Investment Management

Momentum strategies

Yan-Shing Chen
Spring, 2024

1
 Outline
• Momentum effect and its sources
– Jegadeesh and Titman (1993)
• Past winners outperform past losers
– Chan, Jegadeesh, and Lakonishok (1996)
• Compare earnings vs. price momentum
– Jegadeesh and Titman (2001)
• Delayed over-reaction to information drives the momentum
effect
– Carhart (1997)
• Use momentum effect to explain the one-year persistence of
mutual fund performance
• Four-factor model (market, size, B/M, and momentum)
– Daniel and Moskowitz (2016)
• When and why does momentum crash?
2
 Jegadeesh and Titman (1993)
• Trading strategies (J-month/K-month)
– At the beginning of each month t, rank securities in
ascending order on the basis of their returns in the past J
months (J = portfolio formation period)
– Form ten portfolios
• Top decile (P1): Losers
• Bottom decile (P10): Winners
• Stocks are equally weighted within each decile
– In each month t from 1965 through 1989
• Buy the winner portfolio and sell the loser
• Hold this position for K months (K = holding period)

3
– Closes out the position initiated in month t – K
• That is, revise the weights on 1/K of the securities in
the entire portfolio in any given month and carry over
the rest from the previous month
– Portfolios are rebalanced monthly to maintain equal
weights
– 16 strategies (J or K = 3, 6, 9, 12)

4
• Example of overlapping holding period (J = 6, K = 3)
– Each month, 1/3 of the holdings is revised

5
• Findings
– The returns of all the zero-cost (buy-sell) portfolios (i.e.,
the returns per dollar long in this portfolio) are positive
– All these returns are statistically significant except for the
3-month/3-month strategy that does not skip a week
– The 6-month formation period produces returns of about
1% per month regardless of the holding period
– Most successful zero-cost strategy: J = 12/K = 3
– These holding period returns are slightly higher when
there is a 1-week lag
• Skip a week between J and K helps avoid some of the
bid-ask spread, price pressure, and lagged reaction
effect (very short-term reversal)

6
Skip a week

7
• Decomposition of momentum profits
– Single factor model
– μi = the (unconditional) expected return of security i
– ft = the (unconditional) unexpected return on a factor-
mimicking portfolio
– bi = factor sensitivity
– eit = the firm-specific component

8
– The three sources of momuntum
• Stocks have different expected return μi. Winners
(losers) are those with higher (lower) systematic risk
• Factor return ft is serially correlated. Momentum profits
will partly comes from picking stocks with a high b
when the conditional expectation of the factor portfolio
return is high
• Firm-specific risk eit is serially correlated due to
investors’ unreaction to firm-specific information

9
• Empirical results
– The beta of the portfolio of past losers (1.36) is higher than
the beta of the portfolio of past winners (1.28)
• Momentum profits are not driven by the differential
exposure to systematic risk
– The serial covariance of 6-month returns of the equally
weighted index is negative (–0.0028)
• Serial covariance of factor portfolio returns is unlikely
to be the source of momentum profits
– The estimates of the serial covariance of market model
residuals for individual stocks are on average positive
(0.0012)
• Momentum profits arise from investors’ under-reaction
to firm-specific information

10
– Risk-adjusted returns (α from the following model)

– rmt is the return on the value-weighted market index

– α (P10 – P1) are all positive and significant across all size
and beta subsamples
– Abnormal performance mainly comes from buy side
transactions
• α from losers are negative but not significant

11
12
• Profit after accounting for transaction costs
– On average, the momentum trading rule results in a
turnover of 84.8% semiannually
– Assume a 0.5% one-way transaction cost
– The risk-adjusted return net of transaction cost is 9.29%
per year, which is reliably different from zero

13
• Calendar month returns
– P10 – P1 loses about 7% on average in each January, and
earn 1.66% on average in other months
• Losers perform better in January because investors sell
losers in the year end to save tax
– P10 – P1 realizes positive returns in 67% of the months,
and 71% of the months when January is excluded
– The negative January performance concentrated on small
firms
– The returns are particularly high in April
• Corporations must transfer money to their pension
funds prior to April 15
• Pension fund assets are primarily invested by portfolio
managers who follow the momentum rules
14
– Returns on P10 – P1 by calendar months (6/6)

15
– Proportion of positive returns (6/6)

16
• Performance in event time
– Track the average portfolio returns in each of the 36
months following the portfolio formation date
– With the exception of month 1, the average return in each
month is positive in the first year
– The average return is negative in each month in year 2 as
well as in the first half of year 3 and virtually zero
thereafter
– The cumulative returns reach a maximum of 9.5% at the
end of 12 months but decline to about 4% by the end of
month 36 (long-run reversal)

17
18
• Stock returns around earnings announcement dates
– The 3-day returns (days –2 to 0) of the individual stocks in
P1 and P10 groups are calculated around each of their
quarterly earnings announcements that occur within 36
months after the portfolio formation date
– In the first 6 months, the announcement date returns of the
past winners exceed the announcement date returns of the
past losers by over 0.7% on average
• Represents about 25% of the zero-cost portfolio returns
over this holding period (2 announcements × 0.7%/5.1%)
– From months 8 through 20 the differences in
announcement date returns are negative
– How to interpret the short-run momentum and long-run
reversal?
19
20
 Chan, Jegadeesh, and Lakonishok (1996)
• Earnings and price momentums coexist
– Earnings momentum strategy
may benefit from
underreaction to information
related to short-term
earnings
– Price momentum strategy
may benefit from the
market’s slow response to a
broader set of information
about future profitability
Post-earning announcement price drift (PEAD)

21
• Sample and methodology
– Stocks listed on the New York (NYSE), American (AMEX),
and Nasdaq stock markets
– At the beginning of every month from January 1977 to
January 1993, rank stocks on the basis of either past
returns or a measure of earnings news
– The ranked stocks are then assigned to one of ten decile
portfolios, where the breakpoints are based only on NYSE
stocks
– All stocks are equally weighted within a given portfolio

22
• Ranking variables
– Price momentum: Compound returns over the six months
prior to portfolio formation

– Earnings momentum (1):

– SUE (standardized unexpected earnings): EPS most
recently announced as of month t minus EPS four quarters
ago, divided by the standard deviation of unexpected
earnings over the preceding eight quarters

– Measures earnings surprise over a longer period

23
– Earnings momentum (2):
– ABR: Cumulative abnormal stock return around the most
recent announcement date of earnings up to month t

rmj is the return on the equally-weighted market index

– Captures the change over a window of only a few days in

the market’s views about earnings

24
• Methodology
– For each of our momentum strategies, report buy-and-
hold returns in the periods subsequent to portfolio
formation
– Skip the first five days after portfolio formation
– If a stock is delisted after it is included in a portfolio but
before the end of the holding period over which returns
are calculated, we replace its return until the end of the
period with the return on a value-weighted market index
– At the end of the period we rebalance all the remaining
stocks in the original portfolio to equal weights in order to
calculate returns in subsequent periods

25
• Price momentum
– Winners outperform losers by 8.8% in the first six months,
and by 15.4% in the first year

26
– Market continues to be caught by surprise at the two quarterly
earnings announcements following portfolio formation (slow
responses to information in past returns)

27
• Earnings momentum (SUE)
– Winners outperform losers by 6.8% in the first six months, and
by 7.5% in the first year

28
– Past SUE contains information that is not incorporated into the
stock price
– No significant difference after the third announcement

29
• Earnings momentum (ABR)
– Winners outperform losers by 5.9% in the first six months
and by 8.3% in the first year

30
– Past ABR contains information that is not incorporated into the
stock price

31
• Summary of findings
– Sorting stocks on the basis of past returns and earnings
surprise yield large differences in subsequent returns
– The profitability of earnings momentum strategies is
smaller and less persistent than that of price momentum
strategy
– Market does not incorporate the news in past prices or
earnings promptly, so that there are drifts in subsequent
returns

32
• Two-way independent sort (ABR and R6)
– 3-3 outperforms 1-1 by 7.9 percent in the first six months
– Controlling for R6, ABR positively predicts return

33
– Controlling for ABR, R6 positively predicts return
– Each variable has predictive power given the other variable

34
• Two-way independent sort (SUE and R6)
– 3-3 outperforms 1-1 by 8.1 percent in the first six months
– Each variable has predictive power given the other variable

35
• Is there any contradiction between the stories
behind the momentum and contrarian strategies?
– Momentum: Buy past winners and sell past losers
• Market digests new information slowly

– Contrarian: Buy past losers and sell past winners

• Investors extrapolate past performance too far into the
future
– Are the winners and losers the same in the two strategies?
(see page 1711 of the paper)
– Is there any common element in the two stories?

36
 Jegadeesh and Titman (2001)
• How to explain the short-run momentum and long-
run reversal?
– Delayed overreaction to information
• Conservatism bias + representative heuristic
• Self-attribution bias (overconfidence)
• Rational investors with partial information

39
• Portfolio formation
– Sample period: 1965–1997
– Exclude all stocks priced below $5 at the beginning of the
holding period and all stocks in the smallest NYSE size
decile
– Overlapping portfolios (J = K = 6)
• A momentum decile portfolio in any particular month
holds stocks ranked in that decile in any of the previous
six ranking months
• December winner portfolio comprises 10 percent of the
stocks with the highest returns over the previous June
to November period, the previous May to October, and
so on up to the previous January to June period
– Each monthly cohort is assigned an equal weight in this
portfolio 40
• Monthly returns for momentum portfolios

41
• Portfolio characteristics
– Both winners and losers tend to be smaller firms
– Losers are more sensitive to all three Fama-French factors

42
• Risk-adjusted returns (α) are positive

43
• Post-holding period returns
– A dramatic reversal of returns in the second through fifth
years over 1965–1981

44
• Delayed overreaction to information
– Explanation 1: Conservatism bias + representative heuristic
• Barberies et al. (1998)
– Conservatism bias leads to initial underreaction
– Representative heuristic leads to long-run reversal
• Investors mistakenly conclude that firms realizing
extraordinary earnings growths will continue to
experience similar extraordinary growth in the future
• Push stock prices of winners above the fundamental
value

45
– Explanation 2: overconfidence and self-attribution bias
• Daniel, Hirshleifer, and Subrahmanyam (1998)
– The overconfident-informed traders overweight the private
signal relative to the prior (underestimate the error variance in
the signal) (date 1), causing the stock price to overreact
– When a later public signal confirms the trade (good news arrives
after a buy, or bad news after a sell) (date 2), the investor’s
confidence rises further due to self-attribution bias, but
disconfirming information causes confidence to fall only
modestly
• Thus, public information can trigger further overreaction to a
preceding private signal
– Such continuing overreaction causes momentum in security
prices, but that such momentum is eventually reversed as
further public information (date 3) gradually draws the price
back toward fundamentals
– Thus, biased self-attribution implies short-run momentum and
long-term reversals 46
 Solid line: no overconfidence
Solid bold line: overconfidence
Dash line: overconfidence + self attribution

Date 1: informed investors receive a noisy but unbiased private signal

Date 2: a nosiy but unbiased public signals arrives
Date 3: true value reveals 47
Momentum returns by country (Chui, Titman, Wei, 2010)
Winner (loser): prior 6 month return in top (bottom) 30%

48
Momentum does not exist in most southern and eastern Asian coutries, in which
people show less individualism and overcondifence
49
– Explanation 3: Rational investors with partial information
• Hong and Stein (1999)
– Two groups of investors who trade based on different sets
of information
– “Informed traders” obtain signals about future cash flows
but ignore information in the past history of prices
– “Technical traders” trade based on a limited history of
prices and do not observe fundamental information
– The information obtained by informed traders is
transmitted with a delay and cause underreaction
– The technical traders extrapolate based on past prices and
tend to push prices of past winners above their
fundamental values

50
 Carhart (1997)
• A four factor model

– PR1YR is the equal-weight average of firms with the

highest 30 percent eleven-month returns lagged one
month minus the equal-weight average of firms with the
lowest 30 percent eleven-month returns lagged one
month
– The portfolios include all NYSE, Amex, and Nasdaq stocks
and are re-formed monthly

51
• How to explain short-term mutual fund performance
persistence?
– Superior stock-picking skill of fund managers (X)
– Common factor (monentum) in stock returns (O)
– Persistent difference in mutual fund expenses and
transaction costs (O)

52
• Summary of findings
– Short-run mutual fund returns persist strongly
– Most of the persistence is explained by common-factor
sensitivities, expenses, and transaction costs
– The net gain in returns from buying the decile of past
winner funds and selling the decile of losers is 8 percent
per year
– 4.6 percent can be explained with size, book-to-market
and one-year momentum in stock returns; 0.7 percent
with expense ratios; and 1 percent with transaction costs

53
• Methodology
– On January 1 of each year, form ten equal-weighted
portfolios of mutual funds based on the previous year
reported returns
• Reported returns are net of all operating expenses and
security-level transaction costs, but do not include sales
charges
– Hold the portfolios for one year, then re-form them
• Yields a time series of monthly returns on each decile
portfolio from 1963 to 1993
– Funds that disappear during the course of the year are
included in the equal-weighted average until they
disappear, then the portfolio weights are readjusted
appropriately

54
• Findings
– Short-term persistence of fund performance
• The post-formation monthly excess returns on the
decile portfolios decrease nearly monotonically in
portfolio rank
• P1 (winners) – P10 (losers) = 8% (= 0.67% × 12)
– The CAPM does not explain the relative returns on these
portfolios (betas on P1 and P10 are similar)
– 4-factor model explains most of the spread and pattern in
these portfolio
• P1 has higher loadings on SMB and PR1YR
• PR1YR explains 31 basis points out of the 67-basis-point
monthly spread of P1 – P10

55
56
• Why can PR1YR explain short-term performance
persistence?
1. Fund managers successfully follow momentum strategies
• The abnormal performance of winner funds should be
long-lasting
2. Some mutual funds just happen by chance to hold
relatively larger positions in last year’s winning stocks
• The abnormal performance should reverse in the long
run

57
• Performance persistence of individual funds
– Construct a contingency table of initial and subsequent
one-year mutual fund rankings

– Only 20% of funds in P1 or P10 decile stay in the same

group in the next year
– Last year’s winners frequently become next year’s losers
and vice versa
• Consistent with gambling behavior by mutual funds
– Year-to-year rankings on most funds appear largely
random

58
59
• Post-formation returns on portfolios of mutual funds
sorted on lagged one-year return
– One-year performance persistence is mostly eliminated
after one year
– Except for the persistent underperformance by the worst
funds, mean returns across deciles do not differ
statistically significantly after one year

60
Formation year: the year subsequent to the initial ranking
(sample year 1962-1987)

61
• Loadings on PR1YR
–1 year Formation Year +1 year
P1 0.18 0.29 0.14
P10 0.00 –0.09 0.04

– Returns on the top and bottom decile funds are not nearly
so strongly related to the one-year momentum effect in
stock returns outside of the ranking and formation years
– Winner funds happen by chance to hold relatively larger
positions in last year’s winning stocks

62
• Fund characteristics and fund abnormal returns
– Fama-MacBeth regression

– α = one-month abnormal returns from the 4-factor model

– Factor loadings (b, s, h, and p) are estimated over the prior

three years
– x: Expense ratio, turnover (Mturn), total net assets
(ln(TNA)), or load fees
– When x are TNA, maximum load, and the turnover, returns
are measured after adding back expense ratios

63
– Mutual funds do not recoup their investment costs through
higher returns
• The coefficients on expense ratio, turnover, and load fees
are negative and significant
– Every 100-basis-point increase in expense ratios, annual
abnormal return (net of expenses) drops by about 154 basis
points
• Ideally, this coefficient should be positive, which means the
investment in expenses could be fully recovered by
additional returns

64
• 買0050還是006208?
年度 報酬率 總內扣費用
台灣50 0050 006208 0050 006208
2023 29.34 28.79 29.01 0.43 0.25
2022 –21.49 –21.79 –21.69 0.43 0.24
2021 22.42 21.72 22.10 0.46 0.35
2020 32.26 31.63 31.87 0.43 0.36
2019 33.71 32.97 33.06 0.43 0.34
2018 –4.41 –4.89 –4.84 0.44 0.32
2017 18.83 18.39 18.61 0.42 0.41
2016 19.12 18.67 18.66 0.42 0.43
2015 –5.69 –6.08 –6.22 0.42 0.48
2014 17.61 16.96 16.77 0.43 0.48
1. 指數報酬率以每年最後一交易日的價格計算
2. 總內扣費用(總開銷)包含經理費、保管費、交易成本，與其他費用
3. 2017年開始006208經理費由0.30%降至0.15%，0050經理費為0.32% (2024)
65
 Daniel and Moskowitz (2016)
• When and why does momentum crash?
– Momentum strategies can experience infrequent and
persistent strings of negative returns
– Momentum crashes occur in panic states, following market
declines and when market volatility is high, and are
contemporaneous with market rebounds
– Large changes in market beta can help to explain some of
the large negative returns earned by momentum strategies
– When the market has fallen significantly over the
momentum formation period, winners tend to be low-beta
firms and losers tend to be high-beta firms
– When market bebounds quickly, momentum strategies
crash because they have negative betas
66
What are the implications for portfolio management?

67
68