overnight
period.

6 Robustness Tests

6.1 CRSP Data

Our main return sample consists of overnight and 15-minute intraday returns computed
from the transaction prices available in the TAQ database. This choice of high-frequency
returns is relatively new and differs from the majority of existing literature on either news
or momentum, which typically relies on lower-frequency returns at, for example, daily or
monthly frequencies. An advantage of the higher-frequency data is that it enables a sharper
distinction between news-driven and non-news driven returns, which in turn enhances the
ability to separate the different reactions to information and non-information based price
changes. However, a possible concern of high-frequency data is that it is noisy in several
ways. First, not only do the well-known microstructure issues such as bid-ask bounce and
stale price cause the observed returns to be less informative about the real underlying price
process, but data recording errors are also likely to appear in the raw intraday data, yielding
anomalous returns.12 Second, it is possible that the news time stamp is imprecise about
the actual news release time or more importantly the true information event. However,
as Tetlock (2010) argues, news and information events usually occur on the same day, so
the recorded news event and the true information event might be synchronous at the daily
level. Thus, it would be useful to determine if the news momentum discovered from TAQ is
sensitive to the sampling frequency and whether high-frequency data enhances or diminishes
the findings over the daily return data.
12
We implement a set of clearing rules commonly used in high-frequency econometrics literature to elim-
inate possible errors in the high-frequency data.

25
 We use the daily CRSP data to repeat the single-sort analysis in Section 3.1. In particular,
for the predictor of news returns, we classify the distribution adjusted close-to-close daily
returns from CRSP into news and non-news categories based on whether at least one news
event occurred during the close-to-close period. In the notation of Equation (1), M becomes
1 because there is only one observation per day, and we define future overall returns to be
predicted using the CRSP daily returns. We then sort the five-day-ahead overall returns
into decile portfolios using the previous one-day news return. Panel A of Table 8 displays
the average monthly returns of portfolios in the same format as Panel A of Table 2.
The news momentum pattern remains in those daily returns. The winner-minus-loser
portfolio generates a monthly risk-adjusted average return of 1.98% with a t-statistic of 6.05.
However, recall that the risk-adjusted spread from sorting news returns constructed from
the higher-frequency data is 3.37% per month (t-statistic=11.76) in Panel A of Table 2.
The spread magnitude and statistical significance constructed from sorting daily or coarser
frequency returns are much weaker than those obtained from sorting high-frequency returns.
We also replaced the predictor of news returns based on daily data with those based on
high-frequency data while keeping the response variable of overall returns based on CRSP
data. The same portfolio strategy generates a risk-adjusted monthly spread of 3.48% with a
t-statistic of 12.15 (untabulated). Our usage of the new measure of news returns purges this
component out of the daily news-driven return and is thus crucial in building stronger news
momentum strategy.

6.2 Mid-Quote Returns

Return data of frequency from 15 minutes to one week computed from transaction data
might contain measurement errors arising from microstructure noises. Measurement errors
threaten inferences based on transaction return data alone. A stylized example is the spu-
rious reversal pattern due to negative autocorrelation induced by bid-ask bounce at shorter
horizons (Roll, 1984), for which, returns of quoted prices are commonly used to address

26
the issue (e.g., Kaul and Nimalendran 1990; Gutierrez and Kelley 2008). However, it is
unlikely that the news momentum pattern is driven by microstructure noise such as bid-
ask bounce and non-synchronous trading. Indeed, negative autocorrelations due to bid-ask
bounce and positive cross-correlations due to non-synchronous trading both contribute to
the cross-sectional return reversal pattern rather than the momentum pattern documented
here. Nevertheless, it is useful, at least from a practical point of view, to investigate the news
momentum patterns using quoted prices. We carry out such analysis by first computing the
overnight and 15-minute returns using the mid-quote price and then aggregating them into
news and non-news returns according to Equation (1). We repeat the calendar time strategy
of Section 3.1 for these returns computed from quoted prices (summarized in Panel B of
Table 8). We find that the news momentum pattern remains intact when returns are formed
from mid-quote prices. The winner-minus-loser portfolio generates a monthly alpha of 3.42%
with a t-statistic of 12.75 – remarkably similar to the profitability of 3.37% per month and
a t-statistic of 11.76 reported in Table 2 from the transaction data.

6.3 Open-to-Open Returns

The main results in Section 3 demonstrate the predictability of news returns aggregated
over the close-to-close period on the subsequent five-day close-to-close overall returns. Using
close-to-close returns make the results more sensitive to the intraday information than the
overnight information. An interesting question is to investigate the performance of the news
momentum strategy using the open-to-open returns to allow for more overnight effects. To
do so, we compute the news returns accumulated over the period from 9:45 a.m. on day
t − 1 to 9:45 a.m. on day t and use them to forecast the five-day ahead overall returns
computed from 10:00 a.m. on day t to 10:00 a.m. on day t + 5. We repeat the same decile
portfolio strategies as in Section 3.1 and summarize their returns and FFC4 alphas in Panel
C of Table 8. The average winner-minus-loser spread of the decile portfolios based on open-
to-open returns is 3.92% per month with a robust t-statistic of 12.91. The alpha of the

27
strategy controlling for the four factors is 3.94% per month with a t-statistic of 13.03. These
performance measures are slightly higher than their counterparts in Table 2, which are based
on close-to-close returns.

6.4 Firm-Specific News Return

The main results in Section 3 sort stocks by their cumulative news returns, which do not
adjust for the riskiness of the stocks. To determine whether doing so would affect our
findings, we now identify news momentum based on risk-adjusted news returns. We use the
pre-event returns to estimate the regression models as follows:

Ri,t,overall − Rrf,t = αi + βi (Rm,t − Rf,t ) + i,t ,

where Ri,t,overall is firm i’s overall daily return, Rm,t is the market return, and Rrf,t is the
risk-free rate. We fit the model using a rolling OLS approach with a window size of 252 days
j
before the event day. We then compute the market-adjusted high-frequency returns mri,t
for firm i, interval j, and day t as

j j j
mri,t = ri,t − βi rm,t , j = 1, 2, ..., M,

j j
where ri,t and rm,t are the j th overnight or intraday simple return for firm i and the market,
respectively, on day t. We use the high-frequency return of the actively traded S&P 500
j j
ETF (ticker SPY) as a proxy for rm,t . We use the market-risk-adjusted returns mri,t in
place of signal construction of Equation (1) and repeat the single-sort analysis. Panel D of
Table 8 shows the results of sorting by market-risk-adjusted returns. The effects of news
momentum remain largely unchanged after adjusting for market risks. The High-Low spread
and its t-statistic are close to those in Table 2, in which the market risk is unadjusted when
constructing the signal.

28
6.5 Characteristic Adjustment

The single-sort exercise in Section 3.1 uses the Fama-French-Carhart factors for adjust-
ing risks in the news momentum strategy returns. In this section, we implement the
characteristic-based benchmark methods of Daniel et al. (1997) and Wermers (2003) as an
alternative way to adjust for risks. We use the benchmark portfolio assignments to compute
the equal-weighted daily 5 × 5 × 5 size, book-to-market ratio, and momentum benchmark
returns based on all NYSE/AMEX/NASDAQ data in the CRSP database.13 A firm’s bench-
mark adjusted return is then a firm’s daily overall return minus the daily return of one of
the 125 benchmarks to which the firm belongs to on that day. These benchmark-adjusted
returns are used in place of the raw overall returns for repeating the single-sort exercise
conducted in Section 3.1.
Panel E of Table 8 displays the results of sorting returns in excess of the characteristic-
matched benchmark returns. The news momentum effect remains strong. The zero-cost
winner-minus-loser portfolio generates a monthly return of 3.06% with a robust t-statistic of
11.14. This spread is slightly smaller than the spread of 3.34% for the raw returns in Table 2,
suggesting some profits of news momentum might be attributable to portfolio characteristics.

6.6 Earnings Announcements

The news momentum pattern we document does not differentiate the types of news stories
that could have different impacts on return continuation. It has long been shown that the
announcement of earnings news tends to trigger a stock’s returns to drift in the same direction
of the earnings surprise for several weeks after the announcement (Ball and Brown, 1968;
Bernard and Thomas, 1989). One explanation of the short-term news momentum pattern
is that it is merely a reconfirmation of the post-earnings-announcement drift pattern and
13
The benchmark assignments, updated monthly, are available via http://www.smith.umd.edu/
faculty/rwermers/ftpsite/Dgtw/coverpage.htm. To compute daily returns of the benchmark portfo-
lios, we convert this monthly assignment into daily frequency by assuming constant daily assignment within
each month.

29
non-announcement observations contribute none to the momentum profits. If so, we should
observe small positive return spreads after excluding the firm-trading day observations on
the earnings announcement dates. Therefore, we construct news momentum portfolios as
before except that we drop all firms that announce their earnings on the same day of the
portfolio formation. We identify quarterly earnings announcement using the announcement
dates from Compustat. Since the time stamp of the earnings announcement is unavailable
in Compustat and earnings announcement can occur before, during, or after the regular
trading hours, we are unable to match the returns that immediately reflect the information
on earnings announcement. To conservatively remove the effect of earnings announcement,
we exclude both the earnings announcement days and the days right after the announcement
from our sample.
Panel F in Table 8 displays the resulting profits. The difference in average returns between
the High and Low news return decile portfolios is 1.85% per month after adjusting for risks
and has a robust t-statistic of 5.92. The earnings announcement cannot entirely account for
the news momentum pattern. However, the fact that the spread and t-statistic are smaller
than their counterparts produced by sorting all samples in Table 2 suggests that earnings
announcement contributes substantially to the news momentum profits.

6.7 Extreme Price Changes

Savor (2012) studies large stock price movements with absolute daily returns exceeding
10%. He groups these extreme price movements to those accompanied with sell-side analyst
recommendations and others without, and finds evidence of return drift following extreme
price changes with analyst recommendations and reversal following those without analyst
recommendations. In this subsection, we assess whether the pervasive news momentum effect
we identify is sensitive to this type of extreme price movement.
Specifically, we test the return predictive power of news returns after excluding the firm-
trading day observations with extreme price changes. Following Savor (2012), we first identify

30
extreme price movements. We calculate a firm’s daily abnormal returns relative to the
four-factor model and classify a trading day for this firm into the set of extreme price
changes if the daily abnormal return exceeds 10% in magnitude. We then track analyst
recommendations from the I/B/E/S. If the price change is accompanied by at least five
analyst recommendations issued during the previous 12 months, we consider it information
driven. Our computation indicates that the set of information-based extreme price changes
represents a small proportion of our sample. We exclude these observations from our sample
and then compute the news momentum profits.
Panel G of Table 8 displays the results, which indicate that the news momentum effect
remains strong after removing these extreme observations. The last column labeled “High–
Low” shows the spread between the average returns of the winner and loser portfolios. The
raw return spread is 2.93% per month with a t-statistic of 10.75, and the four-factor alpha
is 2.97% per month with a t-statistic of 10.70.

6.8 News Clustering

To examine whether our results may be attributable to news clustering, i.e., positive (neg-
ative) news stories tend to be followed by positive (negative) news stories (see, e.g., Wang
et al. (2016)), we change the computation of holding period returns to our news momen-
tum strategy by including only non-news returns, which are not driven by news that arrives
subsequently. Panel H of Table 8 reports the performance of the news momentum strategy
using this metric. The results indicate that the news momentum effect remains strong, even
when we consider only non-news-driven returns in the holding period. The column labeled
“High–Low” shows the spread in average returns between the winner and loser portfolios,
which is 2.85% per month with a t-statistic of 10.51; the four-factor alpha is 2.88% per
month with a t-statistic of 10.53. Thus, news clustering alone could not explain the news
momentum.

31
6.9 Weekend News

If investor underreaction drives the news momentum effect, should we observe stronger mo-
mentum for news arriving over the weekend? To answer this question, we consider how
news momentum return varies over each day of the week. Specifically, we construct a news
momentum strategy that buys (sells) stocks with the highest (lowest) news return in the
previous day with a one-day holding period. Then, we regress daily returns to this strategy
on (1) five dummy variables that represent each day of the week without an intercept, or (2)
a dummy variable of Monday with an intercept. The coefficient on the Monday dummy is of
particular interest, because the return on Monday to the news momentum strategy is based
on news arriving over Friday and the weekend. If there is stronger market underreaction to
Friday and weekend news, we would expect that (1) in the first regression, the coefficient for
the Monday dummy variable would be larger than that for other days of the week; and (2)
in the second regression, the coefficient for the Monday dummy variable would be positive.
Results in Table 9 show supporting evidence. In Column (I), the coefficient for the
Monday dummy is 65 bps per day, which is the largest among the five weekdays; it is more
than twice as large as the coefficient for Tuesday, which is 28 bps per day. Column (II)
presents a formal statistical test on the equality of news momentum returns on Monday and
other days. The slope coefficient for the Monday dummy is 23 bps per day, with a t-statistic
of 2.84. These results suggest stronger market underreaction to Friday and weekend news.

6.10 Drift After Headlines

Chan (2003) studies stock return patterns following the month with headline news. He
finds evidence of post-news drift for stocks with headlines and reversal for stocks without
identifiable news. In particular, he groups stocks into news and non-news sets based on if
they had at least one news headline during a given month t, and finds that news stocks
experience less reversal in month t + 1 and then drift for most of the subsequent months in
the following year. This predictive return profile actually differs from ours, where momentum

32
already exists and is stronger at shorter horizons of hourly to daily holding periods, as evident
in Figure 4. To formally illustrate the difference between Chan (2003)’s effect and ours, we
replicate Chan (2003)’s strategy in our sample as follows.
At the end of each month, we consider a news group consisting of all stocks that have
at least one news story during that month. We then sort them into ten portfolios based
on their monthly return and compute the equal-weighted return of a self-financing portfolio
that buys stocks in the top decile with high returns and sells stocks in the bottom decile with
low returns with K = 1, 3, and 6 month holding periods. Following Jegadeesh and Titman
(1993), this Chan (2003) strategy includes portfolios with overlapping holding periods. That
1
is, we revise the weights on K
of the securities in our news momentum strategy in any given
month and carry over the rest from the previous month.
Table 10 summarizes the portfolio returns in monthly percentage. The rows labeled
“Return” and “FFC4” respectively report the average raw returns and Fama-French-Carhart
four-factor alphas for each portfolio. The column labeled “10–1” reports the difference
in returns between Portfolio 10 and Portfolio 1, with Newey-West robust t-statistics in
parentheses. We see that Chan (2003)’s effect does not exist in our sample as returns seem
to exhibit weakly reversal rather than momentum for those news stocks. The four-factor
alphas are −0.85%, −0.44% and −0.27% per month with t-statistics of −1.74, −1.85 and
−1.86 for one-, three- and six-month holding periods, respectively.

7 Conclusion

We decompose daily stock returns into news-driven and non-news driven components, by
matching a comprehensive sample of intraday firm-level news arrivals with high-frequency
price movements of individual stocks. Consistent with prior literature, we find that non-
news driven returns precede a reversal. For news-driven returns, however, we find strong
evidence of return continuation. A strategy of news momentum that buys stocks with high
news returns and sells stocks with low news returns generates an annualized return of 40.08%

33
in the following week, with a four-factor alpha of 40.44% controlling for the market, size,
value, and momentum. We attribute this effect of news momentum to the autocorrelation
component and find it is particularly pronounced for overnight news, and among small firms
with less analyst coverage, higher volatility, and lower liquidity, which is consistent with
imperfect investor reaction to news and limits to arbitrage. We further demonstrate that
the news momentum strategy has the potential to be profitable after transaction cost.
The pervasive phenomenon of news momentum that we identify is naturally connected to
the stock price momentum observed in stock markets for more than one and a half centuries
around the globe (e.g., Chui et al. (2010); Chabot et al. (2014)). Due to the short span
of our high-frequency news and price data, we could not fully examine the extent to which
underreaction to firm-level information drives the stock price momentum. This line of study
would be a fruitful area for future research.

34
References
Amihud, Y. 2002. Illiquidity and stock returns: cross-section and time-series effects. Journal of

Financial Markets, 5:31–56.

Ball, R. and Brown, P. 1968. An empirical evaluation of accounting income numbers. Journal of

Accounting Research, pages 159–178.

Barndorff-Nielsen, O. E., Hansen, P. R., Lunde, A., and Shephard, N. 2009. Realized kernels in

practice: trades and quotes. The Econometrics Journal, 12(3):1–32.

Barroso, P. and Santa-Clara, P. 2015. Momentum has its moments. Journal of Financial Economics,

116(1):111–120.

Bernard, V. L. and Thomas, J. K. 1989. Post-earnings-announcement drift: delayed price response

or risk premium? Journal of Accounting Research, pages 1–36.

Bollerslev, T., Li, S. Z., and Todorov, V. 2016. Roughing up beta: continuous vs. discontinu-

ous betas, and the cross-section of expected stock returns. Journal of Financial Economics,

120(3):464–490.

Carhart, M. 1997. On persistence in mutual fund performance. The Journal of Finance, 52:57–82.

Chabot, B., Ghysels, E., and Jagannathan, R. 2014. Momentum trading, return chasing, and

predictable crashes. Working paper, Federal Reserve Bank of Chicago, University of North

Carolina at Chapel Hill, and Northwestern University.

Chan, W. S. 2003. Stock price reaction to news and no-news: drift and reversal after headlines.

Journal of Financial Economics, 70(2):223–260.

Chordia, T., Roll, R., and Subrahmanyam, A. 2000. Commonality in liquidity. Journal of Financial

Economics, 56(1):3–28.

Chui, A. C., Titman, S., and Wei, K. J. 2010. Individualism and momentum around the world.

The Journal of Finance, 65(1):361–392.

35
Cohen, L. and Frazzini, A. 2008. Economic links and predictable returns. The Journal of Finance,

63(4):1977–2011.

Daniel, K., Grinblatt, M., Titman, S., and Wermers, R. 1997. Measuring mutual fund performance

with characteristic-based benchmarks. The Journal of Finance, 52(3):1035–1058.

Daniel, K., Hirshleifer, D., and Subrahmanyam, A. 1998. Investor psychology and security market

under-and overreactions. The Journal of Finance, 53(6):1839–1885.

Daniel, K. D. and Moskowitz, T. J. 2016. Momentum crashes. Journal of Financial Economics,

122(2):221–247.

DellaVigna, S. and Pollet, J. M. 2009. Investor inattention and Friday earnings announcements.

The Journal of Finance, 64(2):709–749.

Fama, E. F. 1970. Efficient capital markets: a review of theory and empirical work. The Journal

of Finance, 25(2):383–417.

Fama, E. F. 1998. Market efficiency, long-term returns, and behavioral finance. Journal of Financial

Economics, 49(3):283–306.

Fama, E. F. and French, K. R. 1993. Common risk factors in the returns on stocks and bonds.

Journal of Financial Economics, 33:3–56.

Fama, E. F. and French, K. R. 2015. A five-factor asset pricing model. Journal of Financial

Economics, 116(1):1–22.

Fama, E. F. and MacBeth, J. D. 1973. Risk, return, and equilibrium: empirical tests. Journal of

Political Economy, 81(3):607–636.

Gutierrez, R. C. and Kelley, E. K. 2008. The long-lasting momentum in weekly returns. The

Journal of Finance, 63(1):415–447.

Hasbrouck, J. 2009. Trading costs and returns for US equities: estimating effective costs from daily

data. The Journal of Finance, 64(3):1445–1477.

36
Heston, S. L., Korajczyk, R. A., and Sadka, R. 2010. Intraday patterns in the cross-section of stock

returns. The Journal of Finance, 65(4):1369–1407.

Hirshleifer, D., Lim, S. S., and Teoh, S. H. 2009. Driven to distraction: extraneous events and

underreaction to earnings news. The Journal of Finance, 64(5):2289–2325.

Jegadeesh, N. and Titman, S. 1993. Returns to buying winners and selling losers: implications for

stock market efficiency. The Journal of Finance, 48:65–92.

Jegadeesh, N. and Titman, S. 2011. Momentum. Annual Review of Financial Economics, 3(1):493–

509.

Jiang, H. and Sun, Z. 2015. News and corporate bond liquidity. Working paper, Michigan State

University and University of California at Irvine.

Kaul, G. and Nimalendran, M. 1990. Price reversals: bid-ask errors or market overreaction? Journal

of Financial Economics, 28(1):67–93.

Kelley, E. K. and Tetlock, P. C. 2017. Retail short selling and stock prices. The Review of Financial

Studies, 30(3):801–834.

Lehmann, B. N. 1990. Fads, Martingales, and Market Efficiency. Quarterly Journal of Economics,

105:1–28.

Lesmond, D. A., Schill, M. J., and Zhou, C. 2004. The illusory nature of momentum profits. Journal

of Financial Economics, 71(2):349–380.

Lewellen, J. 2002. Momentum and autocorrelation in stock returns. Review of Financial Studies,

15(2):533–564.

Lo, A. W. and MacKinlay, A. C. 1990. When are contrarian profits due to stock market overreac-

tion? Review of Financial Studies, 3(2):175–205.

Lou, D., Polk, C., and Skouras, S. 2017. A tug of war: overnight versus intraday expected returns.

Working paper, London School of Economics and Athens University.

37
Lucca, D. O. and Moench, E. 2015. The pre-FOMC announcement drift. The Journal of Finance,

70(1):329–371.

Moskowitz, T. J. and Grinblatt, M. 1999. Do industries explain momentum? The Journal of

Finance, 54(4):1249–1290.

Nagel, S. 2012. Evaporating liquidity. Review of Financial Studies, 25(7):2005–2039.

Novy-Marx, R. and Velikov, M. 2016. A taxonomy of anomalies and their trading costs. Review of

Financial Studies, 29(1):104–147.

Patton, A. J. and Verardo, M. 2012. Does beta move with news? Firm-specific information flows

and learning about profitability. Review of Financial Studies, 25(9):2789–2839.

Roll, R. 1984. A simple implicit measure of the effective bid-ask spread in an efficient market. The

Journal of Finance, 39(4):1127–1139.

Savor, P. 2012. Stock returns after major price shocks: the impact of information. Journal of

Financial Economics, 106(3):635–659.

Savor, P. and Wilson, M. 2013. How much do investors care about macroeconomic risk? Evi-

dence from scheduled economic announcements. Journal of Financial and Quantitative Analysis,

48(2):343–375.

Savor, P. and Wilson, M. 2016. Earnings announcements and systematic risk. The Journal of

Finance, 71(1):83–138.

Shleifer, A. and Vishny, R. W. 1997. The limits of arbitrage. The Journal of Finance, 52(1):35–55.

Tetlock, P. C. 2010. Does public financial news resolve asymmetric information? Review of

Financial Studies, 23(9):3520–3557.

Tetlock, P. C. 2011. All the news that’s fit to reprint: Do investors react to stale information?

Review of Financial Studies, 24(5):1481–1512.

Tetlock, P. C., Saar-Tsechansky, M., and Macskassy, S. 2008. More than words: quantifying

language to measure firms’ fundamentals. The Journal of Finance, 63(3):1437–1467.

38
Wang, Y., Zhang, B., and Zhu, X. 2016. The momentum of news. Working paper, Central University

of Finance and Economics, University of New South Wales, and Shanghai University of Finance

and Economics.

Wermers, R. 2003. Is money really ‘smart’ ? New evidence on the relation between mutual fund

flows, manager behavior, and performance persistence. Working paper, University of Maryland.

39
Figure 1: Timeline of News Momentum Strategy
This figure shows the timeline for the news momentum strategy. At the end (market close) of each day
t, we sort stocks into decile portfolios based on their news returns on day t (Ri,t,news ) and compute the
equal-weighted return of a self-financing portfolio that buys stocks in the top decile with high news returns
and sells stocks in the bottom decile with low news returns with a one-week holding period until day t + 5.
Following Jegadeesh and Titman (1993), our news momentum strategy includes portfolios with overlapping
holding periods. That is, we revise the weights on 15 of the securities in our news momentum strategy on
any given day and carry over the rest from the previous day.

40
Figure 2: Performance of News Momentum Strategy
This figure shows cumulative gains of the news momentum strategy (the blue solid line) and the performance
of the Jegadeesh and Titman (1993) momentum strategy (the red dotted line). At the end (market close) of
each day t, we sort stocks into decile portfolios based on their news returns on day t (Ri,t,news ) and compute
the equal-weighted return of a self-financing portfolio that buys stocks in the top decile with high news
returns and sells stocks in the bottom decile with low news returns with a one-week holding period until
day t + 5. Following Jegadeesh and Titman (1993), our news momentum strategy includes portfolios with
overlapping holding periods. That is, we revise the weights on 15 of the securities in our news momentum
strategy on any given day and carry over the rest from the previous day. Let Rwinner,t+1 and Rloser,t+1 be
the returns of the long and short legs of our news momentum strategy, respectively. The cumulative portfolio
value is computed as Wt+1 = Wt (1 + Rwinner,t+1 − Rloser,t+1 + Rrf,t+1 ) where Rrf,t+1 is the risk-free rate
on day (t + 1) and the initial investment is W1 = $1. Plotted is the time series of {Wt }. The return to the
Jegadeesh and Titman (1993) momentum strategy based on past one year return and one month holding
period comes from the data library of Ken French. The scale in the figure is based on the logarithm with
base 10.

41
Figure 3: Performance of News Momentum Strategy over Event Time
This figure shows the cumulative returns and their 95% confidence intervals of news momentum and non-
news reversal strategies for each event day. Specifically, at the end of each day t, we form decile portfolios
based on day t’s news returns Ri,t,news or non-news returns Ri,t,non-news and then compute the cumulative
overall returns from day t + 1 to day t + k for each decile portfolio and event day k. The spreads between
the cumulative returns in the highest and lowest deciles are the cumulative profits of winner-minus-loser
portfolios. Plotted are the average of the cumulative returns and its 95% confidence interval against the
event day k.

42
 (a) (b)

Figure 4: Overnight News Momentum Strategy over Event Time

Panel A shows the average gross cumulative returns to the news momentum strategies using the overnight
news and non-news returns and their 95% confidence intervals over event time. Specifically, at 10:00 a.m. on
each trading day t, we form decile portfolios based on the overnight news or non-news returns from the close
of day t − 1 to 9:45 a.m. of day t and then compute the cumulative overall returns every 30 minute from 10
a.m. on day t to 4:00 p.m. on day t + 4. The average difference in cumulative returns between the highest
and lowest deciles (the average cumulative return to the winner-minus-loser portfolio) is plotted against the
event day k, with the 95% confidence intervals. Panel B uses a return-matching approach to examine the
incremental value of news return, taking into account any mechanical return periodicity documented by
Heston et al. (2010). Specifically, at 10:00 a.m. on each trading day t and for every overnight news return,
we find a stock with a similar magnitude of non-news return (with the smallest absolute value in return
difference). Similar to the news momentum strategy, we construct a momentum strategy using this return-
matched sample. We plot the difference in average cumulative returns between the two strategies over event
time and the 95% confidence intervals.

43
Table 1: Descriptive Statistics
This table reports the descriptive statistics of our main variables. The sample consists of stocks listed on
NYSE/AMEX/NASDAQ with CRSP share code 10 or 11 for the period between March 2000 and October 2012, and with
prices above $1 at the end of the portfolio formation period. Panel A reports the time-series average of the cross-sectional
mean, standard deviation, and quantiles of each variable. Panel B reports the time-series average of the cross-sectional cor-
relations of these variables. Rnews (Rnon-news ) is the daily news (non-news) return aggregated from overnight and intraday
15-minute news (non-news) returns using transaction prices from TAQ and news releases from the RavenPack database. Size
is the product of the closing price and the number of shares outstanding, updated each day. BM is the book-to-market ratio
in June of year t, which is computed as the ratio of the book value of common equity in fiscal year t − 1 to the market value
of equity in December of year t − 1. RV is the daily realized variance, which is the sum of squared overnight and 15-minute
intraday returns from 4:00 p.m. on the previous day to 4:00 p.m. on the current day. Turnover is the total number of shares
traded in the prior five-day period over the average number of shares outstanding for the same period. ILLQ is the illiquidity
measure of Amihud (2002), which is the average daily ratio of the absolute stock return to the dollar trading volume over the
five-day period preceding each day. Mom is the cumulative returns from prior day 252 to day 21 for a given day t. Analyst is
analyst coverage, which is the number of sell-side analysts forecasting annual firm earnings in each month.

Panel A: Cross-Sectional Summary Statistics

Variable Mean Std P1 P25 Median P75 P99
Rnews 0.18% 4.33% -12.69% -0.79% 0.05% 1.00% 14.11%
Rnon-news 0.06% 3.31% -8.38% -1.37% -0.03% 1.34% 9.93%
Log(Size) 19.87 1.95 15.92 18.46 19.77 21.14 24.80
BM
√ 0.78 1.29 0.03 0.33 0.57 0.92 3.95
RV 3.37% 2.59% 0.81% 1.81% 2.65% 4.06% 13.14%
Turnover 0.01 0.02 0.00 0.00 0.00 0.01 0.06
ILLQ 0.32 3.19 0.00 0.00 0.01 0.05 6.10
Mom 6.84% 39.43% -99.32% -14.33% 7.19% 28.35% 110.51%
Analyst 6.89 6.78 0.00 1.54 4.84 10.06 28.49

Panel B: Cross-Sectional Correlations

√
Rnews Rnon-news Log(Size) BM RV Turnover ILLQ Mom Analyst
Rnews 1.000 0.062 -0.003 0.003 0.015 0.030 0.001 0.003 -0.003
Rnon-news 0.062 1.000 0.002 0.007 0.074 0.051 0.004 0.003 -0.009
Log(Size) -0.003 0.002 1.000 -0.224 -0.538 0.164 -0.252 0.051 0.770
BM
√ 0.003 0.007 -0.224 1.000 0.058 -0.077 0.133 0.041 -0.211
RV 0.015 0.074 -0.538 0.058 1.000 0.232 0.355 -0.042 -0.285
Turnover 0.030 0.051 0.164 -0.077 0.232 1.000 -0.077 0.034 0.207
ILLQ 0.001 0.004 -0.252 0.133 0.355 -0.077 1.000 -0.037 -0.154
Mom 0.003 0.003 0.051 0.041 -0.042 0.034 -0.037 1.000 -0.063
Analyst -0.003 -0.009 0.770 -0.211 -0.285 0.207 -0.154 -0.063 1.000

44
Table 2: Performance of the News Momentum Strategy
This table reports the performance of the news momentum strategy. We aggregate overnight and 15-minute news into daily
news returns following Equation 1. At the end (market close) of each day t, we sort stocks into ten portfolios based on their
news returns on day t (Ri,t,news ) and compute the equal-weighted return of a self-financing portfolio that buys stocks in the
top decile with high news returns and sells stocks in the bottom decile with low news returns with a one-week holding period
until day t + 5. Following Jegadeesh and Titman (1993), our news momentum strategy includes portfolios with overlapping
1
holding periods. That is, we revise the weights on 5
of the securities in our news momentum strategy on any given day and
carry over the rest from the previous day. We compute daily holding-period returns using transaction prices at 4:00 p.m.
from the TAQ database. Panel A summarizes the portfolio returns in monthly percentage. The rows labeled “Return” and
‘FFC4” respectively report the average raw returns and Fama-French-Carhart four-factor alphas for each portfolio. The column
labeled “10–1” reports the difference in returns between Portfolio 10 and Portfolio 1, with Newey-West robust t-statistics in
parentheses. Panels B reports the loadings on the four-factor models for the decile and spread portfolios. Our sample includes
stocks listed on NYSE/AMEX/NASDAQ over the period between March 2000 and October 2012 with prices above $1 at the
end of the portfolio formation period. We multiply daily returns by 21 to get monthly returns in percentage.

1 2 3 4 5 6 7 8 9 10 10–1
Panel A: Portfolio Returns and Alphas
Return -0.78 0.40 0.87 1.07 1.10 1.31 1.37 1.45 1.67 2.55 3.34
(-1.21) (0.64) (1.51) (1.98) (2.11) (2.55) (2.55) (2.57) (2.78) (4.07)(11.72)
FFC4 -1.11 0.05 0.50 0.70 0.73 0.91 0.98 1.06 1.32 2.26 3.37
(-4.44) (0.27) (3.63) (5.50) (5.98) (7.40) (7.28) (7.53) (7.59) (9.44)(11.76)
Panel B: Portfolio Betas from the Fama-French-Carhart Model
MKT 1.00 1.03 1.00 0.96 0.94 0.92 0.96 0.99 1.01 0.99 -0.01
(60.62)(109.88)(130.01)(118.03)(109.04)(126.18)(118.71)(126.85) (95.72)(76.04) (-0.58)
SMB 0.73 0.60 0.48 0.40 0.37 0.39 0.42 0.53 0.63 0.74 0.00
(23.70) (31.37) (25.87) (23.23) (21.04) (22.90) (22.98) (33.40) (30.22)(26.67) (0.13)
HML 0.05 0.11 0.16 0.18 0.20 0.24 0.20 0.17 0.09 -0.01 -0.06
(1.56) (4.72) (10.12) (11.00) (13.20) (13.32) (12.99) (8.45) (5.08) (-0.19) (-1.32)
UMD -0.25 -0.21 -0.14 -0.10 -0.06 -0.06 -0.07 -0.10 -0.15 -0.19 0.06
(-11.20) (-15.43) (-13.16) (-9.34) (-5.60) (-5.34) (-6.42) (-8.15)(-12.10) (-9.97) (2.44)

45
Table 3: Fama-MacBeth (1973) Regressions
This table reports the estimated regression coefficients and Newey-West t-statistics (in parentheses) from Fama-MacBeth cross-
sectional regressions predicting five-day ahead stock returns using news and non-news returns in the past day. The sample
consists of stocks listed on NYSE/AMEX/NASDAQ for the period between March 2000 and October 2012 with prices above $1
as of the portfolio formation. Rnews (Rnon-news ) is the daily news (non-news) returns that are aggregated from overnight and
intraday 15-minute news (non-news) returns based on transaction prices computed from merging TAQ and RavenPack news
database. Size is the product of the closing price and the number of shares outstanding and is updated daily using the daily
data of a firm. BM is the book-to-market ratio in June of year t which is computed as the ratio of the book value of common
equity in fiscal year t − 1 to the market value of equity (size) in December of year t − 1 and is updated every July. RV is the
daily realized variance which is the sum of squared overnight and 15-minute intraday returns from 4:00 p.m. on the previous
day to 4:00 p.m. on the current day. ILLQ is the illiquidity measure of Amihud (2002) which is the average daily ratio of the
absolute stock return to the dollar trading volume over the five-day period preceding each day. Mom is the cumulative returns
from prior day 252 to day 21 for a given day t and is updated daily.

(I) (II) (III) (IV) (V)

Intercept 0.0024 0.0023 0.0023 0.0039 0.0050
(2.27) (2.20) (2.17) (2.77) (2.80)
Rnews 0.0425 0.0574 0.0614 0.0641
(8.28) (11.02) (10.64) (11.45)
Rnon-news -0.0792 -0.0809 -0.0737 -0.0427
(-24.29) (-24.80) (-22.39) (-11.59)
Dummy(No News) 0.0001 0.0000 0.0001 -0.0007 -0.0006
(0.50) (0.01) (0.33) (-4.06) (-4.50)
Log(Size) -0.0003 -0.0005
(-2.46) (-2.02)
BM 0.0007 0.0010
(2.96) (2.83)
Mom 0.0006 0.0008
(0.86) (1.07)
RV -0.0481
(-0.56)
ILLQ -0.0001
(-0.65)
Adj-R2 (%) 0.11 0.66 0.75 3.06 5.20
#Obs 10,097,270 10,097,270 10,097,270 8,674,126 6,245,682

46
Table 4: Investment Value of News Momentum
This table reports descriptive statistics of daily returns of six investment strategies: the return of the news momentum strategy
(NEWS); the excess return on the market (MKT); the average return on small-sized firms minus the average return on big-sized
firms (SMB); the average return on high book-to-market ratio firms minus that on low book-to-market ratio firms (HML); the
average return on the high prior 2 to 12 month return portfolios minus that on the low prior 2 to 12 month return portfolio
(UMD); and the average return on the low prior month return portfolio minus that on the high prior month return portfolio
(REV). Daily returns of MKT, SMB and HML, UMD, and REV come from Kenneth French’s website. All return numbers
are in monthly percentage (multiplying the percentage daily returns by 21). Panel A summarizes the average monthly return
(Mean), standard deviation (SD), Sharpe ratio (SR), skewness, kurtosis, and maximum drawdown (MaxDD) of each portfolio.
Panel B reports the time-series correlation coefficients among the six strategies. Panel C reports the performance measures of
the two tangency (maximum Sharpe ratio) portfolios from the two investment opportunity sets: one with the five traditional
strategies and the other with the addition of the news momentum strategy.

Panel A: Strategy Performance

Mean SD SR Skewness Kurtosis MaxDD
NEWS 3.34 3.51 0.95 0.51 12.14 15.92
MKT 0.20 6.26 0.03 0.00 9.54 59.20
SMB 0.16 2.86 0.06 -0.37 7.75 27.11
HML 0.64 3.09 0.21 0.02 8.73 28.63
UMD 0.11 5.08 0.02 -0.79 10.63 63.66
REV 1.31 4.85 0.27 1.44 19.30 24.67

Panel B: Time-Series Correlations

NEWS MKT SMB HML UMD REV
NEWS 1.00 -0.05 0.02 -0.07 0.11 -0.07
MKT -0.05 1.00 0.12 0.00 -0.38 0.39
SMB 0.02 0.12 1.00 -0.15 0.12 0.03
HML -0.07 0.00 -0.15 1.00 -0.23 -0.18
UMD 0.11 -0.38 0.12 -0.23 1.00 -0.11
REV -0.07 0.39 0.03 -0.18 -0.11 1.00

Panel C: Tangency Portfolio Performance

Mean SD SR Skewness Kurtosis MaxDD
Excluding News Momentum 0.73 1.83 0.40 0.71 13.14 11.79
Including News Momentum 2.19 2.05 1.07 0.89 11.77 6.06

47
Table 5: Gross Return, Transaction Cost and Net Return of the News Momentum Strategy
This table reports the daily gross return, transaction cost, and net return of the news momentum strategy in basis points. Daily
news returns are aggregated from overnight and 15-minute news returns computed from matching the transaction prices in the
TAQ database and the news stories in the RavenPack news databases. Daily overall returns are computed using the transaction
prices at 4:00 p.m. from the TAQ database. At the end (market close) of each day t, we sort stocks into ten portfolios based
on their news returns on day t (Ri,t,news ) and compute the average portfolio return and proportional effective spread (PES)
return of a self-financing portfolio that buys stocks in the top decile with high news returns and sells stocks in the bottom
decile with low news returns with a one-week holding period until day t + 5. Following Jegadeesh and Titman (1993), our news
1
momentum strategy includes portfolios with overlapping holding periods. That is, we revise the weights on 5
of the securities
in our news momentum strategy on any given day and carry over the rest from the previous day. Thus, we assume a daily
turnover of 20% for estimating the transaction cost. Panels A and B respectively summarize the daily equal- and value-weighted
portfolio returns and proportional effective spreads in basis points. The rows labeled “Gross Return”, “PES based on 20%
turnover” and ‘Net Return” respectively report the average unadjusted returns, PES, and net returns for each portfolio. The
column labeled “10–1” reports the difference in returns between Portfolio 10 and Portfolio 1. Our sample includes stocks listed
on NYSE/AMEX/NASDAQ over the period between March 2000 and October 2012 with prices above $1 at the end of the
portfolio formation period.

1 2 3 4 5 6 7 8 9 10 10–1
Panel A: Equal-Weighted Portfolios
Gross Return -3.7 1.9 4.1 5.1 5.2 6.3 6.5 6.9 7.9 12.2 15.9
PES based on 20% turnover 10.5 7.3 5.7 4.8 4.5 4.7 5.1 6.1 7.8 11.3 21.7
Net Return -5.8

Panel B: Value-Weighted Portfolios

Gross Return -2.5 2.2 1.6 3.1 1.6 2.5 2.8 2.8 3.8 9.2 11.7
PES based on 20% turnover 4.0 2.6 2.1 1.9 1.8 1.9 1.9 2.2 2.8 4.3 8.3
Net Return 3.5

48
Table 6: Decomposing the Profits of the News Momentum Strategy
This table reports the Lo and MacKinlay (1990) decomposition of the news momentum strategy. For the individual stock news
momentum, we use the 970 stocks that have complete return observations over the period of March 2000 and October 2012.
For the industry news momentum, we form equal-weighted industry portfolios based on the Fama and French (1997) industry
classification. The column labeled “Auto” is the first component in Eq. (4), capturing the autocovariance in stock returns;
“Cross” is the second component, capturing the cross-autocovariance; “Dispersion” is the third component, representing the
dispersion in expected stock returns captured by news returns; and “Total” is the total return to the news momentum strategy.
Panel A reports the estimates based on raw returns. Panel B reports the estimates based on alphas from the four-factor model.
All return numbers are in monthly percentage, by multiplying daily returns by 21.

Panel A: Raw Return Panel B: FFC4 Adjustment

Auto Cross Dispersion Total Auto Cross Dispersion Total
3.79 -0.05 0.04 3.79 3.30 -0.02 0.06 3.34
Individual Stocks
(5.29) (-0.16) (6.03) (5.35) (-0.51) (5.41)

2.07 -0.98 0.01 1.09 1.87 -0.98 0.02 0.91

Industry Portfolios
(2.46) (-1.27) (4.92) (6.49) (-5.21) (4.49)

49
Table 7: Performance of the News Momentum Strategy: Double Sorts
This table reports the performance of portfolios sorted on news momentum and firm characteristics for the period between March
2000 and October 2012. At the end of each day t, we independently sort stocks into 3 portfolios based on news returns and 3
portfolios based on one the firm characteristics. Then we consider the performance of the 9 portfolios from the intersection of
the double sorts. Similar to Table 2, we use portfolios with overlapping holding periods. Our set of firm characteristics includes
the following. Size is the product of the closing price and the number of shares outstanding updated daily using the daily data of
a firm. Analyst Coverage is analyst coverage which is the monthly number of sell-side analysts forecasting annual firm earnings
from I/B/E/S. Volatility is the daily realized variance which is the sum of squared overnight and 15-minute intraday returns
from 4:00 p.m. on the previous day to 4:00 p.m. on the current day. ILLQ is the illiquidity measure of Amihud (2002) which
is the average daily ratio of the absolute stock return to the dollar trading volume over the five-day period preceding each day.
Mom is the cumulative returns from prior day 252 to day 21 for a given day t and is updated daily. In each panel, the first three
rows (columns) represent three levels of the control variable (the news return). The row (column) labeled “Diff” (“High–Low”)
reports the difference in returns between Portfolio 3 and Portfolio 1 constructed according to the control variable (the news
return). The column labeled “FFC4” reports the Fama-French-Carhart four-factor alphas. The corresponding Newey-West
t-statistics are reported in parentheses. We report returns in monthly percentage.

1 2 3 High–Low FFC4 1 2 3 High–Low FFC4

Panel A: Size Panel B: Size-Adjusted Analyst Coverage
Small 0.21 1.87 2.38 2.17 2.19 Low 0.55 1.18 2.26 1.71 1.72
(0.36) (3.39) (4.20) (9.12) (9.05) (0.90) (2.22) (3.72) (7.39) (7.42)
2 0.30 1.18 1.75 1.45 1.45 2 1.17 1.47 2.48 1.31 1.32
(0.46) (2.09) (2.68) (7.23) (7.28) (1.98) (2.84) (4.28) (6.77) (6.86)
Large 0.38 0.80 0.92 0.54 0.51 High 1.10 1.37 2.16 1.06 1.05
(0.63) (1.61) (1.58) (3.03) (2.89) (1.74) (2.54) (3.50) (6.30) (6.27)
Diff 0.17 -1.07 -1.46 -1.63 -1.68 Diff 0.55 0.18 -0.10 -0.65 -0.67
(0.50) (-3.58) (-4.71) (-5.60) (-5.60) (2.09) (0.90) (-0.40) (-2.43) (-2.50)

Panel C: Size-Adjusted Volatility Panel D: ILLQ

Low 0.85 1.12 1.86 1.01 1.01 Low 0.50 0.88 0.78 0.28 0.27
(1.81) (2.67) (4.03) (4.94) (4.95) (0.83) (1.78) (1.34) (1.57) (1.54)
2 0.66 1.04 1.93 1.27 1.28 2 0.25 1.19 1.48 1.23 1.21
(1.13) (1.84) (3.37) (6.85) (6.75) (0.38) (2.16) (2.32) (6.06) (6.08)
High -0.12 0.82 1.70 1.82 1.84 High 0.25 1.57 2.20 1.94 1.95
(-0.16) (1.07) (2.26) (8.28) (8.37) (0.43) (2.87) (3.87) (9.64) (9.67)
Diff -0.97 -0.30 -0.16 0.81 0.83 Diff -0.25 0.70 1.42 1.66 1.68
(-2.23) (-0.65) (-0.38) (2.82) (2.88) (-0.80) (2.61) (5.03) (6.59) (6.65)

Panel E: Mom
Loser 0.12 1.25 1.83 1.71 1.72
(0.17) (1.96) (2.65) (7.90) (8.00)
2 0.26 1.08 2.08 1.82 1.83
(0.47) (2.20) (3.88) (10.02) (9.84)
Winner 0.48 1.18 1.60 1.12 1.12
(0.81) (2.23) (2.71) (5.38) (5.36)
Diff 0.36 -0.07 -0.23 -0.60 -0.60
(0.88) (-0.19) (-0.58) (-2.07) (-2.07)

50
Table 8: Robustness Tests
This table reports the performance of the news momentum strategy under different designs. The main sample consists of stocks
listed on NYSE/AMEX/NASDAQ over the period between March 2000 and October 2012 with prices above $1 at the end of
the portfolio formation period. The rows labeled “Return” and ‘FFC4 alpha” respectively report the average return and Fama-
French-Carhart four-factor alpha for each decile portfolio in monthly percentage. The column labeled “High–Low” reports the
difference in returns between Portfolio 5 and Portfolio 1, with Newey-West robust t-statistics in parentheses. In Panel A, we
use the daily returns from the CRSP database to compute both the daily news and overall returns and then repeat the news
momentum strategy using these returns. In Panel B, we use the high-frequency mid-quote prices to aggregate overnight and
15-minute returns into the daily news and overall returns and then repeat the news momentum strategy on the basis of these
quoted returns. In Panel C, we aggregate 15-minute and overnight news returns over the period of 9:45 a.m. on day t − 1 to
9:45 a.m. on day t and forecast the five-day ahead overall returns over the period of 10:00 a.m. on day t to 10:00 a.m. on day
t + 5. In Panel D, we use the market-risk-adjusted returns to construct trading signal and repeat the news momentum strategy.
In Panel E, we first implement the characteristic-based benchmark methods of Daniel et al. (1997) and Wermers (2003) to
adjust risks. We use the benchmark portfolio assignments to compute the daily equal-weighted 5 × 5 × 5 size, book-to-market
ratio, and momentum benchmark returns based on all NYSE/AMEX/NASDAQ data in the CRSP database. We then subtract
a firm’s daily overall return by the daily return of one of the 125 benchmarks to which the firm belongs to on that day. We
use these benchmark-adjusted returns in place of the raw overall returns when repeating the news momentum strategy. In
Panel F, we first identify quarterly earnings announcements using dates from Compustat. Since the time of the day of the
earnings announcement is unavailable in Compustat and earnings announcement can occur before, during, or after the regular
trading hours, we are unable to match the returns that immediately reflect the information on earnings announcement. To
conservatively remove the effect of earnings announcement, we thus exclude from the samples both the earnings announcement
day and the day after the announcement. In Panel G, we first identify events of information-based major price changes by
following Savor (2012) and then exclude from firm-date samples those observations corresponding to information-based major
price changes. In Panel H, we remove the effect of news clustering by only forecasting the non-news-driven returns.

1 2 9 10 High–Low 1 2 9 10 High–Low
Panel A: CRSP Return Panel B: Mid-Quote Return
Return 0.00 0.85 1.33 1.98 1.99 Return -0.87 0.55 1.76 2.52 3.40
(-0.00) (1.35) (2.27) (3.15) (6.01) (-1.35) (0.91) (2.98) (4.12) (12.55)
FFC4 -0.35 0.48 0.96 1.63 1.98 FFC4 -1.18 0.20 1.41 2.24 3.42
(-1.20) (2.66) (5.73) (6.92) (6.05) (-4.84) (1.20) (8.02) (9.80) (12.75)

Panel C: Open-to-Open Return Panel D: Market-Risk-Adjusted Returns

Return -0.60 0.57 1.98 3.33 3.92 Return -0.8 0.58 1.76 2.49 3.29
(-0.90) (0.95) (3.37) (5.33) (12.91) (-1.27) (0.95) (2.97) (4.04) (12.26)
FFC4 -0.68 0.46 1.88 3.26 3.94 FFC4 -2.23 -0.86 0.31 1.1 3.33
(-1.09) (0.82) (3.39) (5.48) (13.03) (-10.08) (-4.70) (1.75) (5.20) (12.30)

Panel E: DGTW Characteristic Adjustment Panel F: Eliminating Earnings Announcements

Return -1.68 -0.53 0.67 1.39 3.06 Return 0.20 0.71 1.45 1.99 1.79
(-6.50) (-2.79) (3.67) (5.53) (11.14) (0.29) (1.13) (2.35) (3.03) (5.76)
FFC4 -1.64 -0.53 0.67 1.45 3.09 FFC4 -0.17 0.38 1.10 1.68 1.85
(-7.42) (-3.54) (4.39) (6.88) (11.24) (-0.65) (2.23) (6.05) (6.62) (5.92)

Panel G: Eliminating Extreme Price Changes Panel H: Excluding News Clustering

Return -0.57 0.40 1.64 2.36 2.93 Return -0.92 0.25 1.29 1.92 2.85
(-0.91) (0.65) (2.77) (3.89) (10.75) (-1.49) (0.43) (2.24) (3.17) (10.51)
FFC4 -0.92 0.03 1.28 2.05 2.97 FFC4 -1.24 -0.09 0.95 1.63 2.88
(-3.92) (0.16) (7.63) (8.90) (10.70) (-5.26) (-0.53) (5.69) (7.07) (10.53)

51
Table 9: Weekend News
This table shows how news momentum return varies over each day of the week. Specifically, we construct a news momentum
strategy that buys (sells) stocks with the highest (lowest) 10% news return in the previous day with a one-day holding period.
Then we regress the resulting daily returns in basis points on five dummy variables that represent each day of the week without
an intercept in Regression (I), or a dummy variable of Monday with an intercept in Regression (II). The coefficient on the
Monday dummy is of particular interest, because the return on Monday to the news momentum strategy is based on news
arriving over Friday and the weekend.

(I) (II)
Intercept 41.69
(12.16)
Mondays 64.69 23.00
(8.82) (2.84)
Tuesdays 28.23
(3.80)
Wednesdays 45.97
(7.09)
Thursdays 41.99
(6.04)
Fridays 50.61
(7.79)

52
Table 10: Performance of Chan (2003)’s Strategy
This table reports the performance of the Chan (2003)’s strategy. At the end of each month, we consider a news group consisting
of all stocks that have at least one news story during that month and then sort them into ten portfolios based on their monthly
return in order to compute the equal-weighted return of a self-financing portfolio that buys stocks in the top decile with high
returns and sells stocks in the bottom decile with low returns with K = 1, 3, and 6 month holding periods. Following Jegadeesh
and Titman (1993), this Chan (2003)’s strategy includes portfolios with overlapping holding periods. That is, we revise the
1
weights on K
of the securities in our news momentum strategy in any given month and carry over the rest from the previous
month. We compute monthly holding-period returns using transaction prices at 4:00 p.m. from the TAQ database. The rows
labeled “Return” and “FFC4” respectively report the average raw returns and Fama-French-Carhart four-factor alphas for each
portfolio. The column labeled “10–1” reports the difference in returns between Portfolio 10 and Portfolio 1, with Newey-West
robust t-statistics in parentheses.

1 2 3 4 5 6 7 8 9 10 10–1
K=1
Return 2.15 1.38 1.27 1.17 1.06 1.11 1.07 1.14 0.95 1.39 -0.76
(3.15) (2.53) (2.82) (2.82) (2.74) (2.95) (2.85) (2.98) (2.19) (2.49) (-1.43)
FFC4 1.78 0.98 0.86 0.75 0.60 0.69 0.65 0.74 0.48 0.94 -0.85
(5.30) (4.50) (5.46) (5.71) (5.46) (6.52) (6.26) (6.84) (3.31) (4.06) (-1.74)

K=3
Return 1.75 1.27 1.17 1.07 1.04 1.02 0.97 0.99 1.01 1.28 -0.47
(2.71) (2.47) (2.65) (2.65) (2.72) (2.77) (2.60) (2.58) (2.46) (2.71) (-1.52)
FFC4 1.37 0.85 0.74 0.67 0.60 0.60 0.56 0.58 0.60 0.93 -0.44
(6.60) (7.16) (8.10) (7.97) (7.59) (8.02) (7.76) (7.60) (5.85) (6.68) (-1.85)

K=6
Return 1.56 1.18 1.13 1.05 1.02 1.02 1.03 1.02 1.10 1.33 -0.23
(2.69) (2.42) (2.60) (2.64) (2.69) (2.73) (2.73) (2.65) (2.65) (2.79) (-1.11)
FFC4 1.17 0.75 0.68 0.62 0.56 0.57 0.57 0.58 0.65 0.90 -0.27
(7.04) (7.60) (8.80) (8.68) (8.33) (8.80) (9.78) (8.77) (8.75) (8.14) (-1.86)

53