THE JOURNAL OF FINANCE * VOL. LVI, NO.

2 * APRIL 2001

Profitability of Momentum Strategies:

An Evaluation of Alternative Explanations
NARASIMHANJEGADEESH and SHERIDAN TITMAN*

ABSTRACT
This paper evaluates various explanations for the profitability of momentum strat-
egies documented in Jegadeesh and Titman (1993). The evidence indicates that
momentum profits have continued in the 1990s, suggesting that the original re-
sults were not a product of data snooping bias. The paper also examines the pre-
dictions of recent behavioral models that propose that momentum profits are due
to delayed overreactions that are eventually reversed. Our evidence provides sup-
port for the behavioral models, but this support should be tempered with caution.

Many portfolio managers and stock analysts subscribe to the view that mo-
mentum strategies yield significant profits. Jegadeesh and Titman (1993)
examine a variety of momentum strategies and document that strategies
that buy stocks with high returns over the previous 3 to 12 months and sell
stocks with poor returns over the same time period earn profits of about one
percent per month for the following year.1 Although these results have been
well accepted, the source of the profits and the interpretation of the evidence
are widely debated. Although some have argued that the results provide
strong evidence of "market inefficiency," others have argued that the returns
from these strategies are either compensation for risk, or alternatively, the
product of data mining.
The criticism that observed empirical regularities arise because of data
mining is typically the hardest to address because empirical research in
nonexperimental settings is limited by data availability. Fortunately, with

* Narashimhan Jegadeesh is from the University of Illinois at Urbana-Champaign and Sheri-

dan Titman is from the University of Texas at Austin and the NBER. This paper has benefited
from the excellent research assistance of Fei Zou and helpful comments from Werner DeBondt,
David Hirshleifer, Rene Stultz, an anonymous referee, and finance workshop participants at
the University of Chicago, University of Illinois at Urbana-Champaign, Indiana University,
NBER Behavioral Finance Conference, University of Texas at Austin and Vanderbilt University.
1 Rouwenhorst (1998) reports that the momentum profits documented by Jegadeesh and
Titman (1993) for the U.S. market also obtain in the European markets. Chui, Titman, and Wei
(2000) document that with the notable exception of Japan and Korea, momentum profits also
obtain in Asian markets. Moskowitz and Grinblatt (1999) and Grundy and Martin (2000) exam-
ine the industry and factor components of momentum profits. Asness (1997), Lee and Swami-
nathan (2000), and Hong, Lim, and Stein (2000) examine the relation between book-to-market
ratios, trading volume and analyst coverage and momentum, and Chan, Jegadeesh, and La-
konishok (1996) examine the relation between earnings momentum and return momentum. See
Haugen (1999) for additional discussion of the momentum effect.

699
700 The Journal of Finance

the passage of time, we now have nine additional years of data that enable
us to perform out-of-sample tests as well as to assess the extent to which
investors may have learned from the earlier return patterns. Using the data
over the 1990 to 1998 sample period, we find that Jegadeesh and Titman
(1993) momentum strategies continue to be profitable and that past winners
outperform past losers by about the same magnitude as in the earlier pe-
riod. This is noteworthy given that other well-known anomalies such as the
small firm effect documented by Banz (1981) and the superior performance
of value stocks relative to growth stocks are not observed after the sample
periods examined in the original studies.2
Given the persistence of this anomaly, it is important to understand its
cause. A number of authors, for example, Barberis, Shleifer, and Vishny (1998),
Daniel, Hirshleifer, and Subrahmanyam (1998), and Hong and Stein (1999),
present behavioral models that are based on the idea that momentum prof-
its arise because of inherent biases in the way that investors interpret in-
formation. Others, however, have argued that it is premature to reject the
rational models and suggest that the profitability of momentum strategies
may simply be compensation for risk. Most notably, Conrad and Kaul (1998)
argue that the profitability of momentum strategies could be entirely due to
cross-sectional variation in expected returns rather than to any predictable
time-series variations in stock returns. Specifically, following Lo and MacKin-
lay (1990), Jegadeesh and Titman (1993), and others, they note that stocks
with high (low) unconditional expected rates of return in adjacent time pe-
riods are expected to have high (low) realized rates of returns in both peri-
ods. Hence, under the Conrad and Kaul (1998) hypothesis, momentum
strategies yield positive returns on average even if the expected returns on
stocks are constant over time.
The behavioral models and Conrad and Kaul's arguments make diametri-
cally opposed predictions about the returns of past winners and losers over
the period following the initial holding period. The behavioral models imply
that the holding period abnormal returns arise because of a delayed over-
reaction to information that pushes the prices of winners (losers) above (be-
low) their long-term values. These models predict that in subsequent time
periods, when the stock prices of the winners and losers revert to their fun-
damental values, the returns of losers should exceed the returns of winners.
In contrast, Conrad and Kaul (1998) suggest that the higher returns of win-

2 The average Fama-French size factor in the sample period 1965 to 1981 (which precedes
the publication of Banz (1981)) is 0.53% per month with a t statistic of 2.34. However, in the
1982 to 1998 sample period, the average size factor is only -0.18% with a t statistic of -1.01.
Similarly, the average book-to-market factor return in the 1990 to 1998 period (subsequent to
the sample period in Fama and French (1993)) is 0.12% per month (t statistic of 0.47), which is
not statistically different from zero. However, there are other out-of-sample results that sup-
port the value/growth phenomenon. For example, Fama and French (1998) and Davis, Fama,
and French (2000) find that this is an international phenomenon and also that this phenom-
enon was observed in sample periods prior to that considered in the early studies.
 Profitability of Momentum Strategies 701

ners in the holding period represent their unconditional expected rates of

return and thus predict that the returns of the momentum portfolio will be
positive on average in any postranking period.
To test the conflicting implications of these theories, we examine the re-
turns of the winner and loser stocks in the 60 months following the forma-
tion date. Consistent with earlier work, we find that over the entire sample
period of 1965 to 1998, the Jegadeesh and Titman (1993) momentum port-
folio yields significant positive returns in the first 12 months following the
formation period. In addition, the cumulative return in months 13 to 60 for
the Jegadeesh and Titman (1993) momentum portfolio is negative, which is
consistent with the behavioral theories but is inconsistent with the Conrad
and Kaul hypothesis.3
Although the negative postholding period returns of the momentum port-
folio appear to support the predictions of the behavioral models, further
analysis suggests that this support should be interpreted with caution. First,
we find strong evidence of return reversals for small firms, but the evidence
is somewhat weak for large firms, particularly when we evaluate portfolio
performance relative to the Fama and French (1993) benchmark. In addi-
tion, although we find strong evidence of return reversals in the 1965 to
1981 period, the evidence of return reversals is substantially weaker in the
1982 to 1998 period. This is noteworthy because there is no distinguishable
difference between either the magnitude or the significance of the momen-
tum profits in the two subperiods.
The remainder of the paper is organized as follows: Section I provides a
brief description of our data and methodology and examines the profitability
of momentum strategies in the 1990s, Section II provides an analysis of the
longer horizon returns, and Section III concludes the paper.

I. Momentum Profits in the 1990s

This section examines whether the profitability of the momentum strat-
egies documented by Jegadeesh and Titman (1993) can be attributed to data
mining. The issue here is fairly straightforward. Stock return data are now
widely available and computing power is fairly cheap. Because there are
potentially large payoffs to any viable model that predicts stock returns (in
terms of publications and/or money management revenues) many academics
and practitioners have, no doubt, independently tested a wide variety of
trading strategies. Therefore, it is difficult to assess the significance of in-
dividual studies that find that a particular trading strategy is profitable.
We address the data mining issue in the context of the Jegadeesh and
Titman (1993) six-month momentum strategy, which was previously shown
to earn abnormal returns of about one percent per month with a t statistic of
3.07 over the 1965 to 1989 sample period. When this strategy is viewed as a

3 In an independent paper, Lee and Swaminathan (2000) examine the relations between
momentum, volume, and long horizon returns to test the predictions of behavioral models.
702 The Journal of Finance

single experiment, standard statistical theory indicates that the probability

of observing a t statistic at least as large as 3.07 under the hypothesis of
market efficiency is less than 0.11 percent. Based on this, Jegadeesh and
Titman (1993) conclude that the hypothesis of market efficiency can be re-
jected at even the most conservative levels of significance. This inference,
however, ignores the fact that there were many other tests independently
carried out by other researchers over the same sample period that were
perhaps not profitable and hence were not reported. The fact that the evi-
dence of momentum profits gained attention can be attributed to the fact
that it yielded the highest test statistic among the many tests that were
carried out collectively. Under this interpretation, the test statistic in Je-
gadeesh and Titman should be viewed as the highest order statistic across
many tests rather than as a conventional test statistic from a single exper-
iment. The distribution of this order statistic, of course, is not normal.
To formalize the statistical analysis, suppose that researchers collectively
tested n independent trading strategies during the Jegadeesh and Titman
(1993) sample period.4 Also, suppose that the momentum strategy yielded
the highest test statistic among these n strategies. The cumulative distri-
bution of the largest order statistic is F , where F is the cumulative stan-
dard normal distribution.5 As stated earlier, if the Jegadeesh and Titman
(1993) test is viewed in isolation then n = 1 and the probability of observing
a t statistic this large is 0.11 percent. However, if n = 100, for example, then
the probability that the largest test statistic is at least 3.07 is about 10
percent. If n = 650, then the p value based on this test statistic drops below
50 percent. So the perception of how strong the Jegadeesh and Titman (1993)
evidence is in rejecting the efficient market hypothesis depends on the read-
ers' priors about how many other independent and unreported tests that
failed to reject market efficiency had been carried out.

A. Portfolio Formation
The advantage of an out-of-sample test is that it significantly reduces the
number of strategies that researchers can potentially search over, greatly
reducing n, and thus increasing the informativeness of the tests. For this
reason, we reexamine the Jegadeesh and Titman (1993) trading strategy in
the time period subsequent to their analysis.
Our sample is constructed from all stocks traded on the New York Stock
Exchange (NYSE), American Stock Exchange, and Nasdaq. We exclude all
stocks priced below $5 at the beginning of the holding period and all stocks
with market capitalizations that would place them in the smallest NYSE
decile. We exclude these stocks to ensure that the results are not driven

4 The distribution of the highest order statistic will have to be numerically computed if the
n trading strategies examined in this sample period are correlated.
5 We are assuming here that the degrees of freedom for the t statistic are sufficiently large
so that the t distribution can be approximated by the standard normal distribution.
 Profitability of Momentum Strategies 703

Formation Period Holding Period Post-Holding Period

(Month -5 to Month 0) (Month 1 to Month 6 or 12) (Month 13 to Month 60)

Figure 1. Time line showing sample periods.

primarily by small and illiquid stocks or by bid-ask bounce.6 Our sample

differs from Jegadeesh and Titman (1993) because we include Nasdaq stocks
but exclude small and low-priced stocks. The addition of Nasdaq stocks and
the deletion of low-priced stocks, however, have very little effect on average
returns over various horizons we consider, but they decrease standard errors
and significantly lower the magnitude of the negative January returns.
Following Jegadeesh and Titman (1993), at the end of each month we rank
the stocks in our sample based on their past six-month returns (Month -5 to
Month 0) and then group the stocks into 10 equally weighted portfolios based
on these ranks. Each portfolio is held for six months (Month 1 to Month 6)
following the ranking month. The various periods we consider are presented
in the time line in Figure 1, which also presents a postholding period (Month
13 to Month 60) that we consider in the next section.
To increase the power of our tests, we construct overlapping portfolios. In
other words, a momentum decile portfolio in any particular month holds
stocks ranked in that decile in any of the previous six ranking months. For
instance, a December winner portfolio comprises 10 percent of the stocks
with the highest returns over the previous June to November period, the
previous May to October, and so on up to the previous January to June
period. Each monthly cohort is assigned an equal weight in this portfolio.

B. Holding Period Returns

Table I presents average monthly returns for the 10 momentum portfolios.
Portfolio P1 comprises stocks with the largest ranking period returns and
P10 comprises stocks with the lowest ranking period returns. The table re-
veals a monotonic relation between returns and momentum ranks over the
1965 to 1989 sample period, confirming the results in Jegadeesh and Titman
(1993). The difference between the P1 and P10 portfolio returns during
this time period is 1.17 percent per month, which is reliably different from

6 Conrad and Kaul (1993) point out that much of the evidence of long horizon mean reversion
in DeBondt and Thaler (1985) is due to the inclusion of low-priced stocks. The results in this
paper, however, are similar both with and without the $5 price screen except in Januaries. The
low-priced stocks exhibit large return reversals in January and, as a result, the momentum
strategies earn larger negative returns in January if these stocks are included. When all cal-
endar months are considered, momentum profits are about one percent per month with or
without the $5 price screen.
704 The Journal of Finance

t
P9 P8 P7 P6 P5 P4 P3 P2 P1 or
P10 the
EWI This
decile 10
P1-Plo median
statistic(Past (Past
Nasdaq table
cutoff).
market percent
equal-weighted
losers) winners) of
The reports
cap excluding
the
the
portfolio
"Small
NYSE stocksof
stocks
1.096.461.230.420.901.051.111.13
1.17
1.191.281.391.65
10
Cap"
1965-1998 stock monthly
with
priced
and
the
All less percent
of returns
1.104.961.170.460.94 1.181.21
1.051.121.15 1.301.411.63 next
"Large
than thefor
Stocksrespectively.
1965-1989
$5
Cap"
at highest
stocks
"EWI"
is the
momentum
with
1.044.711.390.300.771.03 1.12
1.091.09 1.131.21
1.321.69
the returns,
1990-1998 the
subsamples
and
beginning
portfolios
of so highestMomentum
returns
1.137.411.420.280.84 1.26
1.091.141.19 1.261.371.45
1.70 on theon.
comprise
1965-1998 the formedTable
The
returns I
stocks
holding
"All based
Portfolio
in over
Small on
1.201.261.33
1.195.601.340.350.951.17 1.421.50
1.34 1.69 the the
period
stocks"
Cap
1965-1989 past
"All
equal-weighted
and
Returns
previous
sample
index
stockssix
Stocks"
0.985.74 0.991.01
1.650.080.540.89 1.06 1.231.33
1.05 1.73 of six-month
in
1990-1998
theincludes
stocks
sample months,
all returns
in
that P2
1.034.34 1.091.09
0.860.701.001.04 1.121.25
1.051.10 1.56 is and
smallest
each
are stocks
1965-1998 the
held
tradedfor
market
sample.
smaller
Large capon six
1.041.05
1.003.550.850.680.961.00 1.07
1.00 1.52
1.101.24 and the
Cap
1965-1989
decile equal-weighted
larger months.
NYSE,
P1
(NYSE is
than
1.122.590.880.781.091.171.231.20 1.201.191.27
1.19 1.66 portfolio
1990-1998 thesizeAMEX,
of the
 Profitability of Momentum Strategies 705

zero.7 The table reveals that this return pattern continues in the more recent
1990 to 1998 period. In this period, past winners outperformed past losers
by 1.39 percent per month, which is close to the corresponding returns in the
original Jegadeesh and Titman (1993) sample period.
To put the results in perspective, Table I also presents the average equal-
weighted returns for the stocks in the sample. Interestingly, the winners (P1
portfolio) outperform the equal-weighted index by 0.56 percent per month,
whereas the losers (PlO portfolio) underperform the index by 0.67 percent
per month. These results suggest that both winners and losers contribute
about equally to momentum profits.
Table I also separately presents momentum returns generated by small
and large stocks. Firms with market capitalizations above the median NYSE
listed stock at the beginning of each holding period are classified as large
stocks and the rest of the sample is classified as small stocks. We examine
these subsamples separately for several reasons. First, because it is expen-
sive to trade smaller capitalization stocks, it may not be possible to execute
active trading strategies with these stocks. Therefore, from a practical stand-
point, the evidence will be more convincing if we also find momentum prof-
its for larger firms. Second, differences in the out-of-sample returns of
momentum portfolios consisting of large and small stocks can potentially
provide insights about the extent to which investors learn about the profit-
ability of these momentum strategies and exploit them. Specifically, in the
past decade, momentum strategies have become more popular among insti-
tutional investors, perhaps because of the dissemination of information re-
lating to the performance of these strategies. One might expect that the
trading activities of these institutions would eliminate the momentum ef-
fect, at least for the relatively large stocks that they can trade at low costs.
The results in Table I indicate that the momentum effect continues in the
1990s for large stocks as well as small stocks. The differences between win-
ner and loser portfolio returns are about equal across the two subperiods for
both the small and large firm subsamples. In all cases the returns are close
to being monotonically related to past six-month returns. These results also
indicate that the momentum profits come from the buy as well as the sell
side of this strategy.8
Our findings relating to the profits from small versus large stocks and the
long side versus the short side of our trading strategy are intriguing, given
the conventional wisdom that, with learning, profit opportunities will be
sustained longer when there are higher costs of implementing trading strat-

I
The t statistic now is 4.96 compared with that of 3.07 reported by Jegadeesh and Titman
(1993). Although the magnitude of momentum profits here is similar to that in Jegadeesh and
Titman (1993), the variability is now smaller because of the exclusion of small stocks and stocks
priced below $5.
8 This observation should be contrasted with the observation in Hong, Lim, and Stein (2000),
who suggest that most of the momentum profits come from the short side of the transaction.
Their conclusions are perhaps driven by the fact that they form only three momentum portfo-
lios as opposed to the decile portfolios formed here.
706 The Journal of Finance

egies. The transaction costs explanation suggests that momentum profits

will dissipate faster for large stocks, which are cheaper to trade, and that
because of the costs of short-selling, the profits from trading past winners
should be eliminated more quickly than the profits from trading past losers.
These predictions are not supported by the data.

C. Seasonality
Jegadeesh and Titman (1993) find a striking seasonality in momentum
profits. They document that the winners outperform losers in all months
except January, but the losers significantly outperform the winners in Jan-
uary. This seasonality could potentially be a statistical fluke; January is one
of twelve calendar months and it is possible that in any one calendar month
momentum profits are negative. Here again, we can examine the out-of-
sample performance of the strategy in January to examine whether this
seasonality is real or whether it was the result of looking too closely at the
data.
Our unreported analysis that replicates the momentum strategies using
the sample selection criteria in Jegadeesh and Titman (1993) found results
very similar to theirs for the 1990s, suggesting that the earlier finding was
not a statistical fluke. Table II reports the momentum profits in January
and non-January for our sample that excludes both stocks priced under $5
per share and stocks in the smallest size decile. The momentum profits in
January for this sample are also negative in all subperiods but they are only
marginally significant. This indicates that most of the previously reported
negative returns in January are due to small and low-priced stocks, which
are likely to be difficult to trade at the reported CRSP prices. The January
momentum profits, however, are significantly smaller than the momentum
profits in other calendar months in all sample periods.

D. Portfolio Characteristics and Abnormal Returns

This subsection examines the characteristics of the momentum portfolios
and the risk-adjusted momentum portfolio returns. Table III presents the
characteristics of the momentum portfolios. The size decile ranks in this
table are computed using NYSE size decile cutoffs with the size rank of one
being the smallest and the size rank of ten being the largest. Both winners
and losers tend to be smaller firms than the average stock in the sample,
because smaller firms have more volatile returns and are thus more likely to
be in the extreme return sorted portfolios. The average size rank for the
winner portfolio is larger than that for the loser portfolio.
Table III also presents the sensitivities of these portfolios to the three
Fama-French factors. The results indicate that the market betas for winners
and losers are virtually equal. However, the losers are somewhat more sen-
sitive to the size factor than are the winners (the loadings for the losers is
0.55 versus 0.41 for the winners). Moreover, the winners have a loading of
 Profitability of Momentum Strategies 707

Table II
Momentum Portfolio Returns in January and outside January
This table reports the average monthly momentum portfolio returns, the associated t statistics
to test whether the returns are reliably different than zero, and the percentage of monthly
momentum returns that are positive. The table reports returns for January as well as non-
January months, and returns in the 1965-1989, Jegadeesh and Titman (1993) sample period,
the 1990-1998 subsequent period, as well as the entire 1965-1998 period. The sample includes
all stocks traded on the NYSE, AMEX, or Nasdaq, excluding stocks priced less than $5 at the
beginning of the holding period and stocks in the smallest market cap decile (NYSE size decile
cutoff). The momentum portfolios are formed based on past six-month returns and held for six
months. P1 is the equal-weighted portfolio of 10 percent of the stocks with the highest past
six-month returns and P10 is the equal-weighted portfolio of the 10 percent of the stocks with
the lowest past six-month returns.

Percent
P1 PlO P1-PlO t statistic Positive
1965-1989
Jan 4.01 5.67 -1.67 -1.50 36
Feb-Dec 1.42 -0.01 1.43 6.20 69
All 1.63 0.46 1.17 4.96 66
1990-1998
Jan 1.72 2.95 -1.24 -2.08 11
Feb-Dec 1.69 0.06 1.63 5.32 69
All 1.69 0.30 1.39 4.71 64
1965-1998
Jan 3.40 4.95 -1.55 -1.87 29
Feb-Dec 1.49 0.01 1.48 7.89 69
All 1.65 0.42 1.23 6.46 66

-0.245 on the HML factor whereas the losers have a loading of -0.02. These
results indicate that the losers are riskier than the winners because they are
more sensitive to all three Fama-French factors.
Table IV reports the alphas of the various momentum portfolios estimated
by regressing the monthly momentum returns (less the risk-free rate except
for the zero investment P1-PlO portfolio) on the monthly returns of both the
value-weighted index less the risk-free rate and the three Fama-French fac-
tors. The CAPM alpha for the winner minus loser portfolio is about the same
as the raw return difference, as both winners and losers have about the
same betas. Consistent with Fama and French (1996), the Fama-French al-
pha for this portfolio is also reliably positive. The Fama and French alpha
for this portfolio is 1.36 percent, which is larger than the corresponding raw
return of 1.23 percent. This difference arises because the losers are more
sensitive to the Fama-French factors, as reported in Table III.

II. Postholding Period Returns of Momentum Portfolios

A number of hypotheses have been proposed in the literature to explain
the profitability of momentum strategies. This section examines the per-
708 The Journal of Finance

Table III
Portfolio Characteristics
This table reports the characteristics of momentum portfolios. The sample includes all stocks
traded on the NYSE, AMEX, or Nasdaq, excluding stocks priced less than $5 at the beginning
of the holding period and stocks in the smallest market cap decile (NYSE size cutoff). P1 is the
equal-weighted portfolio of 10 percent of the stocks with the highest past six-month returns, P2
is the equal-weighted portfolio of the 10 percent of the stocks with the next highest past six-
month returns, and so on. Average size decile rank is the average rank of the market capital-
ization of equity (based on NYSE size decile cutoffs) of the stocks in each portfolio at the
beginning of the holding period. FF factor sensitivities are the slope coefficients in the Fama-
French three-factor model time-series regressions. "Market" is the market factor (the value-
weighted index minus the risk-free rate), "SMB" is the size factor (small stocks minus big
stocks) and "HML" is the book-to-market factor (high minus low book-to-market stocks). The
sample period is January 1965 to December 1998.

FF Factor Sensitivities
Average Size
Decile Rank Market SMB HML

P1 4.81 1.08 0.41 -0.24

P2 5.32 1.03 0.23 0.00
P3 5.49 1.00 0.19 0.08
P4 5.51 0.99 0.17 0.14
P5 5.49 0.99 0.17 0.17
P6 5.41 0.99 0.19 0.19
P7 5.36 0.99 0.22 0.19
P8 5.26 1.01 0.24 0.16
P9 5.09 1.04 0.30 0.11
PlO 4.56 1.12 0.55 -0.02
P1-Plo 0.25 -0.04 -0.13 -0.22

formance of momentum portfolios over longer horizons to differentiate be-

tween these hypotheses. Specifically, we examine the returns of the portfo-
lios in the periods following the holding periods considered in the previous
section.

A. Market Underreaction
The null hypothesis of our postholding period tests is that the momentum
profits arise because investors underreact to ranking period information,
which is gradually incorporated into stock prices during the holding period.
Barberis et al. (1998) discuss how a "conservatism bias" might lead investors
to underreact to information in a way that is consistent with our null hy-
pothesis. The conservatism bias, identified in experiments by Edwards (1968),
suggests that individuals underweight new information in updating their
priors. If investors act in this way, prices will tend to slowly adjust to infor-
mation, but once the information is fully incorporated in prices, there is no
further predictability in stock returns. This interpretation suggests that the
postholding period returns will be zero.
 Profitability of Momentum Strategies 709

Table IV
CAPM and Fama-French Alphas
This table reports the risk-adjusted returns of momentum portfolios. The sample comprises all
stocks traded on the NYSE, AMEX, or Nasdaq, excluding stocks priced less than $5 at the
beginning of the holding period and stocks in the smallest market cap decile (NYSE size decile
cutoff). P1 is the equal-weighted portfolio of 10 percent of the stocks with the highest past
six-month returns, P2 is the equal-weighted portfolio of the 10 percent of the stocks with the
next highest past six-month returns, and so on. This table reports the intercepts from the
market model regression (CAPM Alpha) and Fama-French three-factor regression (FF Alpha).
The sample period is January 1965 to December 1998. The t statistics are reported in paren-
theses.

CAPM Alpha FF Alpha

P1 0.46 0.50
(3.03) (4.68)
P2 0.29 0.22
(2.86) (3.51)
P3 0.21 0.10
(2.53) (2.31)
P4 0.15 0.02
(1.92) (0.41)
P5 0.13 -0.02
(1.70) (-.43)
P6 0.10 -0.06
(1.22) (-1.37)
P7 0.07 -0.09
(0.75) (-1.70)
P8 -0.02 -0.16
(-0.19) (-2.50)
P9 -0.21 -0.33
(-1.69) (-4.01)
P1O -0.79 -0.85
(-4.59) (-7.54)
P1-PlO 1.24 1.36
(6.50) (-7.04)

B. Behavioral Models
The recent behavioral literature is motivated in part by a body of evidence
that suggests that the postholding period returns may in fact be negative.
For example, Jegadeesh and Titman (1993) present some evidence that the
postholding period average return of their momentum portfolio is negative,
and DeBondt and Thaler (1985) provide stronger evidence of longer-term
overreaction. In addition, the earlier mentioned evidence of return predict-
ability based on book-to-market ratios is consistent with the existence of
overreaction.
To explain the long-term overreaction as well as the shorter-term mo-
mentum, Barberis et al. (1998) present a model that combines the con-
servatism bias with what Tversky and Kahneman (1974) refer to as a
710 The Journal of Finance

"representative heuristic," which is the tendency of individuals to identify

"an uncertain event, or a sample, by the degree to which it is similar to
the parent population." In the context of stock prices, Barberis et al. (1998)
argue that the representative heuristic may lead investors to mistakenly
conclude that firms realizing extraordinary earnings growths will continue
to experience similar extraordinary growth in the future. They argue that,
although the conservatism bias in isolation leads to underreaction, this
behavioral tendency in conjunction with the representative heuristic can
lead to long horizon negative returns for stocks with consistently high re-
turns in the past.9
Daniel et al. (1998) and Hong and Stein (1999) propose alternative mod-
els that are also consistent with short-term momentum and long-term re-
versals. Daniel et al. (1998) argue that informed traders suffer from a
"self-attribution" bias. In their model, investors observe positive signals
about a set of stocks, some of which perform well after the signal is re-
ceived. Because of their cognitive biases, the informed traders attribute
the performance of ex post winners to their stock selection skills and that
of the ex post losers to bad luck. As a result, these investors become over-
confident about their ability to pick winners and thereby overestimate the
precision of their signals for these stocks. Based on their increased confi-
dence in their signals, they push up the prices of the winners above their
fundamental values. The delayed overreaction in this model leads to mo-
mentum profits that are eventually reversed as prices revert to their
fundamentals.
Hong and Stein (1999) do not directly appeal to any behavioral biases on
the part of investors, but they consider two groups of investors who trade
based on different sets of information. The informed investors or the "news
watchers" in their model obtain signals about future cash flows but ignore
information in the past history of prices. The other investors in their model
trade based on a limited history of prices and, in addition, do not observe
fundamental information. The information obtained by the informed inves-
tors is transmitted with a delay and hence is only partially incorporated in
the prices when first revealed to the market. This part of the model con-
tributes to underreaction, resulting in momentum profits. The technical trad-
ers extrapolate based on past prices and tend to push prices of past winners
above their fundamental values. Return reversals obtain when prices even-
tually revert to their fundamentals. Both groups of investors in this model
act rationally in updating their expectations conditional on their informa-
tion sets, but return predictability obtains due to the fact that each group
uses only partial information in updating its expectation.

9 The time horizon over which various biases come into play in the Barberis et al. (1998)
(and in other behavioral models) is unspecified. One could argue that the six-month ranking
period used in this paper may not be long enough for delayed overreaction due to the repre-
sentative heuristic effect. In such an event we would only observe underreaction due to the
conservatism bias.
 Profitability of Momentum Strategies 711

C. The Conrad and Kaul Hypothesis

Conrad and Kaul (1998) start with the hypothesis that stock prices follow
random walks with drifts, and the unconditional drifts vary across stocks.
The Conrad and Kaul (1998) hypothesis suggests that the differences in un-
conditional drifts across stocks explain momentum profits. Because any pre-
dictability under the Conrad and Kaul (1998) hypothesis is due to differences
in unconditional drifts across stocks and is not due to the random compo-
nent of price changes in any particular period, the profits from a momentum
strategy should be the same in any postranking period. In other words, this
hypothesis predicts that the stocks on the long side of the momentum port-
folio should continue to outperform stocks on the short side by the same
magnitude in any postranking period.

D. The Postholding Period Evidence

Figure 2 summarizes (1) the underreaction, (2) the overreaction and price
correction, and (3) the Conrad and Kaul (1998) hypotheses. Although all
three hypotheses imply momentum profits in the holding period, the posthold-
ing period performance of the momentum portfolios differs sharply under
the three hypotheses as discussed above.
To test these competing hypotheses, we examine the returns of the mo-
mentum portfolio following the initial formation date. The theoretical mod-
els do not offer any guidance regarding the length of the postholding period
over which return reversals due to price corrections are expected to occur.
Jegadeesh and Titman (1993) examine momentum portfolio returns up to
three years after portfolio formation with the idea that even if markets are
not fully efficient, the effect of any information will likely be impounded in
prices within this time frame. Recent studies of initial public offerings and
seasoned equity offerings, however, find evidence of underperformance even
five years after the events.10 Therefore we extend the postholding period to
five years in this study.
Figure 3 presents cumulative momentum profits over a 60-month postfor-
mation period. Over the 1965 to 1998 sample period, the results reveal a
dramatic reversal of returns in the second through fifth years. Cumulative
momentum profits increase monotonically until they reach 12.17 percent at
the end of Month 12. From Month 13 to Month 60 the momentum profits are
on average negative. By the end of Month 60 the cumulative momentum
profits decline to -0.44 percent. This evidence is clearly inconsistent with
the Conrad and Kaul (1998) hypothesis and tends to support the behavioral
hypotheses.' I

0 See Loughran and Ritter (1995) and Spiess and Affleck-Graves (1995), among others.
" Jegadeesh and Titman (2000) show that the main results in Conrad and Kaul (1998) are
largely driven by small sample biases in their experiments.
712 The Journal of Finance

returns
Figure
2. 4
under
Long

various
horizon
_____
K
hypotheses.
momentum TnHolding
hypothesis

profits P]rind

under

different

Event
Underreaction
Ti
hypotheses.
me

This

figure

presents
the
----Overreaction
PnRt_Tn11ding
and
expected
Price Perind

pattern
of
Correction

momentum

portfolio
 Profitability of Momentum Strategies 713

le C
event 2
NYSE,
Figure 1 1
-8 -4 -2 4 8 2 4
3.
momentum 6-
period.
AMEX,
or
Stocks
portfolio
Cumulative
Nasdaq. C)
priced

less
The
construction.
than
momentum
C)
sample
$5
'n
at
profits.
11-
the 1O0.%
comprises
This
all C)~
~~C~
figure -1965
beginning
stocks to CIJ
of 1
that 997 CI)
each presents
-
are - ClJ

event -
ClJ
larger Event
1965
cumulative C

month to
than 1 Month
C~)
are 981
the
Cx)

momentum
Cr)
..
smallest 1
excluded
982
from portfolio
to
NYSE 1
the
997
returns
market 0C)')
--
sample.with
capa r
-

See
0)
decile
sample
Table
at of
I
forthe U)

a stocks
U)

U)
traded
beginning
of on
description
of the the
714 The Journal of Finance

Table V
Seasonality in Longer Horizon Momentum Profits
and Fama and French Factors
Panel A presents the average monthly returns on the three Fama and French risk factors.
"Market" is the market factor (value-weighted index minus risk-free rate), "SMB" is the size
factor, and "HML" is the book-to-market factor. Panel B presents average monthly momentum
portfolio (Pl-P10) returns one, two, three, four, and five years after portfolio formation. See
Table I for a description of portfolio construction. Panel C presents the intercepts from the
Fama-French three-factor regressions fitted over all months in the sample period and sepa-
rately within and outside January. The t statistics are reported in parentheses. The sample
period is January 1965 to December 1998.

Panel A. Average Factors

Fama-French Factors
Calendar
Months Market SMB HML

January 1.83 2.29 2.34

(2.00) (3.63) (4.26)
Feb-Dec 0.39 -0.02 0.23
(1.74) (-.12) (1.83)
All 0.51 0.17 0.41
(2.32) (1.21) (3.18)

Panel B. Raw Returns

Calendar Months Months Months Months Months Months

Months 1 to 12 13 to 24 25 to 36 37 to 48 49 to 60 13 to 60

January -1.69 -2.87 -1.49 -0.48 -0.59 -1.36

(-2.49) (-5.46) (-3.50) (-1.35) (-1.37) (-5.12)
Feb-Dec 1.26 0.00 -0.15 -0.20 -0.28 -0.16
(8.31) (-0.04) (-1.55) (-2.31) (-3.11) (-3.01)
All 1.01 -0.24 -0.26 -0.23 -0.31 -0.26
(6.52) (-2.23) (-2.70) (-2.63) (-3.40) (-4.65)

Table V presents further details on the momentum portfolio returns in the

first five years after portfolio formation. The average profit in the first 12
months of the holding period is 1.01 percent per month, the average profit is
-0.24 percent per month in the second year, -0.26 percent in the third year,
-0.23 percent per month in the fourth year, and -0.31 percent per month in
the fifth year.12 The average return of -0.26 percent over the second through
fifth years is reliably less than zero, which is consistent with the behavioral
models that predict that the momentum profits will eventually reverse.
As Table III reports, the loser portfolios have larger sensitivities to the
Fama and French size and book-to-market factors. The negative returns ob-
served in the postholding period may therefore represent compensation for

12 Momentum profits are negative in four of the five years from Year 6 thorough Year 10, but

they are not reliably different from zero.

 Profitability of Momentum Strategies 715

Table V-Continued

Panel C. Fama-French Three-factor Alphas

Calendar Months Months Months Months Months Months

Months 1 to 12 13 to 24 25 to 36 37 to 48 49 to 60 13 to 60

Winner Portfolio (P1)

January -0.06 -0.11 0.29 0.07 0.18 0.11

(- 0.11) (-0.33) (1.13) (0.29) (0.56) (0.66)
Feb-Dec 0.42 -0.21 -0.18 -0.14 -0.18 -0.18
(4.82) (-3.24) (-2.84) (-2.37) (-3.04) (-3.93)
All 0.37 -0.23 -0.16 -0.12 -0.16 -0.17
(4.11) (-3.48) (-2.57) (-2.10) (-2.75) (-3.83)

Loser Portfolio (P10)

January 0.42 1.02 0.38 0.38 0.22 0.50

(1.18) (3.73) (1.36) (1.14) (0.74) (2.76)
Feb-Dec -0.88 -0.29 -0.10 0.02 0.09 -0.07
(-9.22) (-4.35) (-1.58) (0.31) (1.60) (-1.58)
All -0.80 -0.20 -0.06 0.03 0.10 -0.03
(-8.54) (-2.97) (-0.96) (0.54) (1.71) (-0.74)

Momentum Portfolio (Pl-P10)

January -0.48 -1.13 -0.09 -0.31 -0.04 -0.39

(-0.57) (-2.20) (-0.21) (-0.69) (-0.06) (-1.72)
Feb-Dec 1.30 0.08 -0.08 -0.16 -0.28 -0.11
(8.60) (0.87) (-0.90) (- 1.91) (-3.14) (-2.69)
All 1.17 -0.03 -0.10 -0.16 -0.26 -0.14
(7.57) (-0.31) (-1.11) (-1.84) (-2.94) (-3.26)

factor risks. Furthermore, the Fama-French factors exhibit a January sea-

sonal. In our sample period, the Fama-French size factor has an average
return of 2.29 percent in January compared with -0.02 percent outside Jan-
uary, whereas the book-to-market factor has an average return of 2.34 per-
cent in January and 0.23 percent outside January. If the negative returns for
the momentum portfolio are due to their exposures to the Fama-French fac-
tors, then we would expect that a large portion of these negative returns will
also be concentrated in January.
Table V (Panel B) presents the momentum profits in January and outside
January over various horizons. The average postholding period momentum
profit is negative each year and is significantly negative in each January.
The momentum profits outside January are close to zero in the second and
third years following formation but are reliably negative in the fourth and
fifth years.
716 The Journal of Finance

Panel C of Table V presents the Fama-French three-factor alphas for the

zero cost momentum portfolio for both the winner's and loser's portfolios.
The table reveals that the alpha of the zero cost momentum portfolio is
approximately half the size of the raw returns in Month 13 to Month 60. The
alphas are significantly negative only in years four and five. Our separate
analysis of winners and losers indicates that the return reversals observed
for the zero cost momentum portfolio is entirely due to the negative alphas
of the winners. Indeed, the evidence here indicates that the losers as well as
winners experience negative abnormal returns in years two through five
(see Table V, Panel C). This evidence is inconsistent with the idea that the
momentum in loser returns is generated as a result of positive feedback
trading that is later reversed.

E. Subperiod Evidence
To investigate the robustness of long horizon return reversals we examine
the performance of momentum portfolios in two separate time periods, the
1965 to 1981 and 1982 to 1998 subperiods. In addition to being the half-way
point, 1981 represents somewhat of a break point for the Fama and French
factor returns. The Fama-French SMB and HML factors have higher returns
in the pre-1981 period (the monthly returns of the SMB and HML factors
average 0.53 percent and 0.48 percent, respectively; see Table VI) than in
the post-1981 period (the monthly returns of the SMB and HML factors
average -0.18 percent and 0.33 percent, respectively).
The evidence in Table VI and Figure 3 indicates that the momentum strat-
egy is significantly profitable, and quite similar in both subperiods in the
first 12 months following the formation date. The returns in the postholding
periods, however, are quite different in the two subperiods. In the 1965 to
1981 subperiod, the cumulative momentum profit declines from 12.10 per-
cent at the end of Month 12 to 5.25 percent at the end of Month 36 and then
declines further to -6.29 percent at the end of Month 60. In fact, we found
that the momentum profit is negative in each event month after Month 12
in this subperiod. In the 1982 to 1998 subperiod, the cumulative profit de-
creases from 12.24 percent at the end of month 12 to 6.68 percent at the end
of Month 36 and then stays at about the same level for the next 24 months.
Tables VII and VIII replicate Tables V and VI on the large- and small-firm
subsamples. For the large firms we find strong evidence of return reversals
when we examine raw returns. However, the Fama-French alpha in Month
13 to Month 60 is only -0.07 percent per month, which is not statistically
significant. Furthermore, evidence of return reversals is observed only in
the first subperiod and the average postholding period abnormal return in
the second subperiod is only -0.01 percent per month.
For the small stocks, we find somewhat stronger evidence of postholding
period return reversals. Here again, the evidence of return reversals is con-
siderably stronger in the first half of the sample, although as before, the
 Profitability of Momentum Strategies 717

Table VI
Longer Horizons Momentum Profits and Fama
and French Factors-Subperiod Results
This table presents momentum profits and Fama-French factors within two subperiods. Panel
A presents the average monthly returns on the three Fama and French risk factors. "Market"
is the market factor (the value-weighted index minus the risk-free rate), "SMB" is the size
factor (small stocks minus big stocks), and "HML" is the book-to-market factor (high minus low
book-to-market stocks). Panel B presents average monthly momentum portfolio (Pl-P10) re-
turns one, two, three, four, and five years after portfolio formation. See Table I for a description
of portfolio construction. Panel C presents the intercepts from the Fama-French three-factor
regressions fitted within each subperiod. The t statistics are reported in parentheses.

A. Fama-French Factors

Market SMB HML

1965-1981 0.13 0.53 0.48

(0.42) (2.33) (2.60)
1982-1998 0.89 -0.18 0.33
(2.95) (- 1.01) (1.88)

B. Raw Returns

Months Months Months Months Months Months

1 to 12 13 to 24 25 to 36 37 to 48 49 to 60 13 to 60
1965-1981 1.01 -0.30 -0.26 -0.48 -0.50 -0.38
(3.99) (-1.80) (-1.84) (-3.78) (-3.58) (-4.45)
1982-1998 1.02 -0.19 -0.26 0.03 -0.11 -0.13
(5.62) (-1.34) (- 1.99) (0.22) (-1.00) (-1.93)

C. Fama-French Three-Factor Alphas

Months Months Months Months Months Months

1 to 12 13 to 24 25 to 36 37 to 48 49 to 60 13 to 60
1965-1981 1.26 0.01 -0.03 -0.37 -0.36 -0.19
(5.09) (0.04) (-0.28) (-2.94) (-2.63) (-3.11)
1982-1998 1.12 -0.05 -0.12 0.06 -0.19 -0.08
(5.95) (-0.42) (-0.99) (0.56) (-1.70) (-1.34)

magnitude of the momentum profits in the holding period is similar in the

two sample periods.
In unreported tests, we separately examined the performance of momen-
tum portfolios consisting of high-priced and low-priced stocks over different
horizons. Our analysis was motivated by the fact that past losers tend to be
priced lower than past winners. The results for both high-price and low-
price subsamples were quite similar to the results in Tables VII and VIII for
large and small firm subsamples. Specifically, both subsamples exhibit mo-
mentum profits over the 12-month holding period and similar patterns of
reversals over the following four years.
718 The Journal of Finance

Table VII
Long Horizon Momentum Profits for Large Firms
This table presents the momentum portfolio (winners minus losers) returns for large firms. The
large firm sample in this table comprises all stocks traded on the NYSE, AMEX, or Nasdaq
with market capitalizations larger than the median market capitalization of NYSE stocks. All
stocks priced less than $5 at the beginning of the holding period are excluded from the sample.
Panel A presents average monthly raw returns and Panel B presents abnormal returns adjusted
for Fama-French factors.

Months Months Months Months Months Months

Sample Period 1 to 12 13 to 24 25 to 36 37 to 48 49 to 60 13 to 60

Panel A. Raw Returns

1965-1998 0.78 -0.21 -0.18 -0.19 -0.25 -0.21

(4.66) (-1.64) (-1.67) (-1.97) (-2.46) (-3.39)
1965-1981 0.82 -0.28 -0.23 -0.45 -0.32 -0.32
(3.15) (-1.49) (-1.52) (-3.37) (-2.32) (-3.85)
1982-1998 0.74 -0.14 -0.14 0.07 -0.18 -0.10
(3.52) (-0.80) (-0.86) (0.50) (-1.21) (-1.08)

Panel B. Fama-French Alpha

1965-1998 0.96 0.05 0.00 -0.11 -0.22 -0.07

(5.78) (0.41) (-0.03) (-1.17) (-2.18) (-1.57)
1965-1981 1.07 0.03 0.00 -0.34 -0.21 -0.13
(4.17) (0.16) (0.03) (-2.53) (-1.52) (-2.23)
1982-1998 0.89 0.07 0.03 0.10 -0.25 -0.01
(4.15) (0.44) (0.17) (0.75) (-1.63) (-0.16)

III. Conclusions
This paper evaluates various explanations for the momentum profits docu-
mented previously by Jegadeesh and Titman (1993). We first document that
the momentum profits in the eight years subsequent to the Jegadeesh and
Titman (1993) sample period are remarkably similar to the profits found in
the earlier time period. This evidence provides some assurance that the mo-
mentum profits are not entirely due to data snooping biases. Moreover, our
results suggest that market participants have not altered their investment
strategies in a way that would eliminate this source of return predictability.
To learn more about the source of momentum profits, we examine the
returns of the momentum portfolios in the postholding period. By examining
the postholding period performance, we address issues that were raised re-
cently by Conrad and Kaul (1998), who argue that momentum profits arise
because of cross-sectional differences in expected returns rather than because
of time-series return patterns, and Barberis et al. (1998), Daniel et al. (1998),
and Hong and Stein (1999), who present behavioral models that suggest that
the postholding period returns of the momentum portfolio should be negative.
 Profitability of Momentum Strategies 719

Table VIII
Long Horizon Momentum Profits for Small Firms
This table presents the momentum portfolio (winners minus losers) returns for small firms.
The small firm sample in this table comprises all stocks traded on the NYSE, AMEX, or Nasdaq
with market capitalizations smaller than the median market capitalization of NYSE stocks. All
stocks priced less than $5 at the beginning of the holding period and stocks in the smallest
market cap decile (NYSE size decile cutoff) are excluded from the sample. Panel A presents
average monthly raw returns and Panel B presents abnormal returns adjusted for Fama-
French factors.

Months Months Months Months Months Months

Sample Period 1 to 12 13 to 24 25 to 36 37 to 48 49 to 60 13 to 60

Panel A. Raw Returns

1965-1998 1.11 -0.31 -0.32 -0.21 -0.33 -0.29

(7.28) (-2.87) (-3.24) (-2.41) (-3.69) (-5.29)
1965-1981 1.07 -0.38 -0.29 -0.47 -0.55 -0.42
(4.33) (-2.33) (-1.93) (-3.58) (-3.91) (-4.77)
1982-1998 1.16 -0.23 -0.35 0.04 -0.10 -0.16
(6.41) (-1.70) (-2.72) (0.36) (-0.97) (-2.50)

Panel B. Fama-French Alpha

1965-1998 1.27 -0.12 -0.17 -0.15 -0.29 -0.18

(8.31) (-1.30) (-1.87) (-1.66) (-3.26) (-4.07)
1965-1981 1.35 -0.10 -0.07 -0.38 -0.44 -0.25
(5.58) (-0.70) (-0.51) (-2.90) (-3.16) (-3.51)
1982-1998 1.22 -0.13 -0.23 0.08 -0.19 -0.12
(6.47) (-0.96) (-1.84) (0.70) (-1.87) (-2.16)

Our evidence suggests that the performance of the momentum portfolio in

the 13 to 60 months following the portfolio formation month is negative.
Although this evidence clearly rejects the Conrad and Kaul (1998) hypoth-
esis, and is consistent with the behavioral models, for a variety of reasons
our evidence in support of the behavioral models should be tempered with
caution. In particular, although our evidence of momentum profits in the
year following the formation period is extremely robust, evidence of negative
postholding period returns tends to depend on the composition of the sam-
ple, the sample period, and, in some instances, whether the postholding pe-
riod returns are risk adjusted. In other words, positive momentum returns
are sometimes associated with postholding period reversals and sometimes
are not, suggesting that the behavioral models provide at best a partial ex-
planation for the momentum anomaly.

REFERENCES
Asness, Clifford, 1997, The interaction of value and momentum strategies, Financial Analyst
Journal 53, 29-36.
720 The Journal of Finance

Banz, Rolf R., 1981, The relation between return and market value of common stocks, Journal
of Financial Economics 9, 3-18.
Barberis, Nicholas, Andrei Shleifer, and Robert Vishny, 1998, A model of investor sentiment,
Journal of Financial Economics 49, 307-343.
Chan, Louis K., Narasimhan Jegadeesh, and Josef Lakonishok, 1996, Momentum strategies,
Journal of Finance 51, 1681-1713.
Chui, Andy, Sheridan Titman, and K. C. John Wei, 2000, Momentum, ownership structure, and
financial crises: An analysis of Asian stock markets. Working paper, University of Texas at
Austin.
Conrad, Jennifer, and Gautam Kaul, 1993, Long-term overreaction or biases in computed re-
turns? Journal of Finance 48, 39-63.
Conrad, Jennifer, and Gautam Kaul, 1998, An anatomy of trading strategies, Review of Finan-
cial Studies 11, 489-519.
Daniel, Kent, David Hirshleifer, and Avanidhar Subrahmanyam, 1998, Investor psychology and
security market under- and overreactions, Journal of Finance 53, 1839-1886.
Davis, James L., Eugene F. Fama, and Kenneth R. French, 2000, Characteristics, covariances
and average returns, Journal of Finance 55, 389-406.
DeBondt, Werner F. M., and Richard H. Thaler, 1985, Does the stock market overreact? Journal
of Finance 40, 793-805.
Edwards, W., 1968, Conservatism in human information processing, in B. Kleimutz, ed.: Rep-
resentation of Human Judgement (John Wiley and Sons, New York).
Fama, Eugene F., and Kenneth R. French, 1993, Common risk factors in the returns on stocks
and bonds, Journal of Financial Economics 33, 3-56.
Fama, Eugene F., and Kenneth R. French, 1996, Multifactor explanations of asset pricing anom-
alies, Journal of Financial Economics 51, 55-84.
Fama, Eugene F., and Kenneth R. French, 1998, Value versus growth: The international evi-
dence, Journal of Finance 53, 1975-1999.
Grundy, Bruce D., and Spencer J. Martin, 2000, Understanding the nature of risks and the
sources of rewards to momentum investing, Review of Financial Studies, forthcoming.
Haugen, Robert A., 1999, The New Finance (Prentice Hall, Upper Saddle River, NJ).
Hong, Harrison, Terence Lim, and Jeremy C. Stein, 2000, Bad news travels slowly: Size, ana-
lyst coverage, and the profitability of momentum strategies, Journal of Finance 55, 265-295.
Hong, Harrison, and Jeremy C. Stein, 1999, A unified theory of underreaction, momentum
trading, and overreaction in asset markets, Journal of Finance 54, 2143-2184.
Jegadeesh, Narasimhan, and Sheridan Titman, 1993, Returns to buying winners and selling
losers: Implications for stock market efficiency, Journal of Finance 48, 65-91.
Jegadeesh, Narasimhan, and Sheridan Titman, 2000, Cross-sectional and time-series determi-
nants of momentum profits, Working paper, University of Illinois.
Lee, Charles, and Bhaskaran Swaminathan, 2000, Price momentum and trading volume, Jour-
nal of Finance, forthcoming.
Lo, Andrew W., and A. Craig MacKinlay, 1990, When are contrarian profits due to stock market-
overreaction? Review of Financial Studies 3, 175-208.
Loughran, Tim, and Jay R. Ritter, 1995, The new issues puzzle, Journal of Finance 50, 23-51.
Moskowitz, Tobias J., and Mark Grinblatt, 1999, Do industries explain momentum? Journal of
Finance 54, 1249-1290.
Rouwenhorst, K. Geert, 1998, International momentum strategies, Journal of Finance 53, 267-284.
Spiess, D. Katherine, and John Affleck-Graves, 1995, Underperformance in long-run stock re-
turns following seasoned equity offerings, Journal of Financial Economics 38, 243-267.
Tversky, Amos, and Daniel Kahneman, 1974, Judgement under uncertainty: Heuristics and
Biases, Science 185, 1124-1131.