Are individual investors such poor portfolio

managers?
Camille Magron

January 4, 2014
Abstract
In that paper we evaluate individual investors performance with measures which
t their preferences and risk perception. Based on 24,766 individual investors from
a French brokerage between 2003 and 2006, we evidence that choosing alternative
performance measures to the Sharpe ratio result in dierent rankings of investors.
When they are evaluated with a measure consistent with their attitude towards
risk rather than with the Sharpe ratio, a larger proportion of investors beat the
market index. Yet, individual investors underperform a random investing strategy
even with alternative measures. We conclude that the improvement of investors
performance with alternative measures is driven by mechanical eects du to the
skewness of their portfolio rather than good stock picking skills.
JEL Classication: G 11
Keywords: Individual Investor, Performance Evaluation, Performance measures, Be-
havioral Portfolio Theory

LaRGE Research Center, EM Strasbourg Business School, University of Strasbourg. Ple Europen
de Gestion et dEconomie, 61 avenue de la Fort Noire 67085 Strasbourg Cedex. [email protected].
The author is grateful to Matti Keloharju, Kim Oosterlinck, Maxime Merli, Patrick Roger, Marie-Hlne
Broihanne, Pascal Grandin, Georges Gallais-Hamonno and Philippe Cogneau for comments. The author
thanks AFFI PhD Workshop participants and seminar participants at Centre Emile Bernheim (Universit
libre de Bruxelles) for suggestions. The author thanks Tristan Roger for valuable computer programming.
1
Individual traders are often regarded as at best uninformed, at worst fools. Coval,
Hirshleifer, and Shumway (2005).
Financial performance is one main concern of investors, whether they are professionals
or individual investors. The success of an investment strategy and the skills of a trader
are evaluated ex-post by assigning a score to the portfolio, which usually corresponds to
risk-adjusted returns. Concerning individual investors, the global evidence reports that
they do not outperform relevant benchmarks. Barber and Odean (2000) show on 66,465
U.S households that neither the Jensens alpha (Jensen (1968)) nor the intercept test from
the FamaFrench model (Fama and French (1993)) are reliably dierent from zero from
1991 to 1996. Barber and Odean (2000) also nd that the Sharpe ratio (Sharpe (1966)) of
the average household in their sample is 0.179, compared to 0.366 for the market during
the period. Based on the Jensens alpha and the intercept test from the FamaFrench
model, Odean (1999) provides evidence that the stocks that investors buy subsequently
underperform the stocks they sell. On the Taiwanese market, long-short portfolios that
mimic the buysell trades of individual investors earn reliably negative monthly alphas of
11.0%, 3.3%, and 1.9% over horizons of 1, 10, and 25 days respectively (Barber, Lee, Liu,
and Odean (2009)).
Researchers demonstrate that these poor results can be explained by psychological
considerations such as (among others) overcondence, familiarity bias, or loss aversion.
Even if these considerations are unrelated to the information about underlying security
values, they impact the trading choices of individual investors. As a result, individual
investors trade excessively, under-diversify their portfolio and have a tendancy to sell win-
ners too early and to ride losers too long (the so-called disposition eect).
In this paper, we argue that the poor performance of individual investors may be
simply due to a wrong performance measure. Indeed, risk-adjusted return indicators such
as the Jensens alpha, the Fama-French intercept and the four-factor intercept stem from
the Mean-Variance model (Markowitz (1952)). In this paradigm, we evaluate the risk
2
of a choice by the variance of the outcomes. However, surveys reveal that the variance
does not t with the risk perception of individual investors (Unser (2000); Veld and Veld-
Merkoulova (2008)). Therefore, though these performance indexes are widely spread in
the literature, we should interpret them cautiously when they are related to individual
investors. The same argument applies to the Sharpe ratio which is the most popular
performance measure in the nance industry. In The (more than) 100 ways to measure
portfolio performance Cogneau and Hubner (2009a) and Cogneau and Hubner (2009b)
suggest that a number of alternative performance measures overcome the main drawbacks
of the Sharpe ratio. Besides the abovementioned problem on risk perception, the fact that
the Sharpe ratio is founded on the Expected Utility Theory (EUT) questions its use as a
relevant performance measure for individuals investors. According to the EUT, investors
exhibit a uniform attitude towards risk, i.e., they are risk averse throughout (Von Neu-
mann and Morgenstern (1947)). Yet, experimental evidences nd that investors do not
behave as it is assumed in this model of decision-making. Research show that individ-
ual investors exhibit loss aversion, have risk averse preferences for gains combined with
risk seeking preferences for losses
1
(Kahneman and Tversky (1979); Tversky and Kahne-
man (1992)) and target lottery-like outcomes (Friedman and Savage (1948); Mitton and
Vorkink (2007)). To adress for these actual behaviors, the Prospect Theory (Kahneman
and Tversky (1979); Tversky and Kahneman (1992)) and the Behavioral Portfolio Theory
(BPT) (Shefrin and Statman (2000)) have been proposed in the litterature as alternative
models of preferences.
In this work, we demonstrate that individual investors are not such poor managers
when we evaluate their performance with measures correctly weighting their preferences
and risk perception. More precisely we choose performance measure which adjust gains
by the risk associated with losses (downside risk) instead of the total risk. Furthermore we
allow for dierent models of choices with the performance measure proposed by Farinelli
1
This is the certainty eect (risk averse preference for a sure gain over a larger gain that is merely
probable) combined with the reection eect (risk seeking preference for a loss that is merely probable
over a smaller losses that is certain
3
and Tibiletti (2008). This ratio exhibit a great exibility in such a way that attitudes
toward risk can be weighted for gains and losses.
Our empirical study shows that the proportion of individual investors beating the
market is much larger when performance is evaluated with a measure consistent with the
Behavioral Portfolio Theory, compared to the score they obtain with the Sharpe ratio.
As a consequence, we actively support a replacement of the Sharpe ratio by more tted
performance measures when it comes to evaluate the performance of individual investors.
The equivalence of performance measure is a topic of debate in the literature. On one
side, Eling and Schuhmacher (2007) compare 13 performance measures
2
and nd that the
ranking of 2763 hedge funds (over the period 1985-2004) is not signicantly aected by the
choice of the performance measure. In fact, the average rank correlation of performance
mesure with the Sharpe ratio is 97%. Eling (2008) conforms these results on a set of
38,954 mutual funds invested in stocks, bonds, commodities and real estate. In contrast
with these ndings, Zakamouline (2011) nds that the evaluation of performance depends
on the selected performance measure.
We contribute to the litterature on alternative performance measures by focusing on
individual investors instead of hedge funds. Our contribution is threefold. First we sup-
port the results of Zakamouline (2011) and nd that the choice of alternative performance
measures has an impact on the ranking of investors. For instance in 2003, the propor-
tion of investors who are upgraded (downgraded) with another measure than the Sharpe
ratio ranges from 35.94% to 46.45% (5.85% to 36.19%). We show that these proportions
signicantly dier from what is expected with random permutations. Second we show
that, compared to the market index, individual investors are not such poor managers as
reported by the Sharpe ratio ranking. For example in 2006, though only 10% of investors
outperform the market index according to the Sharpe ratio, 60% of the population beat
2
Sharpe ratio, Treynor ratio, Jensens alpha, Omega ratio, Sortino ratio, Kappa3 ratio, Upside Poten-
tial ratio, Calmar ratio, Sterling ratio, Burke ratio, Excess return on value at risk, Conditional Sharpe
ratio and Modied Sharpe ratio.
4
the market with the measure tting the Behavioral Portfolio Theory. With this measure,
30% of investors outperform the market during 4 consecutive years, whereas no investor
beat the market persistently with the Sharpe ratio. Finally we show that randomly cre-
ated portfolio outperform investors in our sample, even with the alternative measures.
We conclude that the improvement of investorsperformance is not driven by their stock
picking skills but rather by mechanical eects linked to the skewness of their portfolio
as a whole. As a result, though our main nding is the importance of the choice of the
measure, we do not conclude that individual investors exhibit particular skills to select
outperforming stocks.
This paper is organized as follows. In Section 1, we challenge the Sharpe ratio and
present the alternative measures analyzed in our study. We present the empirical study
in Section 2 and we conclude in Section 3.
1 The Sharpe ratio and the alternative measures
The Sharpe ratio is usually computed as follows:
Sharpe ratio =
(r
i
r
f
)

i
(1)
in which r
i
is the mean return of the investor i, r
f
is the risk free rate and
i
is the standard
deviation of the portfolio returns. This measure has a simple design and includes two main
information (risk and return). Yet, the Sharpe ratio is not relevant in case of individual
investors.
Firstly, it is impossible to establish a global ranking of investors with the Sharpe ra-
tio. Indeed, when the numerator is positive, the larger the excess return and the lower
the standard deviation, the larger the Sharpe ratio. However in case of a negative nu-
merator, investors cannot be ranked in order of their performance. To illustrate this
limit, consider this small example, with 3 assets, A (r
A
r
f
= 0.10;
A
= 0.40), B
5
(r
B
r
f
= 0.10;
B
= 0.50), and C (r
C
r
f
= 0.05;
C
= 0.40).
Sharpe ratio
A
=
(r
A
r
f
)

A
= 0.25
Sharpe ratio
B
=
(r
B
r
f
)

B
= 0.20
Sharpe ratio
C
=
(r
C
r
f
)

C
= 0.125
A and B exhibit the same excess return, but B is more volatile. Therefore, B is prefered
to A. In term of Sharpe ratio, that means that Sharpe ratio
B
= 0.20 is worse than
Sharpe ratio
A
= 0.25. In that case, the smaller the Sharpe ratio, the better the per-
formance of the asset. Yet, lets compare A and C which exhibit the same volatility, but
r
C
= 0.05 is larger than r
A
= 0.10. Therefore C is prefered to A, which means that
Sharpe ratio
C
= 0.125 is better than Sharpe ratio
A
= 0.25. In that case, the rule is
reversed: the larger the Sharpe ratio, the better the performance of the asset.
Secondly, the variance used as a risk measure in the Sharpe ratio constitutes a major
drawback because it allocates the same weight to positive and negative deviations. Ac-
tually, surveys on risk perceptions reveal that symmetric risk measures are neglected by
investors in favor of asymmetric ones (Unser (2000); Veld and Veld-Merkoulova (2008)).
At the same time, Ang and Chen (2006) show that returns integrate a premium for
the risk of losses. The aggravation that one experience losing a sum of money appears
to be greater than the pleasure associated with gaining the same amount; this eect is
so-called loss aversion (Kahneman and Tversky (1979), Kahneman, Knetsch, and Thaler
(1990), Tversky and Kahneman (1992)).
Numerous alternative measures to the Sharpe ratio have been suggested in the lit-
terature. In that paper, we study alternative measures that have a return/risk design.
This simplicity is one main strenght of the Sharpe ratio. To adress the abovementioned
weaknesses, we select alternative measures that are based on upper and lower partial
6
moments.
Lower partial moments (LPM) measure the risk as negative deviations from a refer-
ence point. Therefore, they evaluate the undesirable volatility (i.e., left-side volatility),
or the so-called downside risk. The lower partial moment of order k (k > 0) for investor
i during the period T is dened as:
LPM
k
(r
i
) =
1
T
T

t=1
[Max(0, r
it
)]
k
(2)
in which is the target return and r
it
is the stock return on date t. The coecient k
weights the deviation from the target return. For example, LPM of order one measures
the expected value of loss and LPM of order two measures the semi-variance. Note that
the semi variance is a measure of the asymmetry of the distribution. In case of symmetric
returns, the semi-variance is equal to half of the variance.
In LPM of order k, if k > 1, the greater the k, the higher the emphasis on the ex-
treme deviations from the benchmark. By contrast, a k < 1 means that the agents main
concern is to fail the target without regard to the amount lost. If moderate deviations
from the target are relatively harmless when compared to large deviation, then a high
order for the lower moment is adapted (k > 1). Figure 1 illustrates this mechanism.
For each graph, returns r
it
presented on the x-axis lie between -2% and 2%. Outcomes
[Max(0, r
it
)]
k
, based on a target return equal to 0.5%, are on the y-axis. We present
2 cases for k: k = 0.5 and k = 1.5. For k < 1 additional percent of return lost provide
diminishing marginal contribution to the shrinkage of the outcome. By contrast, when
k > 1, additional percent of return lost provide increasing marginal contribution to the
shrinkage of the outcome.
Symmetrically to Lower Partial Moments, Upper Partial Moments (UPM) measure
the positive deviation of returns from the target return:
UPM
k
(r
i
) =
1
T
T

t=1
[Max(0, r
it
)]
k
(3)
7
As for lower partial moments, the coecient k (k > 0) enables the user to allocate a
weight to deviations (see gure 2). The higher the order, the higher the agents inclination
towards the extreme outcomes (with outcomes equal to [Max(0, r
it
)]
k
). If small gains
are satisfactory, then an order lower to 1 ts the purpose (k < 1).
To synthesize, partial moments are always positive, allowing a global ranking of in-
vestors. Risk is dened by downside risk and loss aversion can be taken into account with
a higher coecient for lower than upper moments.
The Farinelli-Tibiletti ratio is a performance measure based on Upper Partial Moments
at the numerator and Lower Partial Moments at the denominator.
Farinelli Tibiletti
(pq)
ratio(r
i
) =
(UPM
p
(r
i
))
1/p
(LPM
q
(r
i
))
1/q
(4)
The exibility in the coecients p and q allows designing a performance measure adapted
to investors preferences.
In a rst case, we choose p and q equal to 1 to convey neutral attitude towards risk for
gains and losses. Indeed, p = q = 1 implies that positive and negative deviations from the
target are equally weighted. The Farinelli-Tibiletti(1,1) corresponds to the Omega ratio,
previously proposed by Keating and Shadwick (2002).
Omega ratio(r
i
) =
(UPM
1
(r
i
))
(LPM
1
(r
i
))
(5)
In a second case, we integrate loss aversion in the Farinelli-Tibiletti ratio with a higher
order for the LPM: q = 2. The Farinelli-Tibiletti(1,2) corresponds to the Upside Potential
ratio, previously proposed by Sortino, Van Ver Meer, and Plantinga (1999).
Upside potential ratio(r
i
) =
(UPM
1
(r
i
))
(LPM
2
(r
i
))
1/2
(6)
The Omega ratio and the Upside Potential ratio are the rst measures selected in our
8
study to be compared to the Sharpe ratio (Selected performance measures are summa-
rized in table 1).
The Sharpe ratio microeconomic foundations are based on the Expected Utility The-
ory (EUT) (Von Neumann and Morgenstern (1947)). A performance measure that is
suited to such behaviors reects risk aversion in the domain of gains and in the domain
of losses. To take into account this model, we choose p < 1 and q > 1. Indeed, in the
domain of gains (UPM at the numerator), p < 1 indicates that investors are not neces-
sarily seeking large gains with low probability of occurrence. Instead, they are satised
as soon as outcomes pass the target and additional gains provide a decreasing marginal
contribution to utility. In the domain of losses (LPM at the denominator), q > 1 signies
that large deviations from the target return are not desired. The marginal contribution
of additional losses to (des)utility is increasing.
Yet, Kahneman and Tversky (1979) evidence the inability of the EUT to explain
portfolio choices of investors. More precisely, in EUT investors exhibit uniform attitude
towards risk. Contrary to this assumption, experimental evidence has established a four-
fold pattern of risk attitudes: risk aversion for most gains and low probability losses, and
risk seeking for most losses and low probability gains (Kahneman and Tversky (1979)).
To consider the attitude towards risk for most commons events, Kahneman and
Tversky (1979) introduced a S-shaped utility function (the so-called Value function) in
Prospect Theory. With this Value function, the marginal value of both gains and losses
decreases with their magnitude. The concept of Loss aversion is integrated by a steeper
Value function for losses than for gains. We take into account this model of preferences in
the Farinelli-Tibiletti ratio with p < 1 at the numerator, meaning that the investor is risk
averse in the domain of gains, and q < 1 at the denominator, meaning that the investor
is risk seeking in the domain of losses.
9
Concerning unlikeky outcomes, Kahneman and Tversky (1979) report that individuals
are risk averse for losses and risk seeking for gains. This behavior is in line with Friedman
and Savage (1948) puzzle, the fact that investors who buy insurance policies often buy
lottery tickets at the same time. Friedman and Savage (1948) explain that (p.279) "An
individual who buys re insurance on a house he owns is accepting the certain loss of a
smal l sum (the insurance premium) in preference to the combination of a smal l chance
of a much larger loss (the value of the house) and a large chance of no loss.That is, he
is choosing certainty in preference to uncertainty. An individual who buys a lottery ticket
is subjecting himself to a large chance of losing a smal l amount (the price of the lottery
ticket) plus a small chance of winning a large amount (a prize) in preference to avoiding
both risks. He is choosing uncertainty in preference to certainty".
More recently, researchers observe that individual investors design their portfolio with
the intention of increasing the likelihood of extreme positive returns. In other words, in-
vestors make the distributions of their wealth more lottery-like (Statman (2002); Kumar
(2009); Barberis and Huang (2008)). Kumar (2009) dene lottery-type stocks following
the salient features of state lotteries. Lottery tickets have very low prices relative to the
highest potential payo (i.e., the size of the jackpot); they have low negative expected
returns; their payos are very risky (i.e., the prize distribution has extremely high vari-
ance); and, most importantly, they have an extremely small probability of a huge reward
(i.e., they have positively skewed payos). As with lotteries, if investors are searching for
lottery-like assets, they are likely to seek low-priced, high stock-specic skewness stocks.
Along the same lines, Mitton and Vorkink (2007) nd that investors consciously hold
few stocks to capture positive skewness. Indeed, though increasing the number of assets
in a portfolio enables to decrease the total variance by cancelling specic risk of each
security, it also reduces portfolio skewness. Therefore, a strategic underdiversication is
necessary to capture asymmetric returns. Moreover, underdiversied investors are more
prone to select positively skewed stocks (Mitton and Vorkink (2007)). Goetzmann and
Kumar (2008) prove that investors who tilt their portfolio towards stocks with an asym-
metric distribution and a high variance (small capitalizations, growth stocks, technological
10
sector) hold concentrated portfolios.
In Prospect Theory, these preferences can be adressed by the combination of the Value
function and an overweighting of the probabilities of extreme events. The probability
transformation oset the initial shape of the Value function. As a result, additional
percent of return lost provide decreasing marginal (des)utility, whereas additional percent
of return gained provide increasing marginal utility. Shefrin and Statman (2000) have
emphasized the role of gambling in investment decisions in their Behavioral Portfolio
Theory (BPT). According to BPT, investors proceed in two steps to set their portfolio.
First they satisfy a security criteria, ensuring a minimum return with riskless assets.
Second they invest their remaining wealth in a cheap asset with huge probability of gains
3
.
We can translate such preferences in the Farinelli-Tibiletti ratio with p > 1 at the
numerator and q > 1 at the denominator. With p > 1, investors target exceptional per-
formances and gives importance to large but unlikely excess returns above the benchmark.
q > 1 means that large losses are undesired.
In that paper, we follow Zakamouline (2011) and estimate the Farinelli-Tibiletti ratio
according to 3 (p q) couples: (0.5 2) to be in accordance with the EUT; (0.8 0.85)
to model the Value function, and (1.5 2) to depict BPT investors.
To sum up, alternative performance measures are always positive and consider down-
side risk thanks to lower partial moments at the denominator. The Omega ratio represents
investors with neutral attitude towards risk. The Upside Potential ratio integrates loss
aversion with a stronger coecient for LPM. The Farinelli-Tibiletti ratio(0.5-2) repre-
sents investors who behave as it is assumed in the EUT, the Farinelli-Tibiletti ratio(0.8-
0.85) represents investors whose preferences are conveyed by the Value function and the
Farinelli-Tibiletti ratio(1.5-2) represents investors who behave accordingly with the BPT.
In this rst section, we explained why the Sharpe ratio is not a suitable measure to
estimate the performance of individual investors. Based on ve alternative performance
3
Two versions of BPT co-exist: One mental account and multiple mental accounts. We set in the one
mental account case.
11
measures that overcome the Sharpe ratio main drawbacks, we now evidence that the
evaluation of individual performance is inuenced by the choice of the measure. In a
second step, we show that investors are not such poor managers when their performances
are estimated with alternative measures.
2 Empirical study
2.1 Data
The primary data set is provided by a large European brokerage house. We obtained
daily transactions records of 24,766 French investors over the period 2003-2006. Descrip-
tive statistics related to these investors are presented in table 2. Data in the panel A
indicate that among the 24,766 investors, 80.4% are men. The panel B and C are dedi-
cated to transactions and portfolios respectively, and present yearly results. The 24,766
investors realized 1,882,044 trades over the 4 years. On average, each investor realized
19.2 transactions in 2003, 16.7 transactions in 2004, 18.7 transactions in 2005 and 21.4
in 2006. The investors average purchase (sale) turnover lies between 5.9% (6.2%) in 2004
and 7.8% (8.7%) in 2006. Note that the purchase turnover and the sales turnover are the
values of the amounts bought or sold, respectively, in proportion of the monthly portfolio
value.
From the trade records, we computed the daily stock portfolio of each investor. On
average, the portfolio value of investors worthes 24,241AC in 2003, 27,901AC in 2004, 31,259AC
in 2005 and 36,629AC in 2006. Investors hold an average of 8 stocks in their portfolios.
Yearly returns of portfolios are computed based on weekly returns. The lowest annual
return is observed in 2004 (8.16%) whereas the largest is observed in 2003 (31.40%).
Over the same period, the market index
4
exhibit annual returns of 22.99% (2003), 14.95%
(2004), 28.99% (2005) and 25.23% (2006). The dierence between the average return of
4
Data on the market index is given by the Eurodai general index (computed using the methodology
of the Center for Research in Security Prices, CRSP, and based on approximately 700 stocks over the
period under consideration). European nancial data institute: https://www.eurodai.org
12
investors and the market return in 2003 is explained by an important skewness of investor
(0.76). Computed Jensen s are consistent with these dierences. Indeed, in 2003, is
positive, equal to 6.99% whereas in 2004 it is negative, equal to 3.83%.
It is worth mentionning that in 2003 (resp. 2004, 2005, 2006), 82.8% (resp.18%, 37.7%,
44.5%) of investors hold portfolio with non Gaussian distribution of returns. We test nor-
mality with the Jarque-Bera test at 95% condence level.
Stock price data come from two sources, Eurodai for stocks traded on Euronext and
Bloomberg for the other stocks. Investors trade 2,491 stocks, of which 1,191 are French,
and the main part of the others comes essentially from the U.S. (1020 stocks). Despite
this large part of U.S stocks, more than 90% of trades are realized on French stocks.
In the following section, we demonstrate that the alternative measures chosen in this
study are not equivalent to the Sharpe ratio. This observation is the starting point
required to justify that alternative measures should be favored over the Sharpe ratio
when evaluating individual investors.
2.2 Equivalence between measures
Two measures are said equivalent if they generate the same ranking of the set of investors.
Performance measures are calculated each year using weekly returns. The target return
on performance measures requiring such target return is the risk-free rate. Each year,
we sort investors in decile with each measure, including the Sharpe ratio. The ranking
computed for the Sharpe ratio is sightly dierent from a classic ranking. Though we
cannot rank together several investors who exhibit a negative Sharpe ratio, we can rank a
positive and a negative Sharpe ratio. Indeed, the positive Sharpe ratio is better than the
negative one. We thus rank investors who exhibit a negative Sharpe ratio in the bottom
decile. It follows that the bottom decile may contain more than 10% of investors. More
precisely, in 2003 (resp. 2004 2005 and 2006), the bottom decile contains 4.22% (resp.
25.85%, 7.52% and 8.57%) of investors. We then rank investors who exhibit a positive
13
Sharpe ratio in nine quantiles. These nine quantiles each contains 11.11% of remaining
investors.
We present the rank correlations in table 3. We compute two statistics (Spearman
and Kendall ) and both result in similar conclusions, although the Kendall pro-
vides lower statistics. Since both statistics are supported by researchers in the litterature
(Noether (1981), Griths (1980)), there is no reason to prefer one to the other.
The Omega ratio is the measure that exhibit the higher correlation with the Sharpe
ratio. With the Spearman , the correlation is close to 98% each year. Therefore the
only consideration of the downside risk does not modify the evaluation of investors. The
Upside Potential ratio is the second most correlated measure with the Sharpe ratio. The
inclusion of the consideration of loss aversion does not much more inuence the evaluation
of investors. We observe that the correlation is stronger in 2004, rising to 94.88% with
the Spearman statistic, when the proportion of investors who exhibit normal returns
is the highest (82%). Since the alternative measures are based on deviations from the
benchmark, the ranking by these measures should be closer to the ranking produced by
the Sharpe ratio when returns are normal. Yet, we do not observe a similar increase of
correlation with the other measures, indicating that others eect are interacting.
In order of decreasing correlation, we next nd the Farinelli-Tibiletti(0.5-2) ratio, the
Farinelli-Tibiletti(0.8-0.85) ratio and the Farinelli-Tibiletti(1.5-2) ratio. For the latter, Spear-
man ranges between 30.59% and 41.58%, and Kendall ranges between 23.17% and
32.60% across years. This model seems to be the farthest to the Sharpe ratio.
These strengths of relationship between alternative measures and the Sharpe ratio are
corroborated by the transition matrices presented in tables 4, 5, 6, 7 and 8. In rows
we present the ranking resulting from the evaluation of investors performance with the
Sharpe ratio. In columns, we present the ranking resulting from the evaluation with the
considered alternative measure. For each pair of deciles (i, j), we report the conditional
probability to be ranked in the decile j with the alternative measure when the rank with
14
the Sharpe ratio is the decile i.
5
If the rankings are similar, i.e the decile of investor with the Sharpe ratio (i) and the
decile of investor with the alternative measure (j) are equal, we should observe positive
probabilities only on the diagonal of the matrix.
We see clearly that the rankings resulting from the evaluation of performance with the
Omega ratio and the Sharpe ratio are close. Indeed, most probabilities in the matrices
are null except the values on the main diagonal where i = j and on the second diagonals
where j = i + 1 and j = i 1. For instance in 2003, if an investor is ranked in the rst
decile with the Sharpe ratio, then the probability to be ranked in the rst decile with the
Omega ratio is 100%. Therefore the probability to be in j = 2, ..., 10 is null when i = 1. If
the investor is ranked in decile i with the Sharpe ratio, then there are large probabilities
for him to be ranked in decile j = i or in decile j = i 1 with the Omega ratio. Similar
patterns are observed in 2005 and 2006. Yet in 2004, the diagonal is shift to the right of
the matrix. Hence, an investor ranked in decile i (i = 3 to 8) with the Sharpe ratio has
more probability to be ranked in decile j = i + 1 than in decile j = i with the Omega
ratio.
With the Upside potential ratio, the conditional probabilities are more spread out over
the matrices, but they are null for the extreme pairs of deciles. For instance in 2004 with
i = 10, we observe null probabilities for j = 1 to j = 4, i.e., an investor ranked in the top
decile with the Sharpe ratio cannot be ranked in the rst worst deciles with the Upside
potential ratio.
We obtain even more positive conditional probabilities with the Farinelli-Tibiletti(0.5-2)
and the Farinelli-Tibiletti(0.8-0.85).
With the Farinelli-Tibiletti(1.5-2) ratio, the conditional probabilites are spread out over
the whole matrix. In other words, it is possible to be ranked in each decile j = 1, ..., 10
resulting from the alternative measure evaluation whatever the decile i resulting from the
Sharpe ratio evaluation . We observe similar patterns across years.
5
Based on contingency tables (i.e., the observed frequencies for each pair of deciles (i, j)), Khi2 test
conrms that the ranking of investors according to the Sharpe ratio is not independant from the ranking of
investors according to alternative measures: P(i, j) = P(Decile
Sharpe
= i) P(Decile
MesureAlternative
=
j).
15
These transition matrices indicate that the performance measure does inuence the
evaluation of investors, which corroborates Zakamouline (2011) works. We summarize
the rank permutations in table 9. With the Farinelli-Tibiletti(1.5-2) ratio, the maximum
downgrade (presented in column 1) is -9. In other words, some investors ranked in the best
decile with the Sharpe measure move to the bottom decile with this alternative measure.
We observe the same phenomenon in the opposite direction. For instance in 2003, with
the 3 versions of the Farinelli-Tibiletti ratio, the maximum upgrade (presented in column
2) is 9. With the Omega ratio the maximum upgrade is 2 in 2003 and 2005, and only 1
in 2004 and 2006. Note that the maximum upgrade is always observed with the Farinelli-
Tibiletti(1.5-2) ratio, which is consistent with the fullness of the transitions matrices. We
observe that the results are similar for the Upside Potential ratio that integrates the value
of loss aversion and the Farinelli-Tibiletti(0.5-2) ratio that accounts for the Expected Utility
Theory.
We present the proportion of investors who remains in the same decile in the third
column. In 2006 for instance, this proportion ranges from 17.90% with the Farinelli-
Tibiletti(1.5-2) ratio to 82.64% with the Omega ratio. Over the years, the highest pro-
portion of investors who remain in the same decile is always observed with the Omega
ratio, followed by the Upside potential ratio. By contrast, the largest proportions of in-
vestors who move to a higher/lower decile (see the fourth and fth column of the table)
are observed with the Farinelli-Tibiletti(1.5-2) ratio. With this measure in 2004, 47.33% of
investors are downgraded whereas 36.27% of investors are upgraded.
Hence, a considerable proportion of investors moves to a dierent decile with certain
alternative measures. Computed proportions are similar across years, which supports the
robustness of this observation.
To test whether these rank permutations are signicant, we run Monte-Carlo simula-
tions. More precisely, we create a vector of 24,766 ctitious investors, ranked in deciles.
16
We rank investors #1 to #2477 in the rst decile, investors #2478 to #4955 in the second
decile, and so on until investors #22, 290 to #24, 766 who are ranked in the 10
th
decile.
Based on this initial vector, we compute 1,000 random rank permutations to obtain 1,000
new vectors with permuted deciles. Across the 1,000 permutations, the proportion of
ctious investors who remain in the same decile ranges between 9.63% and 10.38% at
the 95% condence level. Therefore, if our results were driven by chance, we should ob-
serve that the proportion of investors who remain in the same decile when we evaluate
them with an alternative measure rather than with the Sharpe ratio is comprised in this
condence interval. Yet in 2003 for instance, the actual proportions ranges from 17.35%
to 58.20%. Therefore the permutations that we observe are not the result of a random
process.
2.3 Comparison with the market index
Permutations resulting from evaluation with alternative measures apply to the market
index too. In table 10 we present for each year and each measure the decile of the market
index. These grades are based on the initial ranking of investors computed for each
measure. To understand how the choice of the performance measure does inuence the
evaluation of investors, it is interesting to analyze this table in term of percentages. For
example, in 2005, only 10% of investors outperform the market index according to the
Sharpe ratio. Yet, if we refer to the Farinelli-Tibiletti(0.8-0.85) ratio, for the same year,
30% of investors beat the market, while this proportion rises to 60% with the Farinelli-
Tibiletti(1.5-2) ratio. Therefore, evaluating investors with a measure that ts to the S-
shaped Value function leads to worse results (for investors) than with a mesure that ts
the Behavioral Portfolio Theory.
More substantial, in 2004, though only 10% of investors outperform the market index
according to the Sharpe ratio, with the Farinelli-Tibiletti(1.5-2) ratio 90% of the population
is ranked in a better decile. This large dierence is consistent with the results reported
in table 9 and lead us to wonder whether individual investors are such poor managers as
17
studies usually report. Note that the measure that ts the Value function in Prospect
Theory is the second most favorable measure for individual investors.
In 2003, we observe a smaller dierence between the ranks dened according to each
measure. This eect is consistent with the strong outperformance of annual returns of
investors on the market this particular year (see table 2).
To test whether our results are driven by the value of p and q chosen in that paper, we
compute the proportion of investors who beat the maket each year, according to the value
of (p q) couple. Results are presented in gure 3, 4, 5 and 6. p and q coecients lie
between 0 and 4, with a 0.1 incremental unit. It appears that the proportion of investor
who beat the market increases with the coecient p and q.
Depending on the value of (p q), the proportion of investors who beat the market
ranges between 10 % and 90% in 2003, and between 0% and 90% in 2004. In 2005 and
2006, the maximum proportion of investors who beat the market is 70%.
To end with the largest proportion of investors beating the market (darkest area) when
q = 2, p must be at least equal to 1.5 in 2003, 1.3 in 2004, and 2 in 2005. These results
explain why the Farinelli-Tibiletti(1.5-2) ratio gives rise to a larger part of investors beating
the market than the Farinelli-Tibiletti(0.5-2). In 2006, if q = 2, p must be at least 3.9 to
be located in the darkest area. This value is more than twice the value of the highest p in
our computations (p = 1.5). This observation is consistent with the previous result that
only 50% of investors beat the market with the Farinelli-Tibiletti(1.5-2) in 2006. In fact,
with q = 2, the proportion of investors who beat the market increases with the value of
p. The surface is darker and darker as we move to the right side of the gure.
If we compare p = 0.5 and p = 0.8, the proportion of investors who beat the market is
most of time larger with p = 0.8, whatever the value of q. Therefore, the value of p (for
the UPM at the numerator) is more determinant than the value of q (for the LPM at
the denominator) to evaluate the outperformance of investors.
We next examine whether outperformance is persistant over time. In table 11, we
18
present the proportion of investors who are ranked in a higher decile than the market
index with each measure during 1, 2, 3 and 4 years starting from 2003. In other words,
we analyze the proportion of investors who beat the market in 2003, in 2003 and 2004, in
2003, 2004 and 2005 and in 2003, 2004, 2005 and 2006. With the Farinelli-Tibiletti(1.5-2)
ratio, 90% (resp. 81.37%, 48.16%, 30.48%) of investors beat the market during 1 year
(resp. 2 years, 3 years, 4 years). These proportions are the largest ones in the table,
followed by the values obtained with the Farinelli-Tibiletti(0.8-0.85) ratio and the Farinelli-
Tibiletti(0.5-2) ratio. As a comparison, with the Sharpe ratio 42.66% of investors beat the
market in 2003, 4.67% in 2003 and 2004, 1.12% from 2003 to 2005 and less than 0.3%
over the complete period.
Therefore, with the Sharpe ratio we conclude that a handful of them beat persistently
the market. By contrast with the measure that ts BPT, more than a quarter of investors
beat the market during 4 consecutive years.
2.4 Skills or luck?
In the previous section, we provided evidence that the Farinelli-Tibiletti(p-q) ratio that is
consistent with the Behavioral Portfolio Theory promotes the portfolio hold by individual
investors. In other words, with this measure, individual investors perform much better
than with the others. BPT investors tend to increase the likelihood of extreme positive
returns by making the distributions of their wealth more lotery-like,. Mitton and Vorkink
(2007) show that this skewness seeking drives investors to hold underdiversied portfolios.
Consistently with this nding, the median number of stocks hold in portfolio is 6 in our
sample (see descriptive statistics in table 2). According to the works of Statman (1987),
a well diversied portfolio must include at least 30 stocks. Hence, investors hold underdi-
versied portfolios, which conrm that individual investors in our sample try to capture
extreme asymmetric returns. Both results (outperformance of investors with BPT and
underdiversication) jointly indicate that the behavior of investors in our sample is best
modelized with the Behavioral Portfolio Theory.
19
In that section we analyze whether the observed outperformance of investors is solely
mechanical. Actually, it is not surprising to nd that investors who are underdiversied
outperform the market with a measure that promote asymetric returns. Though our re-
sults might be purely driven by mechanical eects, our goal in this paper is to emphasize
that there exists measures more suited to individual investor than the Sharpe ratio. We
indeed show that these measures lead to rened conclusions relatively to their poor trad-
ing ability. Yet, can we really conclude that individual investors select stocks correclty?
Do they really have particular stock picking skills? Are they doing better than they could
do by luck?
To answer these questions, we start with the creation of 24,766 portfolios composed
of stocks picked at random. The weights that we allocate to each stock in the portfolios
are drawn randomly as well. We then compute Sharpe ratios and alternative measures
each year, for each portfolio, based on weekly returns. Lastly, we rank each year and with
each measure the random portfolios.
The number of stocks in each portfolio mimic the number of stocks of investors. More
precisely, the rst portfolio created contains exactly the same number of stocks than the
portfolio of the investor #1; The second portfolio created contains exactly the same num-
ber of stocks than the portfolio of the investor #2; and so on.
Our goal in this section is to analyze the rank of the market index among these
random portfolios. As table 12 shows, with the 3 versions of the Farinelli-Tibiletti ratio,
the market index is each year in the bottom part of the ranking. In other words, 90%
of the random portfolios outperform the market index. Comparing this large proportion
with the results reported in table 10, we remark that investors do not perform better than
the randomly chosen portfolios.
Though the Farinelli-Tibiletti ratio enhances investors performance, an under-diversied
random strategy is even more promoted. Interestingly, the random strategy is promoted
20
by the measures that t the 3 models of decision making considered in our study. Yet,
though the Behavioral Portfolio model implies to underdiversify for capturing skewness,
it is the opposite for the Expected Utility Theory. Indeed, EUT penalizes the deviations
from the target return that arise due to a lack of diversication. We assume that the
good performance of stocks randomly selected overcome this eect.
Consequently,it is the shape of their distribution of returns which boosts investors
performance. The overall increase in performance is not due to the particular stocks
chosen.
With the Sharpe ratio, the Omega ratio, and the Upside potential ratio, the market
index is, as expected, in the top of the ranking. Yet, in 2003, though the index is
ranked in higher deciles with these measures than with the Farinelli-Tibiletti ratios, the
outperformance is not so clear. Indeed, at least 60% of the randomly created portfolios
are ranked in a better decile. We observed a similar eect with the portfolios of investors
(see table 10).
3 Conclusions
In this paper we evidence that it is not reasonable to evaluate investors performance
according to a standard ratio that does not consider their investing criteria. We compare
the evaluation of performance resulting from the Sharpe ratio with the ones resulting from
alternative performance measures. We consider ve measures, designed as the Sharpe
ratio (return to risk ratio). Yet those measures are always positive and enable a ranking
among investors in all cases, whereas the Sharpe ratio has no meaning when it is negative.
Besides this main dierence, risk is dened by negative deviation from a target, rather
than by the variance.
Alternative measures are built with partial moments and are designed to take into
consideration several preferences of investors. The Omega ratio represents neutral atti-
tude towards risk for gains and losses. The Upside Potential ratio integrates the concept
of loss aversion with a stronger weight allocated to losses than to gains. Investors within
the Expected Utility Theory, who are risk averse throughout, are considered with the
21
Farinelli-Tibiletti(0.5-2) ratio. Investors whose preferences are consistent with the Value
function in Prospect Theory (i.e., risk averse for gains and risk seeking for losses) are
represented in the Farinelli-Tibiletti(0.8-0.85) ratio. Lastly, investors who behave as it is
assumed in the Behavioral Portfolio Theory (i.e., risk seeking for gains and risk averse for
losses) are taken into consideration with the Farinelli-Tibiletti(1.5-2) ratio. Considering the
tendancy to seek skewness through underdiversication reported by Mitton and Vorkink
(2007), this model seems to best t the behavior of individual investors.
We rst show that the choice of the performance measure does inuence the ranking of
investors. Indeed, a signicant part of investors moves to a higher/lower decile when we
estimate their performance with an alternative measure. Second, we nd that a greater
proportion of investors outperform the market index with alternative measures, notably
with the Farinelli-Tibiletti(1.5-2) ratio. Furthermore, 30% of investors beat persistently
(over 4 consecutive years) the market with the Farinelli-Tibiletti(1.5-2), compared to 0.3%
with the Sharpe ratio.
Hence estimating performance with a measure that correctly weights skewness seeking
of investors provide evidence that their risk-adjusted returns is far better than it is usu-
ally reported with performance measures that stem from the Mean-Variance paradigm.
Yet, we nd that the improvement of portfolios performance with alternative measure is
mainly due to mechanical eects due to skewness rather than stock picking skills. Indeed,
even when they are evaluated with adequate alternative measures, individual investors
underperform a random investing strategy.
22
Figure 1 Lower Partial Moments
.
Figure 2 Upper Partial Moments
23
Table 1 Alternative performance measures
This table presents the alternatives performance measures con-
sidered in that paper. Attitude towards gain and losses im-
plied by the parameters values are detailed for each measure.
Alternative performance measures Attitude towards gains Attitude towards losses
Omega ratio(r
i
) =
(UPM1(ri))
(LPM1(r
i
))
Small gains and large gains
are weighted equally
Small losses and large
losses are weighted equally
Always positive and Downside risk
Upside potential ratio(r
i
) =
(UPM1(ri))

(LPM2(ri))
Small gains and large gains
are weighted equally
Large losses are undesired
Integration of loss aversion
Farinelli Tibiletti
(0.52)
ratio(r
i
) =
(UPM0.5(ri))
1/0.5

(LPM2(ri))
Small gains are favored
over large but low probable
gains
Large losses are undesired
Consideration of the Expected utility function
Farinelli Tibiletti
(0.80.85)
ratio(r
i
) =
(UPM0.8(ri))
1/0.85
(LPM0.8(ri))
1/0.85
Small gains are favored
over large but low probable
gains
Losses are undesired, no
matter their magnitude
Consideration of the S-Shaped Value function
Farinelli Tibiletti
(1.52)
ratio(r
i
) =
(UPM1.5(ri))
1/1.5

(LPM2(ri))
Large but low probable
gains are favored
Large losses are undesired
Consideration of the Behavioral Portfolio Theory
24
Table 2 Descriptive statistics
This table presents statistics on the dataset during the period 2003 to 2006. Panel
A is related to investors. Panel B provides yearly information on transactions, av-
eraged across investors. The monthly turnover is computed as the market value
of shares purchased in month t, or sold in month t, divided by the mean market
value of all shares held in the portfolio during month t. Panel C reports yearly
information on investors portfolios, averaged across investors. The portfolio value
and the number of stocks are calculated in the mi-year. Annual returns and skew-
ness are computed based on weekly returns. Medians are reported in parentheses.
2003 2004 2005 2006
Panel A : Investors
Number of investors 24,766
Proportion of men 80.4 %
Panel B : Transactions
Total number of trades 444,155 431,022 512,309 651,801
Average number of trades per investor 17.9 (2) 17.4 (2) 20.7 (4) 26.3 (5)
Purchase monthly turnover (%) 7.2 (1.3) 5.9 (1.1) 6.4 (1.2) 7.8 (1.7)
Sale monthly turnover (%) 7.2 (1.5) 6.2 (1.6) 7.1 (2.2) 8.7 (2.9)
Panel C: Portfolios
Portfolio value (Euros) 24,241 27,901 31,259 36,629
(9455) (10,935) (11,293) (13,252)
Number of dierent stocks in portfolio 8.6 (6.3) 8.4 (6.2) 8 (6) 7.8 (5.5)
Annual return (%) 31.40 (27.53) 8.16 (8.47) 28.03 (26.94) 22.05 (20.27)
Annual Jensen (%) 6.99 (5.71) -3.83 (-2.95) 2.54 (3.04) -2.38 (-2.44)
Annual Skewness 0.76 (0.74) -0.09 (-0.16) -0.06 (-0.16) -0.38 (-0.50)
25
Table 3 Rank correlations
This table presents the relationship between the rankings of investors resulting from the
evaluation of investors performance with alternative measures and the Sharpe ratio.
The Spearman and the Kendall are computed each year between 2003 and 2006.
2003 2004 2005 2006
Spearman correlations (%)
Omega ratio 97.92 98,.44 97.90 98.79
Upside Potential ratio 91.81 94.88 91.71 89.59
Farinelli-Tibiletti(0.5-2) ratio 70.74 71.11 85.11 79.72
Farinelli-Tibiletti(0.8-0.85) ratio 66.11 64.38 82.77 76.70
Farinelli-Tibiletti(1.5-2) ratio 33.72 30.59 41.58 42.26
Kendall correlations (%)
Omega ratio 93.22 94.74 93.52 96.06
Upside Potential ratio 81.71 86.35 81.18 77.71
Farinelli-Tibiletti(0.5-2) ratio 57.42 57.04 71.70 65.61
Farinelli-Tibiletti(0.8-0.85) ratio 52.79 51.26 69.74 62.78
Farinelli-Tibiletti(1.5-2) ratio 25.86 23.17 30.94 32.60
26
Table 4 Transition matrices - Sharpe ratio/Omega ratio
This table presents the transition matrices between the ranking result-
ing from the Sharpe ratio evaluation and the Omega ratio evaluation for
2003, 2004, 2005 and 2006. We report the conditional probability to be
ranked in decile j with the Omega ratio (in columns) knowing that the
investor is ranked in decile i according to the Sharpe ratio (in rows).
Sharpe/Omega - 2003
1 2 3 4 5 6 7 8 9 10
1 100,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0
2 55,9 43,9 0,2 0,0 0,0 0,0 0,0 0,0 0,0 0,0
3 0,0 49,8 49,3 0,8 0,0 0,0 0,0 0,0 0,0 0,0
4 0,0 0,0 43,8 52,8 2,9 0,4 0,0 0,0 0,0 0,0
5 0,0 0,0 0,5 39,6 52,6 6,4 0,5 0,2 0,1 0,1
6 0,0 0,0 0,0 0,6 37,7 52,7 8,0 0,6 0,3 0,1
7 0,0 0,0 0,0 0,0 0,6 33,6 56,0 9,2 0,4 0,2
8 0,0 0,0 0,0 0,0 0,0 0,6 28,9 56,7 13,2 0,6
9 0,0 0,0 0,0 0,0 0,0 0,0 0,4 27,1 61,9 10,6
10 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 17,9 82,1
Sharpe/Omega - 2004
1 2 3 4 5 6 7 8 9 10
1 40,4 40,4 19,1 0,0 0,0 0,0 0,0 0,0 0,0 0,0
2 0,0 0,0 63,0 37,0 0,0 0,0 0,0 0,0 0,0 0,0
3 0,0 0,0 0,0 79,8 20,1 0,0 0,0 0,0 0,0 0,0
4 0,0 0,0 0,0 0,3 95,2 4,5 0,0 0,0 0,0 0,0
5 0,0 0,0 0,0 0,0 6,6 91,4 2,0 0,0 0,0 0,0
6 0,0 0,0 0,0 0,0 0,0 23,8 74,7 1,4 0,0 0,0
7 0,0 0,0 0,0 0,0 0,0 0,0 42,8 56,5 0,6 0,0
8 0,0 0,0 0,0 0,0 0,0 0,0 0,0 61,6 38,2 0,2
9 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 79,7 20,3
10 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 1,0 99,0
Sharpe/Omega - 2005
1 2 3 4 5 6 7 8 9 10
1 100,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0
2 29,8 69,5 0,7 0,0 0,0 0,0 0,0 0,0 0,0 0,0
3 0,0 27,1 70,9 1,8 0,1 0,1 0,0 0,0 0,0 0,0
4 0,0 0,0 25,1 69,0 5,0 0,4 0,2 0,1 0,1 0,0
5 0,0 0,0 0,0 26,0 65,8 6,2 1,1 0,4 0,2 0,4
6 0,0 0,0 0,0 0,0 25,8 63,6 8,5 1,4 0,4 0,4
7 0,0 0,0 0,0 0,0 0,0 26,4 55,2 16,8 1,3 0,4
8 0,0 0,0 0,0 0,0 0,0 0,1 30,8 54,4 13,2 1,6
9 0,0 0,0 0,0 0,0 0,0 0,0 0,2 24,4 64,8 10,7
10 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 16,7 83,3
Sharpe/Omega - 2006
1 2 3 4 5 6 7 8 9 10
1 100,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0
2 3,7 94,6 1,7 0,0 0,0 0,0 0,0 0,0 0,0 0,0
3 0,0 4,9 89,9 5,1 0,1 0,0 0,0 0,0 0,0 0,0
4 0,0 0,0 8,0 83,8 7,9 0,2 0,0 0,0 0,0 0,0
5 0,0 0,0 0,0 10,7 78,1 10,3 0,8 0,1 0,0 0,1
6 0,0 0,0 0,0 0,0 13,6 73,7 10,9 1,5 0,2 0,2
7 0,0 0,0 0,0 0,0 0,0 15,4 71,7 11,7 0,8 0,4
8 0,0 0,0 0,0 0,0 0,0 0,0 16,2 72,0 10,8 1,0
9 0,0 0,0 0,0 0,0 0,0 0,0 0,0 14,4 75,6 10,1
10 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 12,2 87,8
27
Table 5 Transition matrices - Sharpe ratio/Upside Potential ratio
This table presents the transition matrices between the ranking resulting from
the Sharpe ratio evaluation and the Upside Potential ratio evaluation for
2003, 2004, 2005 and 2006. We report the conditional probability to be
ranked in decile j with the Upside Potential ratio (in columns) knowing that
the investor is ranked in decile i according to the Sharpe ratio (in rows).
Sharpe/UPR - 2003
1 2 3 4 5 6 7 8 9 10
1 97,9 2,0 0,1 0,0 0,0 0,0 0,0 0,0 0,0 0,0
2 47,3 39,0 10,4 2,5 0,5 0,3 0,0 0,0 0,0 0,0
3 8,0 36,2 32,1 15,9 5,3 1,7 0,6 0,3 0,0 0,0
4 1,0 12,1 32,3 28,9 16,0 6,6 2,0 0,8 0,2 0,0
5 0,3 3,6 12,8 26,8 30,3 17,5 6,2 1,8 0,5 0,2
6 0,0 1,2 4,2 13,3 26,3 30,2 17,8 5,8 1,2 0,1
7 0,1 0,6 1,3 4,2 11,6 25,3 32,5 19,4 4,5 0,6
8 0,0 0,2 0,5 1,7 3,0 8,9 26,1 36,1 21,5 2,0
9 0,0 0,1 0,2 0,4 0,8 3,0 7,6 25,3 46,6 16,0
10 0,0 0,0 0,0 0,0 0,1 0,5 1,1 4,3 19,1 74,9
Sharpe/UPR - 2004
1 2 3 4 5 6 7 8 9 10
1 39,4 34,4 21,2 4,4 0,4 0,1 0,0 0,0 0,0 0,0
2 2,3 12,5 34,2 38,9 11,1 1,1 0,0 0,0 0,0 0,0
3 0,3 3,5 14,6 36,6 31,1 13,5 0,4 0,0 0,0 0,0
4 0,2 1,0 5,5 19,3 42,6 26,1 5,0 0,4 0,0 0,0
5 0,1 0,6 1,8 8,4 23,3 39,6 23,4 2,4 0,4 0,0
6 0,0 0,1 0,5 2,3 8,9 27,8 42,1 16,6 1,7 0,0
7 0,0 0,0 0,0 0,6 1,5 10,0 34,6 42,8 10,1 0,1
8 0,0 0,0 0,0 0,0 0,3 1,1 11,4 42,8 41,8 2,5
9 0,0 0,0 0,0 0,0 0,0 0,4 2,6 14,1 57,6 25,3
10 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,4 7,9 91,6
Sharpe/UPR - 2005
1 2 3 4 5 6 7 8 9 10
1 97,4 2,6 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0
2 29,9 58,3 10,7 0,8 0,0 0,2 0,0 0,0 0,0 0,0
3 1,5 32,5 44,0 17,4 3,6 0,9 0,1 0,0 0,0 0,0
4 0,0 4,0 32,6 36,0 17,3 6,8 2,2 0,7 0,3 0,1
5 0,0 0,2 8,7 28,7 29,5 19,1 8,1 3,6 1,0 1,0
6 0,0 0,0 0,7 11,8 28,0 25,2 18,1 11,3 3,6 1,3
7 0,0 0,0 0,0 1,8 13,9 24,4 23,1 16,8 17,5 2,5
8 0,0 0,0 0,0 0,1 3,4 17,0 25,5 23,1 18,7 12,2
9 0,0 0,0 0,0 0,0 0,2 3,9 17,8 28,2 26,8 23,2
10 0,0 0,0 0,0 0,0 0,0 0,0 1,8 13,0 28,8 56,4
Sharpe/UPR - 2006
1 2 3 4 5 6 7 8 9 10
1 72,8 20,7 5,7 0,8 0,1 0,0 0,0 0,0 0,0 0,0
2 22,1 40,1 19,2 15,9 2,0 0,4 0,1 0,3 0,0 0,0
3 6,3 26,7 32,3 20,3 9,1 3,5 1,5 0,3 0,0 0,0
4 1,2 9,4 25,1 25,7 20,8 11,9 4,5 1,1 0,4 0,0
5 0,1 2,9 11,2 20,8 24,7 17,9 15,2 5,6 1,3 0,2
6 0,0 0,6 4,9 10,5 22,1 26,4 15,3 12,3 6,8 1,0
7 0,0 0,0 1,3 4,2 13,6 21,5 29,5 18,6 8,8 2,5
8 0,0 0,0 0,3 1,4 6,2 13,2 23,1 29,5 20,0 6,3
9 0,0 0,0 0,0 0,1 1,0 4,4 9,4 26,8 37,8 20,5
10 0,0 0,0 0,0 0,0 0,1 0,4 0,9 5,2 24,4 69,0
28
Table 6 Transition matrices - Sharpe ratio/Farinelli-Tibiletti(0.5-2) ratio
This table presents the transition matrices between the ranking resulting from
the Sharpe ratio evaluation and the Farinelli-Tibiletti(0.5-2) ratio evaluation
for 2003, 2004, 2005 and 2006. We report the conditional probability to be
ranked in decile j with the Farinelli-Tibiletti(0.5-2) ratio (in columns) knowing
that the investor is ranked in decile i according to the Sharpe ratio (in rows).
Sharpe/Farinelli-Tibiletti(0.5-2) - 2003
1 2 3 4 5 6 7 8 9 10
1 83,2 12,4 2,9 0,8 0,4 0,2 0,1 0,0 0,0 0,0
2 30,4 32,8 18,0 7,7 4,8 3,6 0,9 0,6 0,3 0,8
3 12,1 22,5 21,7 17,7 11,4 6,7 4,2 2,5 1,0 0,3
4 6,6 10,5 17,1 19,9 15,2 12,0 11,3 4,2 2,3 0,8
5 3,1 7,5 12,8 15,4 16,4 15,9 13,9 10,1 4,2 0,9
6 5,5 5,8 8,0 11,3 14,1 16,4 17,0 13,0 6,8 2,2
7 1,6 4,2 5,8 8,5 11,7 15,5 16,9 17,7 13,3 4,9
8 1,9 2,8 4,3 6,5 10,5 12,2 14,0 19,5 20,8 7,6
9 0,5 2,0 3,2 3,9 5,8 6,9 10,0 17,1 26,6 24,0
10 0,6 1,1 2,0 2,7 3,7 4,5 5,6 9,1 18,4 52,3
Sharpe/Farinelli-Tibiletti(0.5-2) - 2004
1 2 3 4 5 6 7 8 9 10
1 34,1 23,8 15,4 10,1 7,1 4,3 2,6 1,5 1,1 0,0
2 8,5 15,8 17,5 17,8 14,9 11,0 8,7 3,3 2,3 0,3
3 3,8 10,5 12,9 15,2 14,6 13,1 11,8 8,2 8,9 1,0
4 3,8 8,3 13,0 13,2 14,1 12,5 12,4 16,8 5,0 0,9
5 1,5 6,3 10,9 14,5 13,2 14,7 13,9 13,2 9,2 2,7
6 0,5 4,2 7,5 10,1 14,7 15,7 16,1 12,5 12,5 6,2
7 0,2 1,4 4,9 10,3 12,7 15,5 15,6 15,5 14,2 9,7
8 0,2 1,0 2,0 5,2 9,6 14,7 16,7 19,5 19,4 11,8
9 0,2 1,5 5,3 2,8 4,5 8,1 13,4 18,7 24,1 21,4
10 0,0 0,0 0,1 0,6 0,4 1,7 3,4 7,8 20,6 65,4
Sharpe/Farinelli-Tibiletti(0.5-2) - 2005
1 2 3 4 5 6 7 8 9 10
1 72,2 23,0 4,4 0,3 0,0 0,0 0,1 0,0 0,0 0,0
2 35,7 30,3 18,9 10,2 2,6 1,6 0,4 0,2 0,1 0,0
3 10,4 29,5 27,6 16,6 9,1 3,6 1,4 0,9 0,5 0,3
4 2,2 16,1 23,7 21,4 15,3 9,1 5,9 4,0 1,7 0,5
5 0,1 4,3 15,1 24,4 20,5 13,8 9,8 6,8 3,0 2,0
6 0,0 1,0 6,9 15,6 23,2 19,7 14,3 10,1 5,6 3,7
7 0,0 0,0 1,3 6,5 16,8 21,6 23,0 13,1 10,2 7,5
8 0,0 0,0 0,3 1,5 7,6 17,2 22,0 21,4 15,3 14,7
9 0,0 0,0 0,0 0,2 1,5 7,2 17,9 26,1 30,2 17,0
10 0,0 0,0 0,0 0,0 0,0 0,9 3,9 14,1 30,1 51,0
Sharpe/Farinelli-Tibiletti(0.5-2) - 2006
1 2 3 4 5 6 7 8 9 10
1 66,7 20,3 8,1 3,9 0,7 0,3 0,0 0,1 0,0 0,0
2 18,9 28,0 19,1 16,2 6,2 9,9 1,1 0,4 0,2 0,0
3 10,4 24,3 24,6 14,7 10,1 7,3 4,5 2,6 1,4 0,1
4 4,2 13,3 20,1 19,9 14,9 9,4 7,5 6,6 3,3 0,7
5 1,0 9,7 14,6 18,2 19,2 14,1 9,5 7,9 4,2 1,6
6 0,6 2,9 7,8 13,2 19,6 16,8 14,6 11,1 8,0 5,5
7 0,2 1,0 3,5 8,0 13,3 19,3 24,7 13,8 10,3 5,8
8 0,2 0,7 1,5 4,0 10,9 13,9 19,5 21,9 17,1 10,3
9 0,1 0,1 0,6 1,3 4,1 7,2 14,0 24,6 26,0 21,9
10 0,0 0,1 0,1 0,5 0,6 1,4 4,1 10,7 29,0 53,6
29
Table 7 Transition matrices - Sharpe ratio/Farinelli-Tibiletti(0.8-0.85) ratio
This table presents the transition matrices between the ranking resulting from
the Sharpe ratio evaluation and the Farinelli-Tibiletti(0.8-0.85) ratio evaluation
for 2003, 2004, 2005 and 2006. We report the conditional probability to be
ranked in decile j with the Farinelli-Tibiletti(0.8-0.85) ratio (in columns) knowing
that the investor is ranked in decile i according to the Sharpe ratio (in rows).
Sharpe/Farinelli-Tibiletti(0.8-0.85) - 2003
1 2 3 4 5 6 7 8 9 10
1 50,3 22,6 13,2 8,8 3,6 1,4 0,0 0,1 0,0 0,0
2 30,2 21,7 16,3 11,7 7,2 6,0 3,4 1,7 1,5 0,2
3 16,9 21,1 17,9 13,9 12,1 7,5 4,3 4,2 1,9 0,2
4 8,9 13,8 15,9 18,0 13,4 10,5 8,2 7,6 3,0 0,8
5 4,9 9,3 12,3 16,2 16,5 15,1 12,7 7,7 3,9 1,4
6 6,8 6,7 9,2 10,9 15,6 16,4 16,3 9,7 6,1 2,2
7 2,5 5,2 6,7 7,6 11,4 16,8 18,1 16,6 10,5 4,7
8 2,5 3,2 5,3 5,9 7,5 9,7 15,5 20,5 22,6 7,3
9 1,3 2,7 3,2 3,8 4,8 6,6 9,6 16,8 26,9 24,4
10 0,9 1,6 1,9 2,6 3,8 4,7 5,6 9,0 17,4 52,6
Sharpe/Farinelli-Tibiletti(0.8-0.85) - 2004
1 2 3 4 5 6 7 8 9 10
1 34,3 20,1 12,9 9,5 7,1 4,9 4,2 3,0 2,2 1,7
2 8,9 19,1 17,0 14,3 10,7 9,8 7,2 6,9 4,4 1,7
3 4,7 13,3 14,2 13,0 11,1 10,6 8,9 7,2 14,8 2,3
4 2,7 12,1 15,2 13,1 13,1 10,2 14,6 8,8 7,3 3,0
5 1,3 8,5 14,2 15,4 13,0 13,2 11,9 9,7 8,1 4,7
6 0,4 4,7 10,0 14,0 15,5 15,0 13,0 10,9 8,9 7,4
7 0,1 1,9 7,2 12,3 15,7 15,4 14,1 13,0 13,3 6,9
8 0,0 0,4 3,1 7,0 13,0 16,8 16,5 20,3 13,3 9,5
9 0,0 0,0 0,4 2,4 6,1 12,6 17,0 23,1 19,3 19,0
10 0,0 0,0 0,0 0,0 0,4 1,5 4,1 10,8 23,3 59,9
Sharpe/Farinelli-Tibiletti(0.8-0.85) - 2005
1 2 3 4 5 6 7 8 9 10
1 73,1 22,1 3,4 1,0 0,3 0,1 0,0 0,1 0,0 0,0
2 37,2 29,6 14,2 8,5 4,3 2,0 1,6 1,2 1,0 0,3
3 9,5 31,5 27,0 12,7 7,6 4,6 3,2 2,0 1,0 0,8
4 1,1 17,2 27,2 20,8 11,8 7,7 6,5 3,6 2,2 2,0
5 0,0 3,2 19,1 26,6 18,2 11,6 7,3 6,6 4,1 3,4
6 0,0 0,4 5,9 20,0 25,7 18,2 10,8 8,2 5,7 4,9
7 0,0 0,0 0,7 6,5 19,4 23,7 21,8 10,7 8,9 8,2
8 0,0 0,0 0,1 1,1 8,3 21,0 22,6 19,3 13,4 14,3
9 0,0 0,0 0,0 0,2 1,1 7,3 19,6 29,0 28,2 14,7
10 0,0 0,0 0,0 0,0 0,1 0,4 3,2 16,1 32,1 48,1
Sharpe/Farinelli-Tibiletti(0.8-0.85) - 2006
1 2 3 4 5 6 7 8 9 10
1 63,7 17,7 6,6 4,9 2,9 1,7 1,2 1,1 0,3 0,0
2 20,6 25,5 21,6 9,8 14,0 3,1 2,0 1,6 1,2 0,7
3 11,0 29,2 19,9 13,1 7,0 7,2 4,8 4,1 2,9 0,9
4 4,5 14,5 23,2 18,9 11,0 8,6 7,0 6,4 4,4 1,5
5 1,2 9,3 16,7 21,1 16,5 11,4 8,8 6,8 5,8 2,5
6 0,8 2,6 7,3 18,0 18,5 16,9 12,0 10,4 7,3 6,3
7 0,2 1,0 3,1 8,7 15,7 21,8 20,9 11,7 9,8 6,9
8 0,2 0,6 1,1 4,3 10,8 18,5 20,6 18,3 14,4 11,1
9 0,0 0,0 0,4 0,8 2,8 9,2 18,5 25,7 22,8 19,7
10 0,0 0,0 0,0 0,1 0,4 1,4 3,8 13,5 30,4 50,3
30
Table 8 Transition matrices - Sharpe ratio/Farinelli-Tibiletti(1.5-2) ratio
This table presents the transition matrices between the ranking resulting from
the Sharpe ratio evaluation and the Farinelli-Tibiletti(1.5-2) ratio evaluation for
2003, 2004, 2005 and 2006. We report the conditional probability to be
ranked in decile j with the Farinelli-Tibiletti(1.5-2) ratio (in columns) knowing
that the investor is ranked in decile i according to the Sharpe ratio (in rows).
Sharpe/Farinelli-Tibiletti(1.5-2) - 2003
1 2 3 4 5 6 7 8 9 10
1 16,4 9,6 9,1 7,3 6,7 8,8 9,0 12,3 11,1 9,7
2 16,5 15,2 12,1 11,2 10,0 8,5 8,8 7,8 6,1 3,9
3 13,9 16,3 15,5 12,9 11,0 9,3 7,1 6,6 4,8 2,6
4 10,6 13,3 15,3 13,1 11,3 9,9 8,9 8,7 4,9 3,9
5 8,8 11,5 13,4 13,6 12,6 13,1 9,8 8,0 5,1 4,1
6 11,3 9,7 10,6 13,4 12,6 13,5 10,6 7,9 6,5 3,9
7 7,1 7,5 8,3 11,0 14,1 12,6 13,0 11,0 9,7 5,7
8 6,8 6,4 6,9 7,0 9,7 11,8 12,2 14,1 13,9 11,0
9 5,5 5,7 3,9 5,3 5,8 7,4 12,2 15,2 21,7 17,3
10 7,2 4,5 4,2 3,4 4,1 4,4 7,8 9,8 16,8 37,8
Sharpe/Farinelli-Tibiletti(1.5-2) - 2004
1 2 3 4 5 6 7 8 9 10
1 23,9 13,8 10,9 8,6 7,8 7,8 7,2 7,8 6,7 5,5
2 12,9 14,1 11,6 8,9 8,6 8,7 8,9 10,2 8,7 7,3
3 8,8 12,2 10,4 10,0 7,7 7,5 9,4 8,5 9,3 16,2
4 8,8 11,4 10,8 11,1 9,1 9,0 14,2 8,0 8,6 9,1
5 7,0 10,9 13,2 11,4 11,7 10,2 9,2 10,1 8,4 7,9
6 5,2 10,5 12,0 11,1 10,6 12,1 9,6 10,5 10,0 8,4
7 3,3 8,1 11,1 13,5 11,6 12,4 10,7 9,4 12,5 7,4
8 2,0 6,5 9,5 11,9 13,0 14,0 12,5 14,1 9,8 6,7
9 0,9 4,2 7,3 11,8 16,6 12,2 12,1 11,5 12,3 11,1
10 0,0 0,6 1,5 4,5 7,7 10,4 11,8 14,3 20,1 29,0
Sharpe/Farinelli-Tibiletti(1.5-2) - 2005
1 2 3 4 5 6 7 8 9 10
1 37,0 20,7 9,7 9,2 15,1 2,3 2,0 1,7 1,4 0,9
2 26,7 15,0 11,6 10,2 7,2 6,3 6,6 4,8 4,8 6,8
3 16,9 17,1 11,5 9,8 7,4 9,8 7,5 7,3 6,6 6,2
4 12,0 12,0 12,7 11,3 10,0 8,5 9,1 8,5 8,6 7,3
5 7,7 12,5 12,8 10,1 10,4 9,4 9,8 8,6 9,4 9,3
6 4,6 10,3 12,5 9,5 10,5 11,3 11,6 11,1 9,1 9,5
7 2,1 8,0 9,4 11,2 9,9 10,4 10,4 11,3 16,0 11,2
8 1,4 4,9 9,6 11,0 10,5 12,6 11,1 11,2 12,8 15,0
9 0,3 2,5 7,3 11,0 11,8 12,1 12,5 15,5 12,4 14,7
10 0,1 0,5 2,9 6,5 9,0 14,8 16,8 17,4 16,1 15,9
Sharpe/Farinelli-Tibiletti(1.5-2) - 2006
1 2 3 4 5 6 7 8 9 10
1 40,6 11,8 8,8 8,0 5,4 5,8 6,0 5,4 4,3 4,0
2 17,1 13,4 15,6 7,8 6,6 5,4 8,2 7,3 11,7 6,6
3 17,3 18,1 11,3 10,3 8,5 7,4 6,9 7,8 6,1 6,2
4 10,4 16,7 12,5 12,1 10,9 10,5 7,7 6,9 7,2 5,0
5 6,7 12,9 13,7 13,0 14,6 10,9 9,7 6,1 7,7 4,9
6 4,3 10,5 12,1 14,0 12,1 11,5 11,3 7,2 8,6 8,4
7 1,9 8,0 10,4 11,8 13,0 12,0 13,5 14,5 8,1 6,8
8 1,7 5,6 9,1 12,1 13,3 14,2 12,0 12,7 10,4 9,1
9 0,9 2,4 4,9 7,8 10,8 14,1 13,6 15,3 15,8 14,4
10 0,4 0,8 1,6 3,0 4,7 8,0 10,7 14,9 21,5 34,3
31
Table 9 Impact of the choice of alternative measures
This table contains the change in decile of investors when their performance is
evaluated with alternative measures rather than with the Sharpe ratio. The
maximum downgrade and upgrade are presented as well as the proportion of
investors who remain in the same decile, who are downgraded and upgraded.
Max. Max. No change Down- Up-
down- up- (%) graded graded
grade grade (%) (%)
2003
Omega ratio -5 2 58.20 5.85 35.94
Upside Potential ratio -6 8 41.34 19.07 39.59
Farinelli-Tibiletti(0.5-2) ratio -8 9 27.09 32.50 40.41
Farinelli-Tibiletti(0.8-0.85) ratio -8 9 24.28 33.65 42.07
Farinelli-Tibiletti(1.5-2) ratio -9 9 17.35 36.19 46.45
2004
Omega ratio -3 1 36.25 63.67 0.08
Upside Potential ratio -5 6 36.84 55.96 7.20
Farinelli-Tibiletti(0.5-2) ratio -9 8 24.78 50.04 25.19
Farinelli-Tibiletti(0.8-0.85) ratio -9 7 24.22 46.35 29.43
Farinelli-Tibiletti(1.5-2) ratio -9 9 16.40 47.33 36.27
2005
Omega ratio -6 2 68.61 7.36 24.03
Upside Potential ratio -6 4 40.07 22.84 37.09
Farinelli-Tibiletti(0.5-2) ratio -8 5 30.36 27.76 41.88
Farinelli-Tibiletti(0.8-0.85) ratio -8 5 28.97 25.83 45.20
Farinelli-Tibiletti(1.5-2) ratio -9 9 13.87 38.85 47.28
2006
Omega ratio -5 1 82.64 7.41 9.95
Upside Potential ratio -6 5 38.65 29.09 32.26
Farinelli-Tibiletti(0.5-2) ratio -8 8 29.99 32.54 37.47
Farinelli-Tibiletti(0.8-0.85) ratio -8 7 27.21 31.38 41.41
Farinelli-Tibiletti(1.5-2) ratio -9 9 17.90 36.74 45.35
32
Table 10 Decile of the market index performance
This table reports each year between 2003 and 2006 the decile of the mar-
ket index according to 6 performance measures. These grades are based on
the initial ranking computed each year for each measure across 24,766 investors.
2003 2004 2005 2006
Sharpe ratio 6 9 9 9
Omega ratio 6 9 9 9
Upside Potential ratio 4 7 9 10
Farinelli-Tibiletti(0.5-2) ratio 3 7 8 6
Farinelli-Tibiletti(0.8-0.85) ratio 3 6 7 5
Farinelli-Tibiletti(1.5-2) ratio 1 1 4 5
33
Figure 3 Variations of (p-q) in the Farinelli-Tibiletti ratio - 2003
Figure 4 Variations of (p-q) in the Farinelli-Tibiletti ratio - 2004
Figure 5 Variations of (p-q) in the Farinelli-Tibiletti ratio - 2005
Figure 6 Variations of (p-q) in the Farinelli-Tibiletti ratio - 2006
34
Table 11 Performance persistance
This table reports the number and the proportion of investors who are
ranked in a higher decile than the market index with each alternative per-
formance measure. The rst row indicates the number of consecutive year
(1, 2, 3 or 4 starting from 2003) for which investors beat the market.
1 year % 2 years % 3 years % 4 years %
Sharpe ratio 10564 42.66 1157 4.67 277 1.12 66 0.27
Omega ratio 9906 40.00 1249 5.04 254 1.03 56 0.23
Upside Potential ratio 14860 60.00 4469 18.04 609 2.46 0 0.00
Farinelli-Tibiletti(0.5-2) ratio 17336 70.00 5436 21.95 1136 4.59 640 2.58
Farinelli-Tibiletti(0.8-0.85) ratio 17336 70.00 7204 29.09 2118 8.55 1471 5.94
Farinelli-Tibiletti(1.5-2) ratio 22289 90.00 20151 81.37 11927 48.16 7549 30.48
35
Table 12 Market index decile among hazard portfolio
This table reports each year between 2003 and 2006 the decile of the
performance of the market index according to 6 performance measures.
These grades are based on the ranking of 24,677 randomly created port-
folios that mimic the diversication level of investors in our sample.
2003 2004 2004 2006
Sharpe ratio 4 9 10 7
Omega ratio 3 9 10 9
Upside Potential ratio 2 7 9 9
Farinelli-Tibiletti(0.5-2) ratio 1 1 1 1
Farinelli-Tibiletti(0.8-0.85) ratio 1 1 1 1
Farinelli-Tibiletti(1.5-2) ratio 1 1 1 1
36
References
Ang, A. and Y. Chen, J.and Xing (2006). Downside risk. Review of Financial Studies,
19(4), 11911239.
Barber, B. M., Y. Lee, Y. Liu, and T. Odean (2009). Just how much do individual
investors lose by trading? Review of Financial Studies, 22(2), 609632.
Barber, B. M. and T. Odean (2000). Trading is hazardous to your wealth: The common
stock investment performance of individual investors. Journal of Finance, 55(2), 773
806.
Barberis, N. and M. Huang (2008). Stock as lotteries: The implication of the probability
weighting for security prices. American Economic Review, 98, 20662100.
Cogneau, P. and G. Hubner (2009a). The (more than) 100 ways to measure portfolio
performance. part 1: Standardized risk-adjusted measures. Journal of Performance
Measurement, 13(4), 5671.
Cogneau, P. and G. Hubner (2009b). The (more than) 100 ways to measure portfo-
lio performance. part 2: Special measures and comparison. Journal of Performance
Measurement, 14(1), 5669.
Coval, J., D. Hirshleifer, and T. Shumway (2005). Can individual investors beat the
market? Working paper, Harvard Business School.
Eling, M. (2008). Does the measure matter in the mutual fund industry? Financial
Analysts Journal, 64(3), 113.
Eling, M. and F. Schuhmacher (2007). Does the choice of performance measure inuence
the evaluation of hedge funds? Journal of Banking and Finance, 31, 26322647.
Fama, E. F. and K. French (1993). Common risk factors in the returns of bonds and
stocks. Journal of Financial Economics, 33, 356.
37
Farinelli, S. and L. Tibiletti (2008). Sharpe thinking in asset ranking with one-sided
measures. European Journal of Operational Research, 185(3), 15421547.
Friedman, M. and L. Savage (1948). The utility analysis of choices involving risk. Journal
of Political Economy, 56, 279304.
Goetzmann, W. and A. Kumar (2008). Equity portfolio diversication. Review of Finance,
12(3), 433.
Griths, D. (1980). A pragmatic approach to spearmans rank correlation coecient.
Teaching Statistics, 2, 1013.
Jensen, M. C. (1968). The performance of mutual funds and tests of portfolio eciency.
Journal of Finance, 23, 389416.
Kahneman, D., J. Knetsch, and R. H. Thaler (1990). Anomalies: The endowment eect,
loss aversion, and status quo bias. Journal of Economic Perspectives, 5(1), 193206.
Kahneman, D. and A. Tversky (1979). Prospect theory : An analysis of decision under
risk. Econometrica, 47, 263291.
Keating, C. and W. Shadwick (2002). A universal performance measure. Journal of
Performance Measurement, 59-84, 6(3).
Kumar, A. (2009). Who gambles in the stock market? Journal of Finance, 64, 18891933.
Markowitz, H. (1952). Portfolio selection. Journal of Finance, 7, 7791.
Mitton, T. and K. Vorkink (2007). Equilibrium underdiversication and the preference
for skewness. Review of Financial Studies, 20, 12551288.
Noether, G. (1981). Why kendall tau. The Best of Teaching Statistics, 3, 4143.
Odean, T. (1999). Do investors trade too much? American Economic Review, 89, 1279
1298.
Sharpe, W. (1966). Mutual fund performance. Journal of Business, 39, 119138.
38
Shefrin, H. and M. Statman (2000). Behavioral portfolio theory. Journal of Financial
and Quantitative Analysis, 35, 127151.
Sortino, F., R. Van Ver Meer, and A. Plantinga (1999). The dutch triangle. Journal of
Portfolio Management, 26, 5059.
Statman, M. (1987). How many stocks make a diversied portfolio? Journal of Financial
and Quantitative Analysis, 22(3), 353363.
Statman, M. (2002). Lottery players/stock traders. Financial Analysts Journal, January-
February, 1421.
Tversky, A. and D. Kahneman (1992). Advances in prospect theory: Cumulative repre-
sentation of uncertainty. Journal of risk and uncertainty, 5, 297323.
Unser, M. (2000). Lower partial moments as measures of perceived risk: An experimental
study. Journal of Economic Psychology, 21, 253280.
Veld, C. and Y. Veld-Merkoulova (2008). The risk perceptions of individual investors.
Journal of Economic Psychology, 29), 226252.
Von Neumann, J. and O. Morgenstern (1947). Theory of Games and Economic Behavior.
Princeton University Press (1ed, 1944).
Zakamouline, V. (2011). The choice of performance measure does inuence the evaluation
of hedge funds. Journal of Performance Measure, 15, 4864.
39