The Journal of Behavioral Finance
Vol. 5, No. 3, 148-153
Copyright © 2004 by
The Institute of Psychology and Markets
Worse Than Chance? Performance and Confidence Among
Professionals and Laypeople in the Stock Market
Gustaf Torngren and Henry Montgomery
stock market professionals (Nj
21) and laypeople (Nj
34) provided thirty-day forecasts for
stocks and estimated
size of their
own errors as well as their own and the other group's mean errors. Both groups pre-
dicted that the errors made
professionals would be half the size of the errors made
of both groups were about
size predicted for the
laypeople. Participants also estimated their ability to pick the best performing stock
from two options. Both groups proved to be overconfident. Professional predictions
were only successful 40% of the time, a performance below what could be expected
from chance alone. Self
and correlations between forecasts and price move-
ments suggested that the professionals based their predictions on specific information
of the stocks without sufficient awareness of the unreliability of this information,
while the laypeople used simple heuristics based on previous price movements.
During the recent worldwide stock market turbu-
lence, fortunes were rapidly made and just as rapidly
lost. The great majority of participants did not antici-
pate either the magnitude or the pace of these events.
The bewilderment has been hard to hide.
This paper addresses two questions raised by more
than a few people in the aftermath: "How much do
stock market professionals really know about the fu-
ture development of stock prices, and how much do
they think they know?" We first discuss psychological
findings concerning expert judgments and experts' and
laypeople's confidence in their own judgments.
People often turn to experts for guidance in coping
with complex and uncertain decision environments, es-
pecially those that could affect their standard of living
such as financial decision making. Although theoreti-
cal and empirical questions have been raised about the
efficacy of predicting fluctuations in the stock market,
a small but generously compensated industry is en-
gaged in providing investors with analysis and advice
aimed at differentiating winners from losers. This as-
sumes that an expert's knowledge helps investors make
better judgments and more accurate predictions, result-
ing in greater return on investment.
Gustaf Torngren is a Ph.D. candidate in psychology at Stock-
Henry Montgomery is a professor of cognitive psychology at
Requests for reprinst should be sent
Gustaf Torngren, Depart-
ment of Psychology, Stockholm University, S-106 91 Stockholm,
Sweden. Email: firstname.lastname@example.org
Research on expertise in general, however, does not
fully confirm this assumption. A simple linear model
can often provide more accurate predictions than ex-
perts are able to (Grove and Meehl ). Moreover,
simple and basic training can provide non-experts with
sufficient knowledge to perform at the same level as
experts (Camerer and Johnson ).
On the other hand, experts use information more
efficiently (Shanteau and Stewart ). Because
experts are experienced in recognizing patterns, they
are able to analyze problems without using
cognitively more costly mental calculations. This ap-
parent contradiction has been labeled the "pro-
cess-performance paradox," and the question has
been "how can experts know so much but perform so
(Camerer and Johnson ).
Studies assessing financial analyst predictions of-
ten give a rather gloomy picture of their validity. As
early as the 1930s, Cowles  showed how stocks
recommended by a number of financial services in
the years between 1928 and 1932 were outperformed
by the average common stock. This trend continues
today. De Bondt  analyzed the results of a lon-
gitudinal study (1952-1987), and found that the ex-
perts could have improved their results by using his-
torical means instead of specific evaluations.
Similarly, Malkiel  found that, from 1988 to
the average equity mutual fund in the U.S. had
a 3.3% lower average annual return than the Standard
& Poor's index during the same period.
The degree of confidence people have in the cor-
rectness of their judgments and predictions is cru-
PERFORMANCE AND CONFIDENCE IN THE STOCK MARKET
cially important in judgment and decision processes.
Researchers on human judgment consistently find
discrepancies between individuals' subjective proba-
bility estimates and the relevant objective probabili-
Typically, calibration between the two is poor.
Moreover, participants' actual performance is gener-
ally lower than what they expected. So their confi-
dence level tends to be higher than their actual perfor-
mance would warrant.
This tendency toward overconfidence has been
found in both laboratory and naturalistic settings.
Studies have typically examined people's confidence
in their ability to answer general knowledge questions
Slovic, and Lichtenstein , Keren
), but similar results have also been found in fi-
nancial settings (Lichtenstein and Fischoff ,
Russo and Schoemaker ).
Studies that examine professionals in their deci-
sion-making environment produce a split picture as to
the level of overconfidence. Experienced estate
agents (Northkraft and Neale ), physicians
(Christensen-Szalanski and Bushyhead ,
Detmer, Fryback, and Gassner ), and business
managers making forecasts of such economic indica-
tors as input/output prices (Aukutsionek and Belianin
) have all been shown to make biased probabil-
ity judgments that are poorly calibrated.
Other studies, however, have shown that some ex-
perts provide well calibrated judgments. Bridge play-
ers predicting making contracts (Keren ), expe-
rienced weather forecasters (Murphy and Winkler
), and horse race bettors (Johnson and Bruce
) have all proven their ability to make realistic
predictions about the accuracy of their judgments.
Certain factors have been shown to contribute to
overconfidence. According to Bradley , people
with a high degree of perceived expertise in the area of
a general knowledge question are likely to have unreal-
istically high expectations of the probability of answer-
ing correctly. Davies, Lohse, and Kotterman 
found that more information, even when redundant,
boosted the degree of overconfidence among finance
students predicting stock market fluctuations. Olsen
 found that professional investment managers
tended to overestimate probabilities of outcomes that
were positive to the respondent and to underestimate
undesired outcomes, which we refer to as desirability
bias. Cultural differences (Yates, Lee, and Shinotsuka
) and gender (Estes and Hosseini ) have
also been shown to influence overconfidence levels.
The cause of these effects has been the subject of an
intense debate during the past ten years. Until the early
1990s, cognitive bias was the dominant explanation for
the overconfidence phenomenon. In this account,
miscalibration is a result of cognitive biases that occur
as a byproduct of
heuristics used in the information
processes underlying people's judgments.
More recently, however, proponents of ecological
models have challenged this approach (see Gigerenzer,
Hoffrage, and Kleinbolting , Juslin , and
Bjorkman  for examples). These models view
"fast and frugal" way of using
the structure of information in its own decision environ-
ment to make reasonable judgments and decisions un-
der realistic conditions. An example of a "fast and fru-
heuristic is recognition, which implies the
inference of higher values to objects that are recognized
Tested as a stock-picking de-
it was actually proven quite "frugal." Portfolios
based solely on company recognition by foreign
laypeople actually outperformed professionally man-
aged funds and the market index (Gigerenzer et al.
). However, in a replication of this study con-
ducted during a period of bearish market behavior
(rather than the bull market that prevailed during the
the results were reversed (Boyd
Despite the criticism by proponents of the ecologi-
cal approach, the concept of overconfidence is still on
the judgment and decision making agenda. Recent
studies in naturalistic settings (Aukutsionek and
Belianin ), as well as experimental studies de-
signed to avoid the problems of underrepresentation,
have still reported overconfidence (Brenner et al.
Griffin and Tversky ).
This article continues the small volume of research
on perceived and actual accuracy in predictions of
stock prices (see, e.g., Davies, Lohse, and Kotterman
Olsen , and De Bondt ). Contrary
to previous research, we compared professional judg-
ments with laypeople judgments, and we asked our
participants to estimate their own group's accuracy as
well as the other group's accuracy. Although many
studies suggest that professionals do not outperform
laypeople who use simple strategies (Grove and Meehl
), there seems to be a lack of research on how
these two groups perceive their own accuracy and the
accuracy of the other group.
It seems reasonable that the market would work
more efficiently if laypeople and professionals could
judge each other's predictive abilities more accurately.
Overconfidence can cause excessive trading, which
can be risky to financial well being (Barber and Odean
). To better understand the background of the
two groups' performance with respect to their actual
and perceived accuracy, we explored how each group
makes predictions. Professionals often use information
that (more or less erroneously) supports their identity
as experts, such as specific information about stocks.
We conducted two studies: a thirty-day study that
ended on March 10, 2001, and a replication that took
place between March
and April 7,2002. All
ipants in both studies submitted their contributions on
the same day, which was vital because stock prices
change continuously. Study
was based on twenty-two
stock market professionals and twenty-nine laypeople.
Study 2 was based on twenty-one stock market profes-
sionals and thirty-four laypeople.
"Stock market professionals"
that regularly and on a professional basis engage in the
evaluation of and/or investment in
portfolio managers, analysts, brokers, and investment
counselors. One person at each company was contacted
by the author, informed about the study, and asked to
find colleagues with working duties who fulfilled the
criteria. With one exception, all the professionals were
men, and they had a mean experience of twelve years.
The laypeople were undergraduate students in psy-
chology at Stockholm University. In Study 1, six par-
ticipants were male and twenty-three were female. In
Study 2, ten participants were male and twenty-four
were female. It is likely that most of the professionals
also took part in Study 2 (uninformed of the
results of Study 1), while the laypeople were not the
same in the two studies.
The questionnaire used twenty stocks from
well-known blue-chip companies (Ericsson, Volvo,
etc.) and from a broad spectrum of industries listed on
the Stockholm stock
Participants were given
the name, industry, and monthly percent price change
for each stock for the previous twelve months and
asked to predict the rate of change for the share price
over the next thirty days. They also picked one of two
stocks shown on each page that they expected to per-
form better during the same period, and they estimated
the probability (between 50%-100%) that they would
choose the best performing stock. Participants rated on
a ten-point scale the extent to which they used each of
four strategies (previous monthly results, other knowl-
intuition, or guessing), with "not at all" and "to a
very high degree" as extremes. Finally, participants es-
timated the mean errors of their own predictions (in
percent), as well as the collective mean errors of the
professionals and laypeople.
We calculated mean values for both groups to illus-
trate the participants' own expectations and their expec-
tations for the two
shows significant dif-
ferences between the professionals' and the laypeople's
estimates of their mean errors: t (43) = 3.22, p < 0.05
(Study 1); t (52) = 4.39, p
0.05 (Study 2). Thus, the
stock market professionals had higher expectations than
the laypeople about their ability to make accurate pre-
Clearly these expectations
The differences in performance, displayed in the actual
mean errors, were small but marginally significant in
Study 1, / (46) = 2.29, p < 0.05. Interestingly, both
groups shared the erroneous belief that the professionals
would produce better results.
also shows a good consistency between the
estimations of participants' own performance and the
estimated performance of other group members. Fur-
thermore, both professionals and laypeople seem to
with the expected performances of the two groups.
The predictions made by stock market professionals
were negatively correlated to the actual outcomes in
both studies. The correlations, -0.17 in Study 1 and
were significant: r(20)=-2.75,p<0.5,
For laypeople, no correlation
was noted. Respondents as a whole were too optimistic
when predicting stock
the average es-
twenty stocks was
by the profes-
during the thirty days. In Study 2, the estima-
laypeople. The actual outcome was
In estimating their ability to pick the winner be-
tween two stocks, the participants were again too opti-
mistic. The laypeople were moderately overconfident,
predicting a 58% rate of correctness in Study 1 and
in Study 2. The actual rate was closer to 50%,
which could be the result of chance alone. But the pro-
fessionals displayed an even higher degree of overcon-
fidence, rating their probabilities of success at 63% in
Study 1 and 67% in Study 2.
Remarkably, as Table 2 shows, the performance of
stock market professionals is repeatedly worse than the
performance of laypeople: t (44) =
p < 0.10
(Study 1); t (52) =
p < 0.05 (Study 2). In addi-
tion, their performance was below the 50% expected
from chance alone: t (20) = -2.63, p< 0.05 (Study 1); t
(20) = -2.90, p < 0.05 (Study 2). The best calibration
occurred when participants estimated their probabili-
ties of having chosen the best performing stock to be in
line with chance. Higher confidence judgments did not
reflect more correct predictions for either the profes-
sionals or for the laypeople.
Finally, participants were asked to rate the extent to
which they used four strategies
as Table 3
that they used strategies based on expert knowledge.
The laypeople reported that they were mainly guessing:
tween professionals and laypeople). They also reported
previous results for
t (49) =
0.05 in Study 1; / (53) =
p < 0.05 in Study 2.
The professionals, however, considered knowledge
(other than monthly performance over the last twelve
months) as the most important influence when making
PERFORMANCE AND CONFIDENCE IN THE STOCK MARKET
Table 1. Estimated and Actual Mean Errors in Percent for Stock Price Predictions
Study 1 Study 2
Professionals Laypeople Professionals
Estimated error in own predictions 8.67
Estimated error of professionals 4.3
Estimated error of laypeople 8
Actual error 10.84
Table 2. Calibration Data on Probability Estimations of Picking the Better Performing Stock Out of Two
Study 1 Study 2
Professionals Laypeople Professionals
Mean prob. rating
Note: Figures in parentheses denote the number of estimations behind the mean in each cell.
Table 3. Reported Use of Judgment Strategies
Previous month's results
Note: Values represent ratings on a scale from 1-10, where
= "not at all," and 10 = "to a very high degree."
0.05 in Study 2.
Further analysis of the forecasts showed that the
laypeople were influenced by the historical price
movements of the stocks. More specifically, the results
of the past month apparently served as an anchor for
their extrapolations. This was shown as a positive cor-
relation between predictions and stock results for the
past month across stocks in both studies: The mean
correlations were 0.18 and 0.28, respectively. Both
mean correlations differed significantly from zero: t
p < 0.05 and t (33) = 4.72, p < 0.05. This
was not true for the professionals, and the differences
were significant: t (50) = -2.07, p < 0.05 in Study 1; t
(54) = -2.77, p
0.05 in Study 2.
Our results indicate clear differences between
stock market professionals and laypeople. However,
the two groups did not differ clearly with respect to
forecast accuracy. Both groups expected the profes-
sionals to outperform the laypeople by a large
margin, but the professionals made as many errors as
the laypeople. When participants estimated their
probabilities of success in picking a winner between
two stocks, the tendency was the same. Both groups
were overconfident, but the professionals overesti-
mated their ability by a greater margin.
In addition to being more overconfident than
laypeople, the professionals' performance was
icantly worse than the 50% correctness rate expected
from chance alone. Furthermore, there was no appar-
ent connection between certainty of having picked
the right stock and actual success for either group.
In expressing their perceived use of different judg-
ment strategies, the two groups differed. Professionals
mainly relied on knowledge, while the laypeople based
their judgments primarily on chance and on the previ-
ous monthly results.
It seems clear that the benefits of experience
and knowledge, which professionals ostensibly
have greater access to, are overrated. Our results
TORNGREN AND MONTGOMERY
suggest that professionals do not have advantages
over laypeople. It is possible, however, that the
judgments participants were required to make here
did not fully correspond to real-life investment de-
cisions. This might help explain the exaggerated
expectations laypeople had for professionals, but
surely the professionals should have been able to
ascertain whether their backgrounds would help
them gain greater accuracy.
Our results agree with prior research that exper-
tise (Bradley ) and access to information
(Davies, Lohse, and Kotterman ) can result
in overconfidence. The results show that profes-
sionals perceived their judgments as having been
based mainly on "knowledge." It seems reasonable
that basing predictions on "knowledge" would gen-
erate higher levels of confidence than "guessing,"
the primary strategy used by laypeople.
Another factor previously shown to contribute to
professional investor overconfidence is the desirability
bias (Olsen ). For stock market professionals,
bull markets are desirable because they offer the possi-
bility of receiving higher bonuses, and research has
suggested that optimism can be career-promoting
(Hong and Kubik ). These factors contribute to
the recognized surplus of recommendations to "buy"
versus "sell" (Carleton, Chen, and Steiner ).
One of our most intriguing findings is the low level
of correctness when professionals were required to
pick a winner from two stocks. The results suggest that
a chimpanzee, a symbol of randomness in economic
contexts, could have outperformed the stock market
professionals in this study. And chimps do not possess
cognitive advantages. On the contrary, the repeated and
significant deviations from the 50% level of correct-
ness predicted by chance as well as the significant
(negative) correlations between the professionals' pre-
dictions and the outcomes suggest that the profession-
als did use some common judgment strategies.
We attempted to determine whether the profession-
predictions were correlated with the volatility of
the stocks or other industry variables, but no clear-cut
evidence was found.
One possible explanation for our results (which was
not ruled out by our study) is that the high expectations
on the performance of the professional could have been
fulfilled under different circumstances, as both studies
were carried out under bear market conditions (falling
It would be interesting to examine the re-
sults under bull market conditions.
In summary, despite the ecologically realistic set-
both groups displayed overconfidence, and it
was most evident among the stock market profession-
The results suggest that the information-based pre-
dictions of the professionals did not outperform the
simple heuristics used by laypeople.
This research has received financial support from
Aukutsionek, S.P., and A.V. Belianin. "Quality of Forecasts and
Business Performance: Study of Russian
Economic Psychology, 22, (2001), pp. 661-^92.
Barber, B and Odean, T (2000), Trading Is Hazardous To Your
Wealth: The Common Stock Investment Performance of Indi-
vidual Investors, Journal of Finance, 55, (2000), pp.773-806.
Bjorkman, M. "Internal Cue Theory: Calibration and Resolution of
Confidence in General Knowledge." Organizational Behavior
and Human Decision Processes, 58, (1994), pp. 386-405.
Boyd, M. "On Ignorance, Intuition, and Investing: A Bear Market
Test of the Recognition Heuristic." Journal of Psychology and
Financial Markets 2, (2001), pp. 150-156.
Bradley, J.V. "Overconfidence in Ignorant Experts." Bulletin of
Psychonomic Society, 17, (1981), pp. 82-84.
Brenner, L., D. Koehler, V. Liberman, and A. Tversky. "Overconfi-
dence in Probability and Frequency Judgments: A Critical Ex-
amination." Organizational Behavior and Human Decision
Processes, 65, (1995), pp. 212-219.
Camerer, C.F., and E. Johnson. "The Process-Performance Paradox
How Can Experts Know So Much and Pre-
dict So Badly?" In K.A. Ericsson and J. Smith, eds.. Toward a
General Theory of Expertise: Prospects and Limits. Cam-
bridge: Cambridge University Press, (1991), pp. 195-217.
Carleton, W.T, C.R. Chen, and T.L. Steiner. "Optimism Biases
Among Brokerage and Non-Brokerage Firms' Equity Recom-
mendations: Agency Costs in the Investment Industry." Finan-
cial Management, 27, (1998), pp. 17-30.
Christensen-Szalanski, J.J.J., and J.B. Bushyhead. "Physicians' Use
of Probabilistic Information in a Real Clinical Setting." Journal
of Experimental Psychology: Human Perception and Perfor-
mance, 7, (1981), pp. 928-935.
"Can Stock Market Forecasters
Davies, F.D., G.L. Lohse, and J.E. Kotterman. "Harmful Effects of
Seemingly Helpful Informationon Forecasts of StockEamings."
Journal of Economic Psychology, 15, (1994), pp. 253-267.
Economists Know About
ket?'Journal of Portfolio Management, 17,(1991),pp. 84-92.
Detmer, D.E., D.G. Fryback, and K. Gassner. "Heuristics and Biases
in Medical Decision Making." Journal of Medical Education,
(1978), pp. 682-683.
Dunning, D., D. Griffin, J. Milojkovic, and L. Ross. "The Overconfi-
dence Effect in Social Prediction." Journal of
Social Psychology, 58, (1990), pp.
R., and J. Hosseini. "The Gender Gap on Wall Street: An
Empirical Analysis of Confidence in Investment Decision
Making." Journal of Psychology, 122, (1988), pp. 577-590.
B., P. Slovic, and S. Lichtenstein. "Knowing With Cer-
tainty: The Appropriateness of Extreme Confidence." Journal
of Experimental Psychology: Human Perception and Perfor-
mance, 3, ('1977), pp. 552-564.
Gigerenzer, G., B. Borges, D. Goldstein, and A. Ortman. "Can Ig-
norance Beat the Stock Market?" In G. Gigerenzer, P.M.
Todd, and the ABC Research Group, eds.. Simple Heuristics
that Make us Smart. Oxford: Oxford University Press, 1999.
Gigerenzer, G., U. Hoffrage, and H. Kleinbolting. "Probabilistic
Mental Models: A Brunswickian Theory of Confidence." Psy-
98, (1991), pp. 506-528.
PERFORMANCE AND CONFIDENCE IN THE STOCK MARKET
Gigerenzer, G,, P,M, Todd, and the ABC Research Group, eds. Sim-
Smart. Oxford: Oxford University
Griffin, D,, and A, Tversky, "The Weighing of Evidence and the De-
terminants of Confidence," Cognitive Psychology, 24, (1992),
Grove, W,M,, and P,E, Meehl, "Comparative Efficiency of Infor-
mal (Subjective, Impressionistic) and Formal (Mechanical,
Algorithmic) Prediction Procedures: The Clinical-Statistical
Controversy," Psychology, Public Policy and Law, 2, (1996),
Hong, H,, and J,D, Kubik, "Analyzing the Analysts: Career Con-
cerns and Biased Earnings Forecasts." Journal of Finance, 58,
Johnson, J,, and A, Bruce, "Calibration of Subjective Probability
Judgments in a Naturalistic Setting," Organizational Behavior
and Human Decision Processes, 85, (2001), pp, 265-290,
Juslin, P, "The Overconfidence Phenomena as a Consequence of In-
formal Experimenter Guided Selection of Almanac Items," Or-
ganizational Behavior and Human Decision Processes, 57,
Keren, G, "On the Ability of Monitoring Non-Veridical Perceptions
and Uncertain Knowledge: Some Calibration Studies," Acta
Psychologica, 67, (1988), pp, 95-119,
Keren, G, "On the Calibration of Probability
cal Comments and Alternative Perspectives," Journal of Behav-
ioral Decision Making, 10, (1997), pp, 269-278,
Klein, G, "Naturalistic Decision Making: Where Are We Going?"
In C,E, Zsambok and S,G, Klein, eds,. Naturalistic Decision
Making. Mahwah, NJ: L, Erlbaum Associates, (1997), pp,
Lichtenstein, S,, and B,
"Do Those Who Know More Also
Know More About How Much They Know?" Organizational
Behavior and Human Decision Performance, 20, (1977), pp,
Lichtenstein, S,, B,
Philips, "Calibration of Proba-
bilities: The State of the Art
and A, Tversky, eds,. Judgment under uncertainty: Heuristics
and biases. Cambridge: Cambridge University Press,
Malkiel, B,G, A Random
Street. New York: W,W,
Norton and Company, 1999,
Murphy, A,H,, and R,L, Winkler, "Probability Forecasting in Meteo-
rology," Journal of the American Statistical Association, 79,
Northkraft, G,B,, and M,A, Neale, "Expert, Amateurs and Real Es-
An Anchoring and Adjustment Perspective on Property
Pricing Decisions," Organizational Behavior and Human Deci-
sion Processes, 39, (1987), pp, 84-97,
Olsen, R, "Desirability Bias Among Professional Investment Man-
Journal of Behavioral Decision Making, 10, (1997),
Russo, J,E,, and PJ,H, Schoemaker, "Managing Overconfidence,"
33, (1992), pp, 7-17,
Shanteau, J,, and T, Stewart, "Why Study Expert Decision
Making? Some Historical Perspectives and Comments," Or-
ganizational Behavior and Human Decision Processes, 53,
J,F,, J,-W, Lee, and H, Shinotsuka, "Beliefs about Overcon-
fidence, Including its Cross-National Variation," Organiza-
tional Behavior and Human Decision Processes, 65, (1996),