1
Persistent doubt: An examination of hedge fund
performance
María de la O. González
School of Economic & Business Sciences, University of Castilla–La Mancha, Plaza
de la Universidad 1, 02071, Albacete, Spain
Email: mariao.gonzalez@uclm.es
Nicolas A. Papageorgiou
HEC Montreal, 3000 Cote Ste Catherine, Montreal, H3T2A7, Canada
E-mail: nicolas.papageorgiou@hec.ca
Frank S. Skinner
Department of Economics and Finance, Brunel University, Uxbridge, London, UB8
3PH, United Kingdom
E-mail: frank.skinner@brunel.ac.uk
Abstract
We examine whether performance persistence is suspicious. Top quintile
portfolios formed on the Sharpe ratio, alpha, and information ratio persistently
outperform similarly constructed mediocre third quintile portfolios throughout our
sample period, but performance is more modest and less persistent when portfolios are
formed on the excess manipulation-proof performance measure (EMPPM). By
selecting funds formed on ranking by Sharpe and information ratios, investors also
select funds that have persistently doubtful performance according to the doubt ratio.
In contrast, portfolios formed on alphas and especially the EMPPM have much less
excess and persistent doubt.

We thank the editor (John Doukas) and an anonymous reviewer for European Financial Management
for excellent comments and suggestions. We are also grateful to Ephraim Clark, Richard McGee, Frank
McGroarty, Thomas and Julia Henker, the seminar participants of the Adam Smith Seminar Series, and
the conference participants of Infinity and FEBS for their helpful comments. We use hedge fund risk
factors from David A. Hsieh’s data library and gratefully acknowledge David A. Hsieh for making
these data publically available. This work was supported by Junta de Comunidades de Castilla–La
Mancha and Ministerio de Economía y Competitividad (grant numbers PEII-2014-019-P and
ECO2014-59664-P, respectively). Any errors are the responsibility of the authors. Corresponding
author: Frank S. Skinner.
2
Keywords: Hedge funds, performance measures, manipulation-proof performance
measure, doubt ratio
JEL classification: G11; G1; G23; G24
3
1. Introduction
Hedge fund performance and performance persistence are a controversial
issue. Many authors, such as Ammann et al. (2013) and Boyson (2008), find that
hedge funds persistently deliver positive alphas out of sample. Other research is more
critical, claiming that performance persistence depends upon the measure used
(Capocci et al. (2005)), that performance persistence is attributable to passive
investments in illiquid assets (Brandon and Wang 2013), and even that performance
persistence for top funds does not exist at all (Slavutskaya 2013). Recently, Dichev
and Yu (2011) document a sharp reduction in buy and hold returns for a very large
sample of hedge funds and commodity trading advisors, from 18.7% from 1980 to
1994 to 9.5% from 1995 to 2008. Clearly, there are doubts concerning the
performance, performance persistence, and measurement of the performance of hedge
funds.
Meanwhile, Goetzmann et al. (2007) demonstrate that common performance
measures such as the Sharpe ratio, alpha, and information ratio can be subject to
manipulation, deliberate or otherwise. To address this issue, the authors develop a
manipulation-proof performance measure (MPPM), so called because it is robust to
the underlying return distribution and to gaming by hedge fund managers.
Additionally, Brown et al. (2010) develop the doubt ratio. This measure is a
diagnostic statistic derived from the MPPM and indicates whether the reported returns
from hedge funds are suspicious. Together, the MPPM and the doubt ratio allow us to
examine recent hedge fund performance and shed light on whether top hedge funds
achieve superior performance, whether this performance persists, whether
performance persistence is doubtful, whether doubts concerning hedge fund
4
performance persist, and whether performance persistence is down to skill,
manipulation, or luck.
We test for persistence in hedge fund performance by forming portfolios
according to five different performance measures that are further stratified by quintile
and strategy. The five measures are the Sharpe ratio; the alpha from Fung and Hsieh’s
(2004) hedge fund asset pricing model augmented by an eighth factor for emerging
markets; the information ratio, obtained as the t-statistic from Fung and Hsieh’s
alphas; the MPPM of Goetzmann et al. (2007); and the doubt ratio of Brown et al.
(2010). The Sharpe ratio, information ratio, and alpha enable us to compare our
results with the recent literature and allow us to highlight how the picture changes
when we evaluate the performance and performance persistence of top quintile hedge
funds with the MPPM and doubt ratio.
Each portfolio is formed monthly, between January 31, 2001, and December
31, 2010, using the prior month’s performance and performance is tracked for two
years. We thus form 120 overlapping out-of-sample portfolios for each strategy and
quintile, each of which is held for two years. This ensures that we only use
information available on the date that the portfolios are formed so that our subsequent
examination of monthly out-of-sample performance mimics the investment activities
of portfolio managers. We then test for superior performance using bootstrap t-tests.
This testing strategy allows us to address several questions. Do hedge funds
deliver consistent superior performance, on average, throughout the sample period
and in the bull and bear markets around the December 31, 2006, pivot date? Can topperforming hedge funds persist in delivering superior performance? If so, for how
long will this superior performance persist? Are the answers to these questions
sensitive to the choice of performance measure? Specifically, is there any evidence
5
that the hedge fund industry manipulates common performance measures? Is the
newly developed MPPM resilient to manipulation? Is performance persistence due to
skill, manipulation, or luck? Finally, if performance and performance persistence are
doubtful, do these doubts persist?
A key innovation in this paper is to reexamine the issue of performance
persistence using the MPPM developed by Goetzmann et al. (2007). Since the MPPM
is a relative performance measure, we benchmark it off the corresponding MPPM for
the Russell 2000 so we are able to draw conclusions as to whether the candidate hedge
fund strategy outperforms a passive buy and hold strategy. Specifically, we subtract
the corresponding MPPM of the Russell 2000 total return index from the
corresponding measure for each hedge fund.
We also examine the reliability of our conclusions by examining the
performance and performance persistence of hedge funds by using the doubt ratio of
Brown et al. (2010). This diagnostic statistic is derived from the MPPM and indicates
if the reported returns from hedge funds are suspicious. We use the doubt ratio to see
if top quintile funds as ranked by the heavily criticized Sharpe ratio, alpha, and
information ratio exhibit suspicious returns. As a control mechanism, we examine top
quintile funds as ranked by the excess MPPM (EMPPM) to see if the doubt ratio
detects suspicious performance.
The latter is a possibility because the doubt ratio does not detect misdeeds on
behalf of hedge fund managers. Instead, it can detect suspicious patterns in hedge
fund returns such as consistent positive returns with low risk. These suspicious returns
could be the result of dubious trading practices, such as income smoothing, or
legitimate trading strategies, such as locking in an arbitrage premium that is realizable
only at maturity of the position. For instance, a positive carry trade in the futures
6
market can have a high doubt ratio due to a sequence of small positive returns being
recognized by daily mark to market. The latter could be a legitimate reason why top
quintile funds formed on the EMPPM can still have a high doubt ratio. As Brown et
al. (2010) point out, a high doubt ratio, while suspicious, is not necessarily evidence
of dubious trading practices but, rather, should incite further investigation to
determine why the doubt ratio is high.
This paper contributes to our understanding of the hedge fund industry by
finding that hedge funds can indeed deliver persistent superior performance, even
under challenging economic conditions, albeit at a much more modest level and for a
shorter time interval than previously reported by authors such as Ammann et al.
(2013) and Boyson (2008). Importantly, this paper finds that new measures of
performance provide a new perspective on the performance of the hedge fund
industry. Specifically, we find that the EMPPM is a much more exacting performance
measure than the more traditional Sharpe ratio, alpha, and information ratio and less
often associated with high doubt ratios. Specifically, we find that top quintile funds as
ranked by the Sharpe and information ratios have larger doubt ratios than the
corresponding mediocre third quintile portfolios and this excess doubt persists for two
years. In contrast, there is much less evidence of manipulation by top portfolios as
ranked by alpha and the EMPPM. Specifically, there is much less excess doubt and
persistence of doubt for portfolios formed on alphas and excess and persistent doubt is
virtually non-existent for top portfolios as ranked by the EMPPM. When
benchmarked off the doubt ratios of the EMPPM-formed top quintile portfolios, the
portfolios formed on the Sharpe and information ratios have significantly higher doubt
ratios, whereas those formed on alpha often have significantly lower doubt ratios.
Interestingly, when benchmarked off the EMPPM, performance persistence is greatly
7
reduced, especially for Sharpe-formed portfolios, but, for information ratio-formed
portfolios, performance continues to persist for some strategies. This suggests that the
information ratio is still able to detect skill that the EMPPM is unable to find,
especially during the more turbulent post-2006 period.
Additionally, use of different performance measures results in different
conclusions. For example, while the Sharpe and information ratios typically indicate
that hedge funds performed better under the robust economic conditions that prevailed
in the first half of our sample period, the alpha and the EMPPM often find just the
opposite, suggesting that surviving hedge funds performed better during the liquidity
and credit crunch conditions that prevailed in the second half of our sample period. As
another example, while top quintile funds stratified by strategy and formed on the
Sharpe ratio, alpha, and information ratio typically persist in delivering superior
performance for two years over the entire sample period and all sub-periods, top funds
formed on the EMPPM typically deliver persistent superior performance for only 12
months during the challenging economic conditions that prevailed in the second half
of our sample period.
We review the literature in the next section and then describe the data and the
performance measures in Section 2. The empirical analysis then proceeds in four
phases. In Section 3, we measure the performance of top quintile hedge fund
portfolios formed on five different performance measures. We test whether
performance was stronger in either of the two sub-periods, which were characterized
by very different economic environments. In Section 4, we examine the persistence of
top quintile performance by testing whether the fifth (highest) quintile funds exhibit
statistically significant superior performance over mediocre third quintile funds 24
months after formation. We also check to determine the degree of persistence three
8
months, six months, 12 months, and 18 months after the portfolio formation date. In
Section 5, we examine excess doubt (differences between the fifth and third quintiles)
and persistence in excess doubt for hedge portfolios formed on the Sharpe ratio, alpha,
information ratio, and EMPPM. Section 6 conducts an experiment to determine if the
differences in performance persistence found by different performance measures are
due to skill, manipulation, or luck. Section 7 summarizes and concludes the paper.
2. Literature review
Hedge fund performance and performance persistence continue to be an active
topic in the finance literature. Although Ackermann et al. (1999) find evidence that
hedge funds outperform mutual funds, they do not find any evidence that hedge funds
outperform market indices. While Brown et al. (1999) find that Sharpe ratios and
Jensen’s alphas of hedge fund portfolios are higher than the Standard & Poor’s (S&P)
500, they are concerned that survival bias could have exaggerated the performance of
hedge funds. Stultz (2007) concludes that, at the very least, hedge funds offer returns
commensurate with risk once hedge fund manager compensation is accounted for.
More recently, Chincarini (2014) compares the effect of investment styles, finding
that funds that follow a quantitative investment style have higher alphas than funds
that follow a qualitative investment style.
Some research strongly supports persistence while other research is more
equivocal. Ammann et al. (2013) and Boyson (2008) find that top-performing hedge
funds formed on Fung and Hsieh’s (2004) alphas continue to provide statistically
significant performance three and two years later, respectively. Ammann et al. find
that strategy distinctiveness as suggested by Wang and Zheng (2013) is the strongest
predictor of performance persistence, while Boyson finds that persistence is
9
particularly strong among small and relatively young funds with a track record of
delivering alpha. Fung et al. (2008) find that funds of funds with a statistically
significant alpha are more likely to continue to deliver positive alpha.
More critically, Kosowski et al. (2007) find evidence that top funds deliver
statistically significant out-of-sample performance when sorted by the information
ratio, but not when sorted by Fung and Hsieh’s (2004) alphas. Capocci et al. (2005)
find that only funds with prior middling alpha performance continue to deliver
middling alphas in both bull and bear markets. In contrast, past top deliverers of
alphas continue to deliver positive alphas only during bullish market conditions. More
recently, Brandon and Wang (2013) find that the superior performance of equity-type
hedge funds largely disappears once liquidity is accounted for and Slavutskaya (2013)
finds that only alpha-sorted bottom-performing funds persist in producing lower
returns in the out-of-sample period. Finally, Hentati-Kaffel and de Peretti (2015) find
that nearly 80% of all hedge fund returns are random, with evidence of performance
persistence concentrated in hedge funds that follow event-driven and relative value
strategies.
Another strand of the hedge fund literature heavily criticizes the use of
common performance measures such as the Sharpe ratio, alpha, and information ratio.
For example, Amin and Kat (2003) question the use of these measures, since they
assume normally distributed returns and/or linear relations with market risk factors.
This strand of research inspired proposals for a wide variety of alternative measures of
performance purporting to resolve issues of measuring performance in the face of
non-normal returns. However, Eling and Shoemaker (2007) find that the ranking of
hedge funds by the Sharpe ratio is virtually identical to that for 12 alternative
performance measures. Goetzmann et al. (2007) point out that common performance
10
measures such as the Sharpe ratio, alpha, and information ratio can be subject to
manipulation, deliberate or otherwise. These issues imply that the use of these
performance measures can lead to misleading conclusions.
3. Data and performance measures
We use the Credit Suisse/Tremont Advisory Shareholder Services (TASS)
database. We select all US dollar funds that have three years of historical performance
to form monthly portfolios commencing on January 31, 2001, and continue to form
monthly portfolios until December 31, 2012, 144 months in all. We choose these data
so that we have three years of data, commencing in January 1998, with which to
estimate historical alphas to later examine 12 years of performance, six years each
pivoting around December 31, 2006. When we examine the number of observations in
the TASS database, we note the exponential growth of the data that seems to have
moderated from January 1998 onward, since from that date, the total number of fund–
year observations grew from 20,000, peaking at 50,000 in 2007 and falling to
approximately 29,000 in 2012.1 By commencing our study in January 1998, we avoid
a possible growth trend in the data.
We collect all monthly holding period returns net of fees from the TASS
database. We adjust for survivorship bias by including all funds, both live and dead.
We also adjust for backfill bias by including data on a given fund only from the date
that the fund was listed in TASS. We estimate the Sharpe ratio as the monthly holding
period return of the hedge fund less the one month T-bill return divided by the
standard deviation of excess returns calculated over the previous two years.
1
The details of the annual fund month observations by strategy are available from the corresponding
author upon request.
11
We combine the hedge fund data with Fung and Hsieh’s (2004) seven factors
along with an eighth emerging market factor, as described in David Hsieh’s data
library. Specifically, the regression model we use to estimate alpha for each hedge
fund i, αi, and the information ratio is
𝐻𝐹𝑅𝑡𝑖 = 𝛼 𝑖 + 𝛽1𝑖 (𝑆&𝑃)𝑡 + 𝛽2𝑖 (𝑆 − 𝐿)𝑡 + 𝛽3𝑖 (10𝑌)𝑡 + 𝛽4𝑖 (𝐶𝑟𝑒𝑑𝑆𝑝𝑟)𝑡 + 𝛽5𝑖 (𝐵𝑑𝑂𝑝𝑡)𝑡
+ 𝛽6𝑖 (𝐹𝑥𝑂𝑝𝑡)𝑡 + 𝛽7𝑖 (𝐶𝑜𝑚𝑂𝑝𝑡)𝑡 + 𝛽8𝑖 (𝐸𝑚𝑒𝑟𝑔𝑒)𝑡 + 𝜀𝑡
(1)
where HFRti is hedge fund i’s monthly return net of fees less the one-month T-bill rate
for date t, S&P is the S&P 500 index total monthly return (equity market factor), S - L
is the Russell 2000 index total return less the S&P 500 index total return (size factor),
10Y is the monthly change in the constant maturity 10-year Treasury yield (bond
factor), CredSpr is the monthly change in the difference between Moody’s Baa yield
less the constant maturity 10-year Treasury yield (credit risk factor), and BdOpt,
FxOpt, and ComOpt are the bond, currency, and commodity trend factors estimated
by bond, currency, and commodity look-back straddles, respectively, and available
from Hsieh’s data library. The final factor, Emerge, is the total monthly return on the
MSCI index (emerging market factor). We run approximately 360,000 constant size
36-month rolling regressions of equation (1) to collect historical alphas for all funds.
Alpha () is the intercept of equation (1) and represents the excess return above that
justified by systematic risk. Following Kosowski et al. (2007), the information ratio is
defined as the t-statistic of alpha in equation (1).
We calculate the MPPM () of Goetzmann et al. (2007) as

1
1 T

ln   (1  rt ) /(1  rft ) (1A )
 (1  A)t  T t 1



(2)
12
where A is the risk aversion parameter, rt is the monthly holding period return of the
hedge fund, rft is the one-month T-bill return, T is the number of observations, and t
is one month. The measure  represents the certainty equivalent excess (over the riskfree rate) monthly return for an investor with risk aversion A employing a utility
function similar to the power utility function. This implies that  is relevant for riskadverse investors with constant relative risk aversion. We estimate  over the
previous two years using a risk aversion parameter A of three.2 We next calculate the
EMPPM, which is the monthly  of the hedge fund less the corresponding  for the
Russell 2000 index.
The doubt ratio is calculated as follows:
𝐷𝑅 = 2 +
 (2)
 (2) −  (3)
Note that (2) is the MPPM as calculated from equation (2), where the risk aversion
parameter A is two and (3) is the MPPM when the risk aversion parameter is three.
This ratio employs a linear interpolation of the degree of risk aversion for two
investors with different degrees of risk aversion. While the change in the degree of
risk aversion is concave, Brown et al. (2010) suggest that a linear interpolation is a
good approximation as long as the degree of risk aversion is not very high. Doubt
2
Initially, we experimented with a variety of risk aversion parameters, from one through 10, finding
that the rankings by  did not change very much. Goetzmann et al. (2007) suggest that the market
believes risk aversion varies between two and four. Since the doubt ratio of Brown et al. (2010) is
calibrated using a ratio of ’s calculated using risk aversion parameters of two and then three, we
chose a value of three so that later we can calculate the  using a value of two. This enables us to
compute a doubt ratio comparable to that of Goetzmann et al. so that we can use their benchmark of a
doubt ratio of 150 as the pivot point for suspicious hedge fund returns.
13
ratios become very large when (2) ≈ (3). This happens when the returns to
investors are always positive with very little variation. This is suspicious and merits
further investigation.
Table 1 reports that these collection procedures capture 176,487 fund–month
observations, approximately one-half from dead funds.3 Monthly returns have the
typical time series characteristics of excessive kurtosis and positive skewness, as do
the information, alpha, and doubt ratios. The Sharpe ratio and the EMPPM both have
excessive kurtosis and left skewness that is traced mostly to dead funds. We observe
that the mean of the doubt ratio is always far larger than the median, clearly indicating
that examining the mean alone as a measure of location can be misleading, a point
that we return to later.
<<Table 1 about here>>
The results of our data collection are broken down by strategy in Table 2. A
striking fact is the huge attrition rate of hedge funds, less than one-half of all of the
hedge funds included in our data survive until the end of our sample period. Live
funds are larger, have a longer history, and perform better than dead funds. The
average fund remains in our data for nearly six years, longer than the four years or
five years reported by Slavutskaya (2013) but in line with Boyson (2008).
<<Table 2 about here>>
Similar to Capocci et al. (2005) and Sadka (2010), the most popular hedge
fund strategy is the fund of funds closely followed by a long–short equity hedge.
Overall, hedge funds have a 37 basis point (4.5%) monthly (annual) average rate of
3
Our understanding is that there are no Madoff funds in the December 31, 2012, version of the TASS
database that we use. We are aware that six feeder funds heavily invested in the Madoff Funds,
specifically Fairfield Sentry Ltd (FFS), Kingate Global Fund Ltd (KING), Optimal Strategic US Equity
Ltd (OPTI), Santa Clara I Fund (SANTA), LuxAlpha Sicav-American Selection (LUX), and Herald
Fund SPC-USA Segregated Portfolio One (HRLD). We checked the live and dead TASS databases by
fund name and close variations of these names and found that none of them was in the database. We
concluded that the Madoff incident did not directly affect our results.
14
return. It is notable that, in the second half of our data, the average return is 19 basis
points (2.3%) monthly (annually), whereas in the first half it is 74 basis points
(9.25%) monthly (annually). Dichev and Yu (2011) also document a sharp reduction
in buy and hold returns for a very large sample of hedge funds and commodity trading
advisors, from, on average, 18.7% from 1980 to 1994 to 9.5% from 1995 to 2008,
while Ibbotson et al. (2011) report an arithmetic mean of 7.86% from 1995 to 2009.
Clearly, our data document a continuing decline in the average rate of return of hedge
funds, as noted by Dichev and Yu (2011).
Except for the dedicated short bias strategy, all funds have a positive monthly
rate of return, on average. Evidently, the EMPPM is more critical than the alpha
measure, since three strategies fail to generate excess returns according to this
measure, whereas the average alphas are negative only for the fund of funds strategy.
Brown et al. (2010) suggest that a doubt ratio of 150 or more is evidence of suspicious
hedge fund returns that should elicit further investigation. By this measure, live and
dead funds trigger suspicion. Moreover, the values of the doubt ratio differ from those
of Brown et al. (2010): Some are very high and others are negative. Most of these
differences can be traced to our use of dead funds, which we include to avoid
survivorship bias, and our longer sample period, which includes three years when the
average return for hedge funds was negative.4
Several hedge fund strategies look suspicious and the most suspicious of all is
the fixed income arbitrage strategy. However, as noted above, the median of the doubt
ratio is typically far smaller than the mean; therefore, to avoid jumping to conclusions
regarding the amount of suspicion we cast on strategies with high doubt ratios, we
include the percentage of doubtful funds for a given strategy in the last column of
4
The details of the reconciliation of our data to those of Brown et al. (2010) are available from the
corresponding author upon request.
15
Table 2.5 We note that most of the strategies that are highly suspicious have only a
modest number of doubtful funds.6 Later, when we examine the empirical doubt
ratios, we need to remind ourselves that high means for the doubt ratios are typically
generated by a modest number of funds.
Interestingly, overall hedge fund performance is much less suspicious in the
second half of the sample period; we therefore investigate the time series
characteristics of our data in Table 3. Generally, hedge funds grow in size and age
throughout the sample period. The hedge fund industry appears to be accident prone,
posting negative returns in 2001, 2008, and 2011. Interestingly, the doubt ratio shows
that hedge fund performance is noticeably less doubtful after negative return years,
hinting that years of poor return also purge suspicious funds.
<<Table 3 about here>>
For robustness, we later split the data into two, using December 2006 as the
pivot date. The rationale is that the earlier sub-period will reflect healthy economic
conditions whereas the second half, containing the credit crunch of 2007 and the
severe recession of 2008 followed by continuing slow growth thereafter, would reflect
more challenging economic conditions. We note that the first half of the sample
period contains some years of poor return, so to check on our conjecture regarding the
sub-periods, we recalculate the sub-period averages according to the six best and
worst return years. The last two rows of Table 2 report that the worst six years are
comparable to the post-2006 period. Specifically, the number of funds, average asset
size, doubt ratio, and percentage of doubtful funds are similar. While the average rate
5
A doubtful fund is defined as one that has 50% of its doubt ratio observations above 150.
The exceptions are fixed income strategies, option strategies, and other, where high average doubt
ratios are paired with a high proportion of doubtful hedge funds. However, there are only a modest
number of funds for each strategy and, therefore, when we calculate the overall doubt ratio for the
entire sample, the mean is still unduly influenced by a modest number of highly doubtful funds.
6
16
of return, Sharpe ratio, and EMPPM are lower, the alpha, and information ratio are
higher in the worst six years compared to the post-2006 period.
Our testing strategy is to construct the top quintile portfolios formed on the
above five measures of performance and compare them to the performance of
similarly formed mediocre third quintile portfolios. These quintile portfolios are
formed out of sample based on the prior month’s performance and each portfolio is
held for 24 months.7 Given our 12-year sample period, from January 31, 2001, to
December 31, 2012, we form 120 overlapping out-of-sample portfolios for each
quintile. The portfolios are equally weighted. Individual funds included in the
formation portfolio that subsequently disappeared during the 24-month valuation
period are assumed to have been reinvested in the remaining funds. It is evident from
Table 3 that the number of hedge funds varies year by year, which is problematic
when we decompose our sample by strategy and by quintile. Therefore, we report our
results for emerging markets, event-driven, fund of funds, long–short equity hedge,
managed futures, and multi-strategy strategies and drop the remaining strategies, since
the t-statistics are unreliable due to missing information in specific months and
quintiles. According to Table 2, most of the retained strategies have modest average
doubt ratios and contain a small number of doubtful funds. However, the managed
futures strategy is an exception, where the average doubt ratio is dominated by a few
managed futures funds that have an extremely high doubt ratio, a fact that we must
bear in mind when interpreting the empirical results of the doubt ratio for the
managed futures strategy.
Note that the prior month’s performance is measured as an average over a period according to the
performance measure at hand. Specifically, the prior month’s alpha are the average alpha over the prior
36 months, the Sharpe ratio over the last 24 months, and the EMPPM and doubt ratio over the last 12
months.
7
17
We use bootstrapped t-tests of differences in the means of the performance
measures to construct one- and two-tailed tests. All t-tests are based on the stationary
bootstrap method of Politis and Romano (1994) with 1,000 iterations. This two-step
procedure resamples with a random block size from the first block bootstrap
procedure to form a stationary pseudo-time series. The original block bootstrap and
the resampled bootstrap select a row of observations for each selection of the
performance measure for all strategies so that any cross-sectional correlation in the
performance measure among the hedge fund strategies is preserved in the resamples.8
4. A first look: In-sample and out-of-sample performance
To determine the maximum performance, we artificially assume that each
month an investor forms and holds for one month an equally weighted portfolio of the
top quintile of funds as defined by a given performance measure. In total, there are
five top quintile funds, one each formed on the Sharpe ratio, information ratio, alpha,
EMPPM, and doubt ratio. This implies that investors can anticipate which funds will
achieve top quintile performance each month for 144 months. The monthly average
for each strategy is reported in Panel A of Table 4. Panel B reports the differences in
performance for the two sub-periods, namely, 2001–2006, and 2007–2012.
<<Table 4 about here>>
Clearly, Panel A of Table 4 shows that hedge fund investment can be
extremely lucrative. Throughout the 12-year period, we calculate an average alpha of
1.29% (16.6%) per month (year). Using Fung and Hsieh’s (2004) seven-factor model,
Boyson (2008) reports a somewhat larger monthly alpha of 1.86% for top quintile
8
The stationary bootstrap is used because the underlying data are monthly. With a sample size of 120,
the recommendation is to resample with each draw of size N 1/3, or five in our case. However, random
samples of five months of data can induce weak non-stationarity in the artificial time series compiled
by the conventional bootstrap procedure. Therefore, we employ the second bootstrap with random
sampling size from the first artificial time series to assure a stationary bootstrapped times series.
18
portfolios formed on alpha, possibly because the author’s data end in 2002 and
therefore does not include the recent challenging market conditions that characterize
the latter half of our sample. Dichev and Yu (2011) employ the same Fung–Hsieh
eight-factor model and find an average in-sample annual alpha of 2.6% for all hedge
funds, which is similar to our sample alpha of 21 basis points monthly (2.5%
annually) for all funds.9
The Sharpe and information ratios are also very high. Interestingly, the
EMPPM is much more modest, achieving a certainty equivalent of 23 basis points
(2.8%) per month (year) and the average doubt ratio for funds formed on the top
quintile of doubt ratios is 991, much higher than the threshold of 150 suggested by
Brown et al. (2010) for funds that start to look suspicious. According to traditional
performance measures, specifically the Sharpe ratio, alpha, and information ratio,
managed futures is by far the top-performing strategy and also the most suspicious
strategy according to the doubt ratio. However, as noted earlier, the average doubt
ratio for the managed futures strategy is dominated by a few suspicious funds.10
Panel B of Table 4 shows that the traditional performance measures are, on
average, significantly higher under the more robust economic conditions that
prevailed up to December 31, 2006, than in the subsequent challenging economic
climate. In contrast, the EMPPM is significantly higher in the more challenging
second half of the sample period. The traditional performance measures typically
9
For the sake of brevity, we report the in-sample top quintile performance statistics in Table 2 and not
the average in-sample statistics. These are available from the corresponding author upon request.
10
We attribute the characteristic high average doubt ratio overall partly to a few funds that typically
have very high doubt ratios and partly to the notion that the doubt ratio does not necessarily detect
misdeeds on behalf of fund managers but, rather, can report a pattern of returns that is suspicious and
requires further investigation. When we look at Table 2, we note a pattern to the doubt ratios. Strategies
such as convertible arbitrage, fixed income arbitrage, and managed futures have very high doubt ratios
on average: 868.21, 1321.50, and 841.68 respectively. A likely explanation is that these strategies
involve locking in an arbitrage premium and holding the illiquid asset until expiry of the position.
These strategies result in a large number of positive stable returns (and the occasional significant
drawdown) that can result in a large doubt ratio.
19
show that individual hedge fund strategies perform better in the first half, whereas the
EMPPM consistently shows that individual hedge fund strategies perform better in the
second half of the sample period.
While Table 4 shows the most optimistic in-sample performance, Table 5
reports a much more realistic view of hedge fund performance based on top quintile
out-of-sample portfolios evaluated two years after formation. For the traditional
performance measures, Panel A of Table 5 shows that the average out-of-sample
performance is more modest but still quite attractive. For example, portfolios formed
on alpha still return 54 basis points per month, on average, which is remarkably
similar to the 50 basis points reported by Boyson (2008) for similarly constructed
portfolios. In contrast, portfolios formed on EMPPM return a mere one basis point per
month in certainty equivalent returns above the Russell 2000 index. Finally, for
portfolios formed on the doubt ratio, the doubt ratio is well above the threshold of 150
for suspicious performance and comparable to its in-sample counterpart.
<<Table 5 about here>>
In contrast to the in-sample results, Panel B of Table 5 shows that portfolios
formed on the EMPPM performed better in the first half. However, when broken
down by individual strategies, we find that this overall result for the EMPPM-formed
portfolios is attributable to the emerging markets strategy, since a clear majority of the
remaining strategies perform significantly better in the second half. Additionally,
alpha-formed portfolios now perform better in the second half. The doubt ratio still
suggests that the top quintile of doubtful funds is significantly more suspicious in the
first half of the sample period.
In summary, this section finds that top quintile hedge funds perform very well
based on the traditional performance measures, namely, the Sharpe ratio, alpha, and
20
information ratio, both in and out of sample 24 months later. The EMPPM is more
exacting, since it finds that the in-sample performance and particularly the out-ofsample performance of top quintile hedge funds are more modest. Statistical tests
suggest that top hedge fund portfolios formed on the Sharpe and information ratios
perform better, on average, under the robust economic conditions that prevailed in the
first half of the sample period. However, alpha- and EMPPM-formed portfolios
perform significantly better and the doubt ratio suggests that hedge fund performance
is significantly less suspicious under the more challenging conditions that prevailed in
the second half of the sample period.
5. Performance persistence
We measure performance persistence by comparing the out-of-sample
performance of portfolios formed on the fifth and third quintiles according to a given
performance measure. We compare the fifth and third quintile portfolios rather than
the fifth and first quintiles, as Boyson (2008), because it is clear that, once formed, the
hedge funds have an attrition rate concentrated in the first quintile.11 To illustrate, we
average and then plot the out-of-sample EMPPM performance for the 120 portfolios
for each of the 24 out-of-sample months.
<<Figures 1 and 2 about here>>
Figure 1 shows that the out-of-sample portfolios formed on the fifth and third
quintiles have very high (23 basis points) and high (six basis points) average excess
(over the Russell 2000 index) certainty equivalent monthly returns, respectively, one
month after formation. This strong performance then gradually falls such that, 24
months after formation, the fifth and third quintile portfolios now return only one and
11
This could be an artifact of our sample period that covers the severe liquidity crisis and subsequent
recessionary and slow growth conditions after 2006.
21
-1 basis point excess certainty equivalent returns, respectively. In contrast, the first
quintile presents a 20 basis point loss initially, but these losses diminish over time to
the extent that, 16 months after the first quintile portfolios are formed, the EMPPM is
reduced to zero and, after 24 months, the first quintile is outperforming the third
quintile. Figure 2 shows a higher attrition rate in the first quintile as more funds
disappear from the out-of-sample first quintile portfolios than for portfolios formed on
the fifth and third quintiles. Consequently, we suspect that the performance of the first
quintile is tainted by a survivorship bias and does not form a valid benchmark for
determining performance persistence.
Therefore, to examine persistence, we compute the difference between the
fifth and third quintile portfolio’s out-of-sample performance 24 months after
formation to determine if the top quintile performance portfolio still delivers superior
performance two years after formation. We conduct a one-tailed bootstrap t-test to
determine if the mean of the fifth quintile portfolio’s performance is statistically
superior to that of the mediocre third quintile portfolio.
Table 6 reports the results of this persistence test. Panel A reports the overall
difference between the out-of-sample fifth and third quintiles by performance measure
and by hedge fund strategy for 120 portfolios formed between January 31, 2001, and
December 31, 2010, and held for two years. Panels B and C report the same
information for the first and second halves of the sample period, respectively,
representing a time series of 60 portfolios each held for 24 months.
<<Table 6 about here>>
Clearly, Panel A of Table 6 shows that, on average, top hedge fund
performance persists up to two years after they are selected. This conclusion is
statistically significant for all measures of performance. Ammann et al. (2013) find
22
three-year performance persistence for top decile hedge portfolios while Boyson
(2008) tests for and finds two-year persistence for top quintile portfolios formed on
Fung and Hsieh’s (2004) alpha. In addition, Capocci et al. (2010) test for and find
alpha persistence for one year throughout their sample period, which includes both
bull and bear markets. Regarding the individual strategies, the alpha and especially
the information ratio provide consistent evidence of performance persistence.
Meanwhile, the Sharpe ratio and especially the EMPPM provide a more critical
assessment of performance persistence, since several individual strategies do not have
statistically significant persistence in performance according to these performance
measures.
The doubt ratio casts a shadow on performance persistence. The most
suspicious funds, which are formed on the fifth quintile of the doubt ratio, have
significant excess doubt above similarly constructed mediocre third quintile
suspicious funds two years later. The large difference in the doubt ratio between the
fifth and third quintile funds suggests that top quintile doubtful funds are more
suspicious because the difference in the doubt ratio is greater than the threshold of
150 suggested by Brown et al. (2010). Looking at the individual strategies, we again
see that several strategies have, to coin a phrase, persistent doubt and, once again, no
strategy has more persistently suspicious returns than the managed futures strategy.
Examining the sub-period results in Panels B and C of Table 6, we see that, on
average, for almost all performance measures, top-performing funds continue to
outperform third quintile funds two years later in both the first and second halves of
the sample period. The exception is the EMPPM, which does not detect any
statistically significant persistence during the challenging economic conditions of the
second half of the sample period. Again, however, the doubt ratio casts a shadow on
23
these results, since it shows evidence of persistence in suspicious returns. The
difference between top and third quintile funds formed on the doubt ratio is far above
the threshold of 150 for suspicious performance in both sub-periods.
Looking at the results by strategy, we see that top funds formed on the Sharpe
ratio, alpha, and information ratio continue to persist in delivering top performance
two years later in both sub-periods. Similarly, doubts persist overall and in both
periods for most individual strategies. However, with the exception of the managed
futures strategy, top quintile portfolios formed on the EMPPM show performance
persistence in the first period alone.
The latter finding leads us to examine for how long performance persists
according to the EMPPM. Table 7 reports the difference in performance between the
fifth and third quintiles of portfolios formed on the EMPPM three months, six
months, 12 months, 18 months, and 24 months after formation. Panel A reports
performance persistence by time horizon for the overall sample period and Panels B
and C report the same information for the two sub-sample periods. With very few
exceptions, performance persists for six months in the overall sample and in both subperiods. After 12 months, however, many individual strategies no longer show
evidence of persistence in performance according to the EMPPM. This decrease in
performance persistence is particularly evident in the second half of the sample. After
18 months, there is no evidence of statistically significant persistence in the second
sub-period. Evidently, portfolios formed on the EMPPM show evidence of
performance persistence after 12 months because of the persistence that prevailed in
the first half of the sample period.
<<Table 7 about here>>
24
In summary, we find that there is performance persistence in hedge funds.
Performance persistence is strongest for the traditional measures of performance and
this persistence continues for up to two years. Doubt also persists, in that strategies
that are doubtful to begin with continue to be doubtful two years later. Interestingly,
the EMPPM is more critical and unable to detect persistence in performance after 12
months for strategies formed under the more challenging economic conditions that
prevailed in the second half of the sample period, specifically for portfolios formed
between January 31, 2006, and December 31, 2010.
6. Doubtful performance strategies?
We now use the doubt ratio to see if selecting top quintile portfolios based on
the previous month’s Sharpe ratio, alpha, and information ratio generates suspicious
out-of-sample performance 24 months after portfolio formation. As a control
mechanism, we also use the doubt ratio to analyze top quintile funds as ranked by the
EMPPM.
Table 8 is similar to Table 5 except that we employ the doubt ratio to evaluate
the portfolios formed on the other four performance measures. Panel A of Table 8
reports the doubt ratios of the out-of-sample top quintile funds by strategy for the
overall period and Panel B reports the differences in the doubt ratio during the two
sub-periods.
<<Table 8 about here>>
Panel A, Table 8, reports that portfolios formed on top quintile Sharpe ratios,
information ratios, and EMPPMs 24 months earlier also have doubt ratios that are
above than the threshold of 150 for suspicious performance. Looking at the doubt
ratio by strategy, we find that about half of the portfolios formed on the Sharpe and
25
information ratios have a doubt ratio above 150, whereas only two portfolios formed
on the alpha and EMPPM performance ratios have doubt ratios above 150.
Interestingly, there is a consensus that top-performing event-driven and managed
futures strategies have suspicious performance no matter which performance measure
is used to form the portfolio. Although the doubt ratios of these strategies can be
dominated by a few suspicious funds, this situation still hints that these strategies
involve some degree of hedging, income smoothing, or both.
Using the doubt ratios of EMPPM-formed portfolios as the benchmark, it is
clear that, except for the long–short equity hedge strategy, top quintile portfolios
formed on the EMPPM have a statistically significant lower out-of-sample doubt ratio
two years later than top quintile portfolios formed on the Sharpe and information
ratios do. Interestingly, alpha-formed portfolios sometimes have a significantly lower
doubt ratio than EMPPM-formed portfolios.12
Panel B of Table 8 reports that top-performing funds, as selected by all but the
alpha measure of performance, have more suspicious performance under the more
robust economic conditions that prevailed in the first half of the sample period.
Moreover, for individual strategies, the majority of statistically significant differences
in the doubt ratio are positive, indicating that performance was more suspicious in the
first half of the sample period.
We now examine whether the doubt ratio of top funds persists. Table 9 is
similar to Table 6, except that we now employ the doubt ratio to evaluate the
12
It is clear that top quintile portfolios formed on Sharpe ratios, alphas, and information ratios have
performance persistence (0.19, 0.43, and 1.93, respectively) overall in Table 6, but different doubt
ratios (823.48, 108.76, and 820.98, respectively) overall in Table 8, suggesting that the performance
persistence of alpha-formed strategies is more reliable. However, comparing these two tables can be
misleading, since the mean of the doubt ratio can be heavily influenced by a few funds with a high
doubt ratio. For example, the lower doubt ratio for alpha-formed funds can be traced to the smaller but
still very large doubt ratio for the managed futures strategy—444.05—when compared to the
corresponding doubt ratios of this strategy for Sharpe and information ratio-formed strategies—
18,703.92 and 14,853.42, respectively—thereby accounting for the difference in the average doubt
ratios for the alpha-formed funds.
26
portfolios. We measure persistence in the doubt ratio by comparing the out-of-sample
doubt ratio of portfolios formed on the fifth and third quintiles according to the
Sharpe ratio, alpha, information ratio, and EMPPM 24 months after portfolio
formation. Funds have excess doubt when the doubt ratios of top-performing funds
have a greater doubt ratio than mediocrely performing funds. Panel A reports the
overall difference in the doubt ratios between the fifth and third quintiles by
performance measure and by hedge fund strategy for the entire sample period and
Panels B and C show the same information for the first and second halves of the
sample period, respectively.
<<Table 9 about here>>
Panel A of Table 9 reports that all Sharpe- and information ratio-formed
portfolios exhibit excess doubt ratios and this extra doubt persists two years after the
portfolio has been formed. In contrast, individual top-performing strategies formed on
the alpha and EMPPM show much less excess doubt that persists more rarely.
Comparing the alpha- and EMPPM-formed strategies, it is clear that excess doubt and
its persistence are lower for EMPPM-formed portfolios. Panels B and C clearly show
that the overall period results carry forward to the first and second sub-periods.
Specifically, top funds formed on the Sharpe and information ratios usually have an
excess doubt ratio that persists two years after portfolio formation while topperforming funds formed on the alpha and especially the EMPPM have much less
excess and persistent doubt for both sub-periods.
Panels B and C of Table 9 show a drop in excess doubt and Table 6 shows a
drop in performance persistence during the second sub-period, particularly for the
Sharpe, information, and doubt ratios. The interesting question is whether the drop in
suspicious hedge fund returns and the performance persistence were due to the more
27
challenging economic conditions prevailing after 2006 or to a purge of suspicious
funds. Evidently, these findings are due to the challenging economic environment
rather than a purging of funds, since Table 2 reports that the average doubt ratio and
the percentage of funds that are suspicious drop by more than half after 2006.
Meanwhile, the doubt ratios and percentages of funds that are suspicious are virtually
the same for live and dead funds. When we eliminate all suspicious funds, defined as
funds for which more than 50% of the observations have a doubt ratio greater than
150, and recompute Table 2, we find that performance, especially as measured by the
Sharpe, information, and doubt ratios, declines substantially, suggesting that
portfolios formed on these measures might have been unduly influenced by doubtful
funds. Moreover, when we purge these doubtful funds from our data, the reductions in
the Sharpe, information, and doubt ratios are about the same for both sub-periods,
suggesting that the reduction in performance persistence is related to economic
conditions rather than to a few doubtful funds.13
In summary, we find evidence that top-performing hedge fund portfolios tend
to have high doubt ratios, regardless of the performance measure used to form top
quintile portfolios. However, top portfolios formed on the EMPPM performance
measure also have less suspicious performance than top portfolios formed on the
Sharpe and information ratio performance measures. It is clear that top portfolios have
more doubt under the more robust economic conditions that prevailed in the first half
of the sample period. When looking at the persistence of doubt, we see that topperforming funds formed on the Sharpe and information ratios usually have excess
doubt that persists over mediocrely performing funds. This finding suggests that,
when selecting funds according to ranking by the Sharpe and information ratio,
13
The results of this exercise are available from the corresponding author upon request.
28
investors are also selecting funds that have suspicious returns. In contrast, portfolios
formed on the alpha and the EMPPM has much less excess doubt that more rarely
persists.
7. Skill, manipulation, or luck?
Another issue is whether the weaker performance persistence of the EMPPM
and stronger performance persistence of the Sharpe ratio, alpha, and information
ratios as reported in Table 6 are due to differences in the degree of overlapping data or
to differences in skill, manipulation, or mere luck.14 To disentangle these possibilities,
we perform the following experiment. We form out-of-sample portfolios by
performance quintile according to the Sharpe ratio, alpha, and information ratios and
EMPPM in the usual way except that we now measure out-of-sample performance for
all of these portfolios two years later according to the EMPPM. In other words, we
recompute Table 6 but evaluate performance according to the EMPPM. Since all
portfolios are assessed using the same performance measure, there is no difference in
the degree of overlap. While the EMPPM is resistant to manipulation, it may not be
able to detect skill. Therefore, comparing the persistence of the EMPPM-formed
strategy with the EMMPM persistence of the portfolios formed on the other measures
of performance will shed light as to whether persistence is due to skill or to something
else, such as luck or manipulation.
<<Table 10 about here>>
Table 10 shows that the Sharpe-formed portfolios show little evidence of
performance persistence, suggesting that the persistence reported in Table 6 is due to
luck or manipulation. Meanwhile, the alpha- and information ratio-formed portfolios
14
Differences in the degree of overlap in forming and evaluating our out-of-sample portfolios occur
because while we use only 12 monthly observations to measure the EMPPM, 24 months to measure the
Sharpe ratio, and 36 months of data to estimate the alpha and information ratios.
29
show much less evidence of performance persistence, suggesting that the persistence
reported in Table 6 is partly due to luck or manipulation. However, the information
ratio-formed portfolios also show clear evidence of skill, especially in the more
challenging second sub-period. For instance, the long–short equity hedge strategy for
information ratio-formed portfolios shows statistically significant performance
persistence two years after portfolio formation This turbulent period contains market
conditions under which one can expect the long–short strategy to be particularly
effective for those with skill and this skill is detected by the information ratio but not
by the EMPPM. We conclude that the lower performance persistence of the EMPPM
is partly due to its limited ability to detect skill and partly because it appears to adjust
for luck or manipulation.
8. Summary and conclusions
In response to the questions raised in the introduction, we find that, indeed, by
all measures of performance, top quintile funds are able to deliver persistent superior
performance out of sample. However, the magnitude of this performance is more
modest than reported in the literature by others. Moreover, the EMPPM results
indicate even more modest and less persistent superior performance. The EMPPM
results show persistence of no longer than 12 months during the more recent
challenging economic conditions rather than the two years of persistence observed for
portfolios formed on traditional measures of performance. Conclusions vary
depending on which performance measure is used to form the top quintile portfolios.
For example, top quintile funds formed on the Sharpe ratio and information ratio
perform better, on average, under the robust economic conditions that prevailed in the
first half of the sample period, while most top quintile portfolios formed on alpha and
30
the EMPPM performed better in the more exacting recessionary and slow growth
conditions that prevailed in the second half of the sample period. Moreover, when we
assess performance persistence based on the EMPPM, we find that at least some
portion of the performance persistence appears to be due to luck or manipulation but
the information ratio still seems able to detect skill, especially in the more turbulent
post-2006 period.
It is clear that top-performing funds tend to have high doubt ratios, particularly
during the first half of the sample period, regardless of the performance measure
employed to select the top quintile funds. Still, top-performing portfolios formed on
the EMPPM have significantly lower doubt ratios than top-performing funds formed
on the Sharpe and information ratios. However, when looking at the persistence of
doubt, we see that top-performing funds formed on the Sharpe and information ratios
generally exhibit excess doubt over mediocrely performing funds that persists over
time. This finding suggests that, when selecting funds according to ranking by Sharpe
and information ratios, investors are also selecting funds that have persistently
suspicious returns. In contrast, portfolios formed on alpha and especially the EMPPM
have much less excess doubt that more rarely persists.
31
References
Ackermann, C., McEnally, C. and Ravenscraft, D., ‘The performance of hedge funds:
Risk, return, and incentives’, Journal of Finance, Vol. 54(3), 1999, pp. 833–874.
Ammann, M., Huber, O. and Schmit, M., ‘Hedge fund characteristics and
performance persistence’, European Financial Management, Vol. 19(2), 2013, pp.
209–250.
Amin, G. and Kat, H. M., ‘Hedge fund performance 1990-2000: Do the “money
machines” really add value?’, Journal of Financial & Quantitative Analysis, Vol.
38(2), 2003, pp. 251–274.
Boyson, N. M., ‘Hedge fund performance persistence: A new approach’, Financial
Analysts Journal, Vol. 64(6), 2008, pp. 27–44.
Brandon, R. and Wang, S., ‘Liquidity risk, return predictability, and hedge fund’s
performance: An empirical study’, Journal of Financial & Quantitative Analysis, Vol.
48(1), 2013, pp. 219–244.
Brown, S., Goetzmann, W. and Ibottson, R., ‘Offshore hedge funds: Survival and
performance, 1989–95’, Journal of Business, Vol. 72(1), 1999, pp. 91–117.
Brown, S., Kang, M. S., In, F. and Lee, G., ‘Resisting the manipulation of
performance metrics: An empirical analysis of the manipulation-proof performance
measure’, Working paper, New York University, 2010.
Capocci, D., Corhay, A. and Hubner, G., ‘Hedge fund performance and persistence in
bull and bear markets’, European Journal of Finance, Vol. 11(5), 2005, pp. 361–392.
Chincarini L., ‘The impact of quantitative methods on hedge fund performance’
European Financial Management, Vol. 20 (5), 2014, pp. 857–890.
Dichev, I. D. and Yu, G., ‘higher risk, lower returns: what hedge fund investors really
earn’, Journal of Financial Economics, Vol. 100(2), 2011, pp. 248–263.
Eling, M. and Schuhmacher, F., ‘Does the choice of performance measure influence the
evaluation of hedge funds?’, Journal of Banking & Finance, Vol. 31(9), 2007, pp. 2632–
2647.
Fung, W. and Hsieh, D. A., ‘Hedge fund benchmarks: A risk-based approach’,
Financial Analysts Journal, Vol. 60(5), 2004, pp. 65–80.
Fung, W., Hsieh, D. A., Naik, N. and Ramadorai, T., ‘Hedge funds: Performance, risk
and capital formation’, Journal of Finance, Vol. 63, 2008, pp. 1777–1803.
Goetzmann, W., Ingersoll, J. E., Spiegel, M. I. and Welch, I., ‘Portfolio performance
manipulation and manipulation-proof performance measures’, Review of Financial
Studies, Vol. 20(5), 2007, pp. 1503–1546.
32
Hentati-Kaffel, R. and de Peretti, P., ‘Detecting performance persistence of hedge
funds’, Economic Modelling, Vol. 47, 2015, pp. 185–192.
Ibbotson, R. G., Chen, P. and Zhu, K., ‘The ABCs of Hedge Funds’, Financial
Analysts Journal, Vol. 67(1), 2011, pp. 15–25.
Kosowski, R., Naik, N. and Teo, M., ‘Do hedge funds deliver alpha? A Bayesian and
bootstrap analysis’, Journal of Financial Economics, Vol. 84(1), 2007, pp. 229–264.
Politis, D. and Romano, J. P., ‘The stationary bootstrap’, Journal of the American
Statistical Association, Vol. 89(428), 1994, pp. 1303–1313.
Sadka, R., ‘Liquidity risk and the cross-section of hedge-fund returns’, Journal of
Financial Economics, Vol. 98 (1), 2010, pp. 54–71.
Slavutskaya, A., ‘Short-term hedge fund performance’, Journal of Banking and
Finance, Vol. 37, 2013, pp. 4404–4431.
Stultz, R., ‘Hedge funds: Past, present and future’, Journal of Economic Perspectives,
Vol. 21(2), 2007, pp. 175–194.
Wang, A. and Zheng, L., ‘The road less traveled: Strategy distinctiveness and hedge
fund performance’, Review of Financial Studies, Vol. 25(1), 2013, pp. 96–143.
33
Table 1: Sample Characteristics of the Performance Measures
This table reports the sample statistics of the performance of hedge funds from
January 31, 2001, until December 31, 2012. All returns are in percent; SR is the
Sharpe ratio, IR is the information ratio, alpha is estimated from an eight-factor
version of Fung and Hsieh’s (2004) model, EMPPM3 is the excess MPPM of
Goetzmann et al. (2007) with a risk aversion parameter of three, and DR is the doubt
ratio of Brown et al. (2010). The EMPPM reveals the monthly excess risk-adjusted
return above the corresponding Russell 2000 return.
Mean
Median
St. Dev
Kurtosis
Skewness
Minimum
Maximum
80th Percentile
60th Percentile
40th Percentile
N
Return
0.37
0.47
4.54
121.77
1.33
-99.99
272.86
2.30
0.92
0.02
176487
Mean
Median
Std. Dev
Kurtosis
Skewness
Minimum
Maximum
80th Percentile
40th Percentile
20th Percentile
N
Return
0.45
0.50
4.48
28.70
0.23
-93.11
108.39
2.45
0.02
-1.50
88063
Mean
Median
Std. Dev
Kurtosis
Skewness
Minimum
Maximum
80th Percentile
40th Percentile
20th Percentile
N
Return
0.30
0.44
4.59
206.02
2.35
-99.99
272.86
2.17
0.01
-1.53
88424
Panel A: Full Sample
SR
IR
Alpha
0.15
0.70
0.18
0.18
0.44
0.13
11.34
2.69
0.80
85337.71
198.15
15.46
-289.70
8.92
0.47
-3340.21
-22.10
-14.05
128.17
106.11
12.31
1.02
1.97
0.63
0.43
0.87
0.26
-0.04
0.03
0.01
176487
176487
176487
Panel B: Live
SR
IR
Alpha
0.20
0.87
0.27
0.21
0.63
0.19
1.33
2.78
0.77
248.39
298.19
6.41
8.16
12.02
0.67
-11.00
-22.10
-5.65
58.00
106.11
7.36
1.02
2.08
0.72
-0.01
0.24
0.07
-0.65
-0.62
-0.19
88063
88063
88063
Panel C: Dead
SR
IR
Alpha
0.09
0.54
0.09
0.15
0.23
0.06
15.96
2.58
0.83
43357.12
63.48
22.69
-207.21
5.13
0.38
-3340.21
-9.66
-14.05
128.17
54.49
12.31
1.02
1.82
0.54
-0.07
-0.17
-0.05
-0.75
-1.09
-0.36
88424
88424
88424
EMPPM3
DR
0.01
161.20
0.00
11.62
0.55
4301.53
20277.03
3215.25
-120.05
26.58
-95.15 -344159.17
1.62 427342.12
0.17
68.94
0.04
22.29
-0.04
3.78
176487
176487
EMPPM3
DR
0.02
154.99
0.00
11.39
0.28
4766.42
224.18
3242.40
-4.64
38.74
-17.58 -240030.69
1.46 427342.12
0.16
64.43
-0.04
4.04
-0.14
-10.13
88063
88063
EMPPM3
DR
0.00
167.38
-0.01
11.90
0.73
3782.19
13260.85
2592.36
-103.55
0.84
-95.15 -344159.17
1.62 266785.96
0.17
73.26
-0.05
3.51
-0.14
-13.73
88424
88424
34
Table 2: Summary of the Sample of Hedge Funds
This table reports the basic sample statistics and performance of hedge funds from January 31, 2001, until December 31, 2012. Statistics are
compiled only from the date of their listing in the TASS database. All returns are in percent, Rf is the risk free rate, RoR is the rate of return net
of fees, SR is the Sharpe ratio, alpha is estimated from an eight-factor version of Fung and Hsieh’s (2004) model, IR is the information ratio, the
EMPPM is the excess MPPM of Goetzmann et al. (2007), and DR is the doubt ratio of Brown et al. (2010). The EMPPM reveals the monthly
excess risk-adjusted return above the corresponding Russell 2000 return. The column % Doubtful is the percentage of funds that have more than
50% of the observations by strategy and more than six months for a given year by time that have a doubt ratio greater than 150. Assets are in
millions of dollars, age is in years, and returns are in percent per month and net of fees.
Strategy
Number Assets
Age
Rf
RoR
SR
Alpha IR
EMPPM DR
%
Doubtful
124
$251.47 6.29
0.18
0.32
0.45
0.15
1.49
-0.02
868.21 8.06
Convertible Arbitrage
24
$25.73
5.83
0.17
-0.08 -0.06 0.15
0.68
-0.07
-4.69
0.00
Dedicated Short Bias
417
$196.34 6.00
0.10
0.59
0.18
0.26
0.77
0.00
78.25
2.88
Emerging Markets
182
$170.66 5.69
0.15
0.36
0.25
0.25
1.20
-0.03
531.01 6.59
Equity Market Neutral
347
$375.06 6.17
0.15
0.48
0.30
0.28
1.45
0.04
148.10 8.36
Event Driven
$302.15 5.73
0.17
0.37
0.69
0.31
2.89
0.01
1321.50 33.33
Fixed Income Arbitrage 114
1273
$206.00 5.87
0.12
0.12
0.10
-0.02
0.10
0.01
42.92
2.59
Fund of Funds
158
$550.54 5.62
0.13
0.42
-1.03 0.31
0.66
0.04
-183.48 1.27
Global Macro
0.63
Long–Short Equity
1265
$155.44 5.95
0.14
0.44
0.10
0.14
0.30
0.01
-51.59
Hedge
295
$257.77 5.71
0.12
0.56
0.26
0.58
1.17
0.01
841.68 1.02
Managed Futures
266
$437.17 6.22
0.12
0.43
0.23
0.27
1.23
0.03
72.38
5.26
Multi-Strategy
12
$92.53
6.12
0.13
0.55
0.48
0.36
2.57
0.06
658.53 16.67
Options Strategy
123
$273.05 5.66
0.11
0.60
0.41
0.49
2.23
0.04
302.45 18.70
Other
4600
$238.48 5.93
0.13
0.37
0.15
0.18
0.70
0.01
161.20 4.04
Grand Total
1922
$256.67 6.27
0.08
0.45
0.20
0.27
0.87
0.02
154.99 4.16
Live Funds
2678
$221.24 5.62
0.18
0.30
0.09
0.09
0.54
0.00
167.38 3.96
Dead Funds
2033
$223.80 5.17
0.23
0.74
0.22
0.21
1.02
-0.02
274.25 28.43
Jan 2001 to Dec 2006
2567
$246.31 6.32
0.08
0.19
0.11
0.17
0.55
0.03
105.98 13.87
Jan 2007 to Dec 2012
2401
$229.88 5.91
0.10
-0.26 0.01
0.24
0.88
0.00
113.93 13.95
Worst Six Years
2199
$247.06 5.95
0.17
1.03
0.29
0.12
0.52
0.03
210.09 27.24
Best Six Years
35
Table 3: Time Series Characteristics of the Sample of Hedge Funds
This table reports the time series statistics of the performance of hedge funds from January
31, 2001, until December 31, 2012. All returns are in percent, Rf is the risk free rate, RoR is
the rate of return net of fees, SR is the Sharpe ratio, alpha is estimated from an eight-factor
version of Fung and Hsieh’s (2004) model, IR is the information ratio, the EMPPM is the
excess MPPM of Goetzmann et al. (2007), and DR is the doubt ratio of Brown et al. (2010).
The EMPPM reveals the monthly excess risk-adjusted return above the corresponding
Russell 2000 return. The column % Doubtful is the percentage of funds that have six or more
monthly observations in a given year with a doubt ratio greater than 150. Assets are in
millions of dollars, age is in years, and returns are in percent per month and net of fees.
Year
Number
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Total
512
151
246
455
333
336
397
428
282
483
567
410
4600
Assets
$147.72
$156.68
$171.35
$223.09
$253.57
$271.30
$314.81
$309.20
$225.02
$225.39
$207.16
$207.62
$238.48
Age
4.42
4.80
5.32
5.09
5.15
5.57
5.86
5.99
6.41
6.68
6.30
6.62
5.93
Rf
0.31
0.13
0.08
0.10
0.25
0.39
0.38
0.14
0.01
0.01
0.00
0.00
0.13
RoR
SR
0.25 0.13
-0.11 0.08
1.39 0.61
0.80 0.44
0.71 0.29
0.93 0.36
0.85 0.32
-1.70 -0.38
1.45 0.31
0.87 0.34
-0.55 -0.13
0.46 0.24
0.37 0.15
Alpha
0.52
0.45
0.22
0.21
0.14
0.09
0.13
0.15
0.03
0.13
0.24
0.28
0.18
IR
1.78
1.66
1.24
1.12
0.81
0.57
0.60
0.49
0.05
0.28
0.66
1.05
0.70
EMPPM
0.07
0.10
0.06
-0.17
-0.04
-0.02
0.01
0.11
0.36
-0.13
-0.09
-0.03
0.01
DR
89.46
18.73
316.84
360.00
318.70
317.86
261.81
76.39
49.00
104.35
48.32
116.31
161.20
%
Doubtful
8.98
31.79
34.15
33.63
31.53
42.26
28.97
7.48
5.67
18.43
11.46
9.51
20.30
36
Table 4: In-Sample Performance
This table reports in-sample performance, where each portfolio is formed monthly from the
top quintile funds according to the performance measure depicted in the column heading.
Portfolios are held for one month from January 31, 2001, until December 31, 2012. Here,
Sharpe is the Sharpe ratio, alpha is the alpha from Fung and Hsieh’s (2004) eight-factor
model, IR is the information ratio, the EMPPM is the excess MPPM over the Russell 2000
index return, and DR is the doubt ratio. The alphas and EMPPM are in percent. The first half
consists of portfolios formed from January 31, 2001, until December 31, 2006, and the
second half consists of portfolios formed from January 1, 2007, until December 31, 2012.
Two-tailed t-tests for differences in the means of the first- and second-period performance
measures are conducted via bootstrap simulation.
Panel A: January 31, 2001, to December 31, 2012
Sharpe
Alpha
IR
EMPPM
DR
Emerging Markets
1.48
1.42
3.89
0.25
605.84
Event Driven
1.57
1.12
4.03
0.19
492.27
Fund of Funds
1.35
1.16
3.38
0.21
188.80
Long–Short Equity Hedge
1.40
1.25
3.12
0.22
169.36
Managed Futures
4.06
1.59
10.08
0.24
21117.66
Multi-Strategy
1.49
1.09
3.65
0.21
226.23
Average
1.62
1.29
4.09
0.23
991.35
Panel B: First Half Less Second Half
Sharpe
Alpha
IR
EMPPM
DR
***
***
**
Emerging Markets
0.06
0.50
0.66
-0.03
-759.21***
**
***
***
***
Event Driven
0.11
0.28
1.31
-0.06
-218.02
Fund of Funds
0.26***
0.10*** 1.47***
-0.05***
115.32***
Long–Short Equity Hedge
0.18***
0.23*** 1.04***
-0.04***
72.41***
Managed Futures
2.78***
0.15*** 13.95***
-0.09*** 21334.15***
***
***
***
Multi-Strategy
0.25
-0.10
1.15
-0.09***
150.38***
Average
0.25***
0.18*** 1.56***
-0.05***
558.92***
***, **
Significant at the 1% and 5% levels, respectively.
37
Table 5: Out-of-Sample Performance
This table reports out-of-sample performance. Each portfolio is formed monthly from the top
quintile funds according to the performance measure depicted in the heading of the column
and its’ performance is calculated 24 months later and reported below. The performance
ratios are as defined in Table 4. The first half consists of portfolios formed from January 31,
2001 until December 31, 2005 and the second half consists of portfolios formed from January
1, 2006 until December 31, 2010. A two-tailed T-test for differences in the means of first and
second period performance measures are conducted via bootstrap simulation.
Panel A: January 31, 2001 to December 31, 2010
Sharpe
Alpha
IR
EMPPM
DR
Emerging Markets
0.35
0.63
3.69
0.05
363.28
Event Driven
0.52
0.54
2.81
0.04
340.12
Fund of Funds
0.21
0.38
1.36
0.00
106.56
Long–Short Equity Hedge
0.15
0.31
1.24
0.00
44.16
Managed Futures
3.89
0.95
11.97
0.08
26956.67
Multi-Strategy
0.28
0.34
1.98
0.00
128.88
Average
0.40
0.54
2.35
0.01
967.84
Panel B: First Half Less Second Half
Sharpe
Alpha
IR
EMPPM
DR
***
***
***
***
Emerging Markets
0.35
0.01
1.20
0.15
-453.34***
***
***
***
***
Event Driven
0.58
0.26
2.03
-0.03
201.89***
Fund of Funds
0.54***
-0.08***
1.61***
0.01***
175.71***
***
***
***
Long–Short Equity Hedge
0.33
-0.05
0.56
0.00
76.41***
***
***
***
***
Managed Futures
4.45
-0.50
19.24
-0.09
20327.98***
Multi-Strategy
0.35***
-0.39***
0.54***
-0.04***
167.81***
***
***
***
***
Average
0.46
-0.13
1.40
0.01
775.25***
***, **
Significant at the 1% and 5% levels, respectively.
38
Table 6: Performance Persistence
This table reports out of sample performance persistence, where each portfolio is formed
monthly from January 31, 2001, until December 31, 2010, by sorting formation period
performance by the fifth and third quintiles according to the Sharpe ratio, alpha, information
ratio (IR), EMPPM, and doubt ratio (DR) and by strategy. The performance of these
portfolios is then calculated 24 months later. The differences between the out-of-sample
performances of the fifth and third quintiles are reported below. One-tailed t-tests for
differences in the means of the fifth and third quintile performance measures are conducted
via bootstrap.
Panel A: January 31, 2001, to December 31, 2010
Sharpe
Alpha
IR
EMPPM
DR
***
***
***
Emerging Markets
0.13
0.53
3.32
0.02
343.35***
***
***
***
Event Driven
0.20
0.32
1.89
0.01
273.03***
***
***
Fund of Funds
-0.01
0.36
1.21
-0.01
63.58***
Long–Short Equity Hedge
-0.01
0.26***
1.09***
-0.01
25.98***
***
***
***
***
Managed Futures
3.55
0.58
11.13
0.03
26943.67***
Multi-Strategy
0.01
0.16***
1.30***
-0.03
84.92***
***
***
***
**
Average
0.19
0.43
1.93
0.03
938.81***
Panel B: January 31, 2001, to December 31, 2005
Sharpe
Alpha
IR
EMPPM
DR
***
***
***
***
Emerging Markets
0.12
0.64
3.58
0.10
91.77***
Event Driven
0.34***
0.43***
2.77***
0.05***
346.61***
**
***
***
***
Fund of Funds
0.05
0.28
2.05
0.03
125.33***
***
***
***
Long–Short Equity Hedge
0.02
0.26
1.39
0.02
54.02***
Managed Futures
5.88***
0.40*** 21.39***
0.08*** 37411.03***
***
***
Multi-Strategy
0.11
-0.02
1.64
0.01***
146.09***
Average
0.27***
0.35***
2.63***
0.05*** 1313.62***
Panel C: January 31, 2006, to December 31, 2010
Sharpe
Alpha
IR
EMPPM
DR
***
***
***
Emerging Markets
0.14
0.87
3.11
-0.06 561.81***
**
***
***
Event Driven
0.06
0.20
1.01
-0.03
198.08***
Fund of Funds
-0.07
0.43***
0.37***
-0.05
1.84
***
Long–Short Equity Hedge
-0.04
0.26
0.80***
-0.05
-2.06
Managed Futures
1.50***
0.76***
1.78***
0.01** 17080.31***
Multi-Strategy
-0.09
0.33***
0.95***
-0.07
22.61***
***
***
***
Average
0.10
0.51
1.24
0.02
564.01***
***, **
Significant at the 1% and 5% levels, respectively.
39
Table 7: Performance Persistence of the EMPPM
This table reports performance persistence according to the EMPPM. Each month we form
portfolios based on the fifth and third quintiles according to the EMPPM and hold them for
24 months. The differences between the fifth and third quintile portfolios according to the
EMPPM is then measured at three months, six months, 12 months, 18 months, and 24 months
after portfolio formation. One-tailed t-tests for differences in the means of the first and third
quintile performance measures are conducted via bootstrap.
Panel A: January 31, 2001, to December 31, 2010
3 mos.
6 mos.
12 mos.
18 mos. 24 mos.
***
***
***
Emerging Markets
0.18
0.13
0.05
0.03***
0.02
***
***
***
Event Driven
0.13
0.08
0.01
0.00
0.01
***
***
***
Fund of Funds
0.14
0.09
0.02
0.01
-0.01
Long–Short Equity Hedge
0.14***
0.10***
0.01***
0.00
-0.01
***
***
Managed Futures
0.17
0.07
-0.06
0.01
0.03***
Multi-Strategy
0.14***
0.10***
0.03***
-0.01
-0.03
***
***
Average
0.14
0.10
0.04***
0.05***
0.03**
Panel B: January 31, 2001, to December 31, 2005
3 mos.
6 mos.
12 mos.
18 mos. 24 mos.
Emerging Markets
0.17***
0.15***
0.10***
0.11***
0.10***
Event Driven
0.12***
0.09***
0.04***
0.04***
0.05***
***
***
***
***
Fund of Funds
0.14
0.10
0.05
0.05
0.03***
Long–Short Equity Hedge
0.14***
0.11***
0.04***
0.03***
0.02***
***
***
Managed Futures
0.16
0.06
-0.06
0.06
0.08***
***
***
***
Multi-Strategy
0.10
0.08
0.02
0.00
0.01***
Average
0.14***
0.10***
0.04***
0.05***
0.05***
Panel C: January 31, 2006, to December 31, 2010
3 mos.
6 mos.
12 mos.
18 mos. 24 mos.
***
***
Emerging Markets
0.18
0.11
0.01
-0.04
-0.06
Event Driven
0.14***
0.08***
-0.02
-0.05
-0.03
***
***
Fund of Funds
0.14
0.08
0.00
-0.04
-0.05
Long–Short Equity Hedge
0.14***
0.09***
-0.01
-0.03
-0.05
***
***
Managed Futures
0.19
0.10
-0.04
-0.01
0.01**
***
***
***
Multi-Strategy
0.17
0.12
0.03
-0.02
-0.07
***
***
***
Average
0.14
0.11
0.03
-0.01
0.02
***, **
Significant at the 1% and 5% levels, respectively.
40
Table 8: Doubtful Performance
This table reports the doubt ratios of top-performing funds. Portfolios are formed monthly
according to the top quintile candidate measure of performance from January 31, 2001, until
December 31, 2010, and are held for 24 months. The doubt ratios of these funds are then
measured 24 months after portfolio formation and reported in this table. Panel A shows the
results for a two-tailed test of differences in the means of the doubt ratios for top portfolios
formed on the Sharpe, alpha, and information ratio (IR) relative to the doubt ratio for top
portfolios formed on the EMPPM. Panel B shows the results for a two-tailed t-test for
differences in the means of first- and second-period performance measures conducted via a
bootstrap simulation.
Panel A: January 31, 2001, to December 31, 2010
Sharpe
Alpha
IR
EMPPM
***
***
***
Emerging Markets
250.15
0.19
246.12
46.00
***
***
Event Driven
313.83
189.54
289.97
201.88
Fund of Funds
76.66***
35.76
102.86***
41.21
Long–Short Equity Hedge
19.42
9.42
30.58
18.46
***
***
***
Managed Futures
18730.92
444.05
14853.42
2453.94
Multi-Strategy
99.89***
72.75
120.34***
71.89
***
***
***
Average
823.48
108.76
820.98
255.25
Panel A: January 31, 2001, to December 31, 2010
Sharpe
Alpha
IR
EMPPM
***
***
***
Emerging Markets
-387.52
86.77
-266.46
24.27***
***
**
***
Event Driven
222.81
-58.10
201.74
-37.14
***
***
***
Fund of Funds
106.15
43.88
153.76
67.87***
Long–Short Equity Hedge
40.78***
43.08***
80.42***
26.87***
***
**
***
Managed Futures
21737.66
284.54
29950.85
4219.03***
Multi-Strategy
105.50***
49.83***
132.41***
82.82***
***
***
Average
512.43
19.48
899.39
140.55***
***, **
Significant at the 1% and 5% levels, respectively.
41
Table 9: Persistent Doubt
This table reports the persistence of doubt ratios of top-performing funds. Each portfolio is
formed monthly from January 31, 2001, until December 31, 2010, and sorted by quintile
according to the Sharpe ratio, alpha, information ratio (IR), and EMPPM. These portfolios
are then held for two years and their respective doubt ratios are calculated. The differences
between the doubt ratios of the fifth and third quintiles are reported below. One-tailed t-tests
for differences in the means of the first and third quintile performance measures are
conducted via bootstrap.
Panel A: January 31, 2001, to December 31, 2010
Sharpe
Alpha
IR
EMPPM
***
***
Emerging Markets
221.39
-44.05
220.65
-89.05
***
***
***
Event Driven
214.15
62.23
224.36
58.28***
***
***
Fund of Funds
12.57
-46.79
49.29
-35.19
**
***
Long–Short Equity Hedge
16.47
1.79
23.95
-9.01
Managed Futures
18656.62*** -8553.14 14524.42***
-357.04
Multi-Strategy
12.95**
-32.08
52.47***
-16.37
***
***
Average
733.12
-197.83
755.99
-65.97
Panel B: January 31, 2001, to December 31, 2005
Sharpe
Alpha
IR
EMPPM
***
***
Emerging Markets
15.13
-8.04
67.32
13.37***
***
***
Event Driven
273.59
-58.17
292.52
-20.69
***
***
Fund of Funds
23.40
-72.91
99.57
-45.86
Long–Short Equity Hedge
6.54***
-9.69
39.66***
-9.38
***
***
Managed Futures
30155.05
-15713.59 31174.34
531.87
Multi-Strategy
3.04
-79.26
80.92***
-32.95
Average
917.75***
-391.10
1213.17***
-134.60
Panel C: January 31, 2006, to December 31, 2010
Sharpe
Alpha
IR
EMPPM
Emerging Markets
421.08***
-80.06
357.48***
-191.48
***
***
***
Event Driven
154.71
182.62
156.21
137.25***
Fund of Funds
1.73
-20.67
-0.99
-24.53
***
***
***
Long–Short Equity Hedge
26.40
13.27
8.24
-8.64
Managed Futures
8314.75*** -1636.12
618.17**
-1419.75
Multi-Strategy
22.85***
15.10***
22.76***
0.91
Average
548.50***
-4.56
298.81***
2.66
***, **
Significant at the 1% and 5% levels, respectively.
42
Table 10: Skill, Manipulation, or Luck
This table reports whether performance is due to skill, manipulation, or luck. Each portfolio
is formed monthly from January 31, 2001, until December 31, 2010, and sorted by quintile
according to the Sharpe ratio, alpha, information ratio (IR), and doubt ratio (DR). These
portfolios are then held for two years and their respective EMPPM values are calculated. The
difference between the EMPPM values of the fifth and the third quintiles are reported below.
One-tailed t-tests for differences in the means of the first and third quintile performance
measures are conducted via bootstrap. We assume that the EMPPM us subject to no
manipulation so that differences in performance persistence are due to skill, manipulation, or
luck.
Panel A: January 31, 2001, to December 31, 2010
Sharpe
Alpha
IR
EMPPM
***
***
Emerging Markets
0.01
0.03
0.05
0.02
Event Driven
0.00
-0.01
-0.01
0.01
Fund of Funds
0.00
0.00
0.00
-0.01
Long–Short Equity Hedge
0.00
0.00
0.01**
-0.01
***
Managed Futures
0.01
-0.02 0.03
0.03***
***
Multi-Strategy
0.00
0.00 0.02
-0.03
Average
-0.01
-0.01
-0.02
0.03**
Panel B: January 31, 2001, to December 31, 2005
Sharpe
Alpha
IR
EMPPM
***
Emerging Markets
0.00
0.06
-0.01
0.10***
***
Event Driven
-0.01
0.04
0.01
0.05***
***
Fund of Funds
0.00
0.02
0.00
0.03***
Long–Short Equity Hedge
-0.01
0.01***
0.00
0.02***
***
Managed Futures
0.00
-0.02 0.02
0.08***
Multi-Strategy
-0.01
-0.02
0.02**
0.01***
***
Average
-0.01
0.03
-0.01
0.05***
Panel C: January 31, 2006, to December 31, 2010
Sharpe
Alpha
IR
EMPPM
**
***
Emerging Markets
0.02
0.00 0.09
-0.06
Event Driven
-0.42
-0.74
-0.67
-0.03
Fund of Funds
-0.01
-0.02
0.00
-0.05
***
Long–Short Equity Hedge
0.00
-0.02 0.01
-0.05
Managed Futures
0.02
-0.01 0.04***
0.01**
Multi-Strategy
0.00
0.02*** 0.03***
-0.07
Average
-0.01
-0.06
-0.02
0.02
***, **
Significant at the 1% and 5% levels, respectively.
43
0.25
0.2
0.15
0.1
EMPPM3
0.05
Q5
Q3
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
-0.05
-0.1
-0.15
Out of Sample Month
-0.2
-0.25
Figure 1. EMPPM3 by Quintile and Out-of-Sample Month
This figure illustrates the EMPPM by out-of-sample month for the fifth (top), third
(mediocre), and first (bottom) quintiles of performing hedge funds.
Q1
44
250
Number of Funds
200
150
Q5
Q3
100
Q1
50
Out of Sample Month
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Figure 2. Number of Hedge Funds by Quintile and Out-of-Sample Month
This figure illustrates the average number of hedge funds by out-of-sample month that remain
in the fifth (top), third (mediocre), and first (bottom) EMPPM3 quintiles of performing hedge
funds.