Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Overconfident Individual Day Traders: Evidence fromTaiwan Futures Market∗
Wei-Yu Kuoa† Tse-Chun Linb‡
aDepartment of International Business, National Chengchi University, TaiwanbSchool of Economics and Finance, University of Hong Kong, Hong Kong
(This version January 16, 2012)
AbstractWe take advantage of a day-trading policy implementation in Taiwan futures market to investigate the
performance of day traders. Since October 2007, investors can commit to be “day traders” by closing theday-trade position on the same day to enjoy 50% deduction for the initial margin. This ex ante natureprovides us a laboratory to explore their trading behavior without potential biases from other trading mo-tivations and disposition effect. The result shows that the 3,470 individual day traders on average make asignificantly loss of 61.5 (26.7) thousand New Taiwan dollars after (before) transaction costs from October2007 to September 2008. This implies that those day traders are not only overconfident in precision of in-formation but also having biased interpretation of information. We also find that trading is only hazardousto the overconfident losers, but not to the winners. Last, the evidence suggests that more experienced in-dividual investors exhibit more aggressive day trading behavior. But these experienced day traders do notlearn their types or gain superior trading skills, which could mitigate the losses due to overconfidence.
JEL Classification: G10, G12, G13, C51.
Keywords: overconfidence, index futures, day traders, individual investors, learning
∗We have benefited from the comments of Xiaohui Gao, Mark Grinblatt, Juhani Linnainmaa, Mark Seasholes,Kam-Ming Wan, Pai Xu, and participants at Paris December 2011 Finance Meeting, 19th SFM Conference, NationalChengchi University and University of Hong Kong. Tse-Chun Lin gratefully acknowledges the research supportfrom the Faculty of Business and Economics at the University of Hong Kong and the Research Grants Council of theHong Kong SAR government. Any remaining errors are ours.
†Tel.: +886-2-2939-8033; fax: +886-2-2938-7699 . E-mail address: [email protected].‡Tel.: +852-2857-8503; fax: +852-2548-1152. E-mail address: [email protected].
1. Introduction
Day traders are both controversial and mysterious. Few studies have been devoted to them mainly
due to the lack of their identity and trading history. Even with comprehensive investor account
data, the day traders can only be identified ex post after tracing their trading record at daily
frequency. However, as pointed out by Barber and Odean (2000), investors could trade for reasons
like consumption needs, tax-motivation, liquidity, portfolio re-balancing, etc. Hence, ideally, we
would like to have an ex ante way of identifying day traders who only trade for profits as it would
provide us a more accurate documentation of their trading activities. This paper takes advantage
of a day-trading policy implementation in Taiwan futures market, which provides us a clear-cut
way to identify day traders ex ante, to study the performance of day traders.
On October 8th 2007, Taiwan Futures Exchange (TAIFEX) implemented a new margin re-
quirement policy that allows investors to indicate their orders as day-trade orders and deposit
only halved required margin. When a day-trade order is executed successfully, the position has to
be closed before the end of the day. Essentially, an investor commits to be literally a “day trader”
ex ante. Unlike the existing literature, which relies on the executed transactions to define day
traders, our day traders reveal themselves by submitting the day-trade orders. Hence, this study
focuses on the day traders identified by the day-trade orders instead of investors who merely buy
and sell the same security on the same day. 1
Our contribution is to take advantage of this new margin rule in the Taiwan futures market to
address the following questions: Are the day traders overconfident in the precision of informa-
tion? Do they have biased interpretation of information? Do they lose more money by trading
more contracts? In other words, is trading hazardous to their wealth? Is there any difference in the
relation between trading frequency and performance among winning and losing day traders? Fi-
nally, do the day traders’ past trading experience and performance affect their day trading behav-
ior? The first two questions are investigated in in Odean (1999) for regular individual investors,
1For the rest of the paper, the day trading and day traders in the Taiwan futures market are referred to the transac-tions executed by day-trade orders and the investors who submit day-trade orders respectively.
1
while the third is addressed in Barber and Odean (2000). However, for reasons to be articulated
below, we believe that it is worth revisiting the questions with our day trader sample. The last
two questions are an insightful extension that provides a more complete picture on the investment
behaviors of day traders.
Several features motivate us to answer these questions by studying the day traders in Taiwan
futures market. First, the day traders have to close their positions before the end of the day, which
leads to an additional liquidation risk. Intuitively, those who engage in day trading are supposed to
be either the most informed or the most confident that they are the informed since they are willing
to take the extra liquidation risk for reduced margin. This unique institutional setup provides us a
sharper angle to revisit the question of investor overconfidence.
Second, the maximum investment horizon for the day traders is just one day by default. There-
fore, the realized end-of-day positions of our day traders are not related to the intra-day returns.
This is an important feature since we can observe the day traders positions without impact from
disposition effect.2 Besides, we do not need to calculate average returns of several holding periods
to accommodate variation of investment horizons among investors as in Odean (1999).
Third, our day traders can easily capitalize their negative information by shorting since there
are no short-sale constraints in the futures market. Jordan and Diltz (2003) point out that the
conventional Wall Street wisdom holds that day traders are profitable when the overall market is
running up. One explanation is the high cost of selling short stocks. However, the cost of shorting
futures is identical to taking a long one. We are thus able to examine the profitability of day
traders for both directional movements of the market.
Finally, our day traders are not trading for consumption or liquidity needs as they cannot get
proceeds from selling short. Instead, they have to deposit the margin. The reason to trade is
also unlikely to be portfolio re-balancing or diversification since they have to close their positions
within a day. Moreover, there is no capital gain tax in Taiwan. We can safely conclude that the
2Barber and Odean (2000) use monthly holding positions and calculate monthly returns to avoid such bias. Bydoing so, they need to assume that trading takes place on the last day of the month and also ignore the intra-monthtrading.
2
main purpose to submit day-trade orders is for leveraging up the position to maximize the trading
profit. Together with previous points, Taiwan futures market provides an ideal environment to
answer the questions raised above.
By using the complete trading record in TAIFEX from October 8 of 2007 to September 30
of 2008, we find that the average net profit for a round-trip trade is −613 Taiwan dollar (one US
dollar is roughly 32 TWD during our sample period) for domestic individual day traders. Ag-
gregating to the investor level, the 3,470 individual day traders make a significantly loss of 61.5
thousand TWD after taking transaction costs into account. The result implies that the individual
day traders are overconfident in the precision of information. Before transaction costs, the indi-
vidual day traders still suffer an average of 26.7 thousand TWD losses. This indicates that those
day traders are not only overconfident but also having biased information. The finding is consis-
tent with Odean (1999) who studies individual traders in the US stock market. For institutional
day traders, the result is similar but statistically insignificant due to a much smaller number of
investors.
We also find that the numbers of short position and long position are similar for the individ-
ual day traders when we differentiate the direction of the round-trip trades. The individual day
traders are not as reluctant to short as US individual investors (see Boehmer and Kelley (2009)).
Interestingly, the average return of short-initiated round-trip trades is higher than which of the
long-initiated round-trip trades for individual investors, with a 160 TWD profit for short positions
and a 676 TWD loss for long positions respectively. It implies that when individual investors are
shorting, they are less overconfident than when they are taking long positions.
In addition, the result shows that the relation between trading frequency and performance is U-
shaped. By performing a quantile regression of profits on contract number for every 5 percentile,
we find that for day traders below the 15th percentile in profit, the more they trade, the more
losses they suffer. However, for day traders above the 60th percentile, the more they trade, they
more profit they make. Collectively, the documented relation between profit and contract number
corroborates that trading is only hazardous to the wealth of overconfident losers, but not to which
3
of the informed or skillful winners. This is consistent with Barber, Lee, Liu, and Odean (2011)
who find cross-sectional differences in day trading skill in Taiwan stock market.
Last, we find that the more experienced individual investors exhibit more aggressive day trad-
ing behavior. However, the profits earned in the past is ony weakly related to the day trading
activity. The higher past profit alone does not contribute to higher day trading activity. The result
also shows that the overconfident day traders do not learn their types from their past performance
as suggested by Gervais and Odean (2001). The trading experience plays a limited role to help
individual investors to earn profit by taking advantage of the extra leverage brought by the new
day trading policy.
The rest of the paper proceeds as follows. Section 2 provides a discussion of the literature
most closely related to our work. Section 3 describes the data on the Taiwan futures market. Sec-
tion 4 outlines our empirical approach, and presents our main results and additional confirmation
exercises. Conclusions and summary of our findings are provided in Section 5.
2. Connection to the literature on day trader behavior and over-
confidence bias
This section provides a brief literature review, and aims to posit our empirical findings in the
context of both day traders and their overconfidence bias.
2.1. Day traders
Harris and Schultz (1998) use Nasdaq’s Small Order Execution System to analyze the perfor-
mance of individual day traders. The data contains around 10,000 round-trip trades over a three
week period. Their individual day traders earn a small average profit, even trading with market
makers. Jordan and Diltz (2003) examine the profitability of a sample of 324 US day traders.
They show that about twice as many day traders lose money as make money. They also document
that around 20 percent of sample day traders were more than marginally profitable. Garvey and
4
Murphy (2005) report consistent intraday trading profits by 15 proprietary stock traders through
96 thousand trades in three months in 2000. Garvey and Murphy (2004) use the same data and
find that those traders earned more than $1.4 million in intraday trading profits.
Instead of studying a small number of US day traders, the two following papers examine a
large sample of day traders by taking advantage of the data availability from two international
markets. Linnainmaa (2005) studies a Finish data set, which contains 22,529 day traders and
spans from January 1995 to November 2002. He shows that individual day traders are reluctant
to close losing day trades, which consequently hurts their performance of portfolios up to -6%
in three months after a holdings change. Barber, Lee, Liu, and Odean (2011) document large
variation in the gross and net returns earned by day traders in Taiwan stock exchange from 1992
to 2006. They find that the spread in returns between top-ranked and bottom-ranked day traders
exceeds 60 basis point per day.
All the studies aforementioned focus on the day traders in the stock market and have a mixed
evidence on the profitability. Hence, we aim to complement the literature by providing new
evidence on the performance of day traders in Taiwan futures market. As far as we know, we are
the first paper exploring the day trading behaviors in the futures market. Besides, as expressed
in the previous section, our day traders are not defined ex post by their round-trip trades on the
same day. Our day traders instead commit to close their positions when they initiate them. The
estimation of their performance is thus free from noises due to other trading needs or behavioral
biases. Therefore, our paper also adds to the existing literature by providing this new perspective.
2.2. Investor overconfidence
There is a plethora of literature on investor overconfidence. We only briefly discuss the most
relevant ones. Odean (1999) and Barber and Odean (1999) use discount brokerage accounts to
demonstrate that overall trading volume in equity markets is excessive. They show that investors
are not only overconfident but also biased in the interpretation of information. Barber and Odean
(2000) show that individual investors who hold common stocks directly pay a tremendous per-
5
formance penalty for active trading. Those that trade most earn an annual return of 11.4 percent,
while the market returns 17.9 percent during 1991 to 1996. They also attribute their findings to
the investor overconfidence.
Barber and Odean (2001) further explore overconfidence among genders and find that men
trade 45 percent more than women. The result shows that trading reduces men’s net returns by
2.65% a year, compared with 1.72% for women. Barber, Lee, Liu, and Odean (2009) analyze a
complete trading history of all stock investors in Taiwan from 1995 to 1999 and find that indi-
vidual investor trading results in systematic and economically large losses. They argue that the
investor overconfidence is one of the major cause of the losses.
Unlike the previous literature, our paper analyzes a specific group of investors i.e., the index
day traders. Our day traders have to close their positions before the end of the day ex ante in
exchange for the halved margin deposit. The commitment to the liquidation constraint intuitively
make the day traders the ideal candidates for studying the overconfident behaviors since they have
the strongest belief to be the informed.
3. Data description
3.1. The Taiwan futures market
The futures contracts of Taiwan Futures Exchange (TAIFEX) are traded by the Electronic Trading
System (ETS) from 8:45 AM to 1:45 PM. The matching rules are price priority and time priority.
The submitted orders are matched on a continuous basis. The four major contracts traded in
the TAIFEX are the Taiwan Stock Exchange Index Futures (hereafter TXF), which is based on
all listed stocks on the Taiwan Stock Exchange, the mini-Taiwan Stock Exchange Index Futures
(hereafter MXF), whose payoff is only one-quarter of the TXF, the Electronic Sector Futures
(hereafter EXF), which is based on the Taiwan Stock Exchange Electronic Sector Index, and the
Finance Sector Futures (hereafter FXF), which is based on the Taiwan Stock Exchange Financial
Sector Index.
6
Our data contains all the trades from October 8 of 2007 to September 30 of 2008. In each
transaction, we observe the investor’s account number, her type (individual or institutional), con-
tract type, and other relevant information. Table 1 shows descriptive statistics of all the transac-
tions in our sample period. The first data trait worth emphasizing is that futures market trading
is dominated by individual investors. The ratio of transactions by individual investors over total
transactions is 71%, which is higher than the one of the Taiwan stock market (63% for 2008 and
72% for 2009). Majority of the transactions are based on the Taiwan Stock Exchange Index, with
55% from TXF and 34% from MXF. The Electronic and Financial Sector Index Futures account
for 5.7% and 5.6% respectively, while all other futures account for less than 1.4%.
3.2. The day trading rules in TAIFEX
On October 8th 2007, TAIFEX implemented a new margin requirement policy to increase the
trading volume of the index futures. For the major four index futures, i.e., TXF, MXF, EXF, and
FXF, the margin requirement is reduced by 50% if an investor indicates her order as a day-trade
order. When a day-trade order is successfully executed, the position has to be closed by the in-
vestor before 1:30 PM, 15 minutes before the market close. Otherwise, the position will be forced
to close by a market order or a limit order that is five ticks within the latest trade. Essentially,
through submitting a day-trade order, an investor commits to be a literally “day trader”. In other
words, the maximum duration of an index futures position that is initiated by a day-trade order is
less than five hours.
In the U.S., the “pattern day traders” are defined as those investors who trade the same stock
four or more times in five business days.3 In this study, index futures investors in Taiwan are
defined as day traders if their positions are established through day-trade orders. Our definition is
thus different from which in the U.S. as we have a clear flag to classify the day traders ex ante.
3Under the rules of NYSE and the Financial Industry Regulatory Authority (FINRA), investors who are deemedpattern day traders must have at least $25,000 in their accounts. See http://www.sec.gov/answers/daytrading.htm formore details.
7
4. The performance of the day traders in Taiwan futures mar-
ket
This section addresses our main research questions related to the index futures day traders: Are
the day traders overconfident in the precision of information? Do they have biased interpretation
of information? Do they lose more money by trading more contracts? In other words, is trading
hazardous to their wealth? Is there any difference in the relation between trading frequency and
performance among winning and losing day traders? Finally, do the day traders’ past trading
experience and performance affect their day trading behavior? We provide evidence on these
questions in turn. Our results are important as they shed light on the possible behavioral biases
for the most aggressive individual traders in the market.
4.1. The definition of day traders and round-trip trade
To analyze the performance of day traders, we first have to filter transactions through day-trade
orders out from those through normal orders. We thus exclude transactions which are not initiated
by the day-trade orders. Second, we need to translate the raw transactions into completed round-
trip trades. Essentially, a round-trip trade is when an initiated position, long or short, is covered.
After the filtration, the profit of a round-trip trade can be easily calculated when an investor’s
net position for an index futures is back to zero again. Our way of calculating the payoff of a
round-trip trade is the same as Jordan and Diltz (2003) and Feng and Seasholes (2005). Finally,
we only include investors who conduct more than 5 day trades during our sample period in the
analysis.
The resulting descriptive statistics of the raw transactions by the day traders are in Table 2.
The transactions from the domestic individual investors and institutional day traders are reported
separately. In total, we have 1.42 million contracts traded by the individual day traders while only
44 thousand contracts traded by the institutional day traders. For domestic individual investors,
8
each transaction contains roughly 1.5 contracts.4 The ratio is much higher for the institutional
day traders. Except for the first month, we have more than one thousand individual day traders in
each month.5 The number stabilizes around 15 hundred per month for the later half of the sample
period. There are only few institutional day traders conducting day trading with average number
of 14 per month. Although the institutional day traders only account only 3% of the day-trade
contracts, it is worth exerting some effort to study whether their performance differs from which
of the individual day traders.
4.2. The poor performance of the aggregate round-trip day trades
Before we dig into the performance of the day traders, looking at the performance of the aggregate
round-trip trades first would offer us a general picture. Table 3 Panel A reports the profit of all the
round-trip trades by individual and institutional day traders. There are 348,000 round-trip trades
by the individual investors. On average, one round-trip day trade suffers a loss of 266 TWD.
After trading fee and transaction tax, the net profit is −613 TWD.6 The profit for institutional day
traders is also negative though insignificant due to a small number of observations. The positive
medians of investors round-trip trades by individual and institutional day traders suggest a skewed
distribution the round-trip trade profit.
Panel B and C report the profits of long-initiated and short-initiated round-trip trades. The
numbers of long position and short position are similar for the individual day traders. It seems
that the individual day traders are not reluctant to short as we anticipated. Several reasons might
explain the phenomenon. First, the cost of initiating a long or short position is symmetric in the
index future market. Second, the daily downside risk is caped at 7% due to price limit in Taiwan
4There is barely any foreign individual day trading in TAIFEX.5The identify of day traders in each month could be largely overlapped over months. In total, we have 3,470
individual day traders.6The transaction tax of Taiwan futures market is 0.8 basis point for initiating or closing one contract. The trading
fee varies among different types of investors and across different brokerage firms. The comprehensive trading feedata is unfortunately unavailable. However, the average trading fee for initiating or closing a contract is around 37.5TWD according to the press release of the Ministry of Finance on 4th of October 2008. We thus apply the fee to allthe transactions we analyze.
9
stock market. Third, the margin call also limits the exposure of the downside risk.
Turning to the performance, we see an intriguingly opposite pattern across the investor types
for the long and short positions. Individual day traders perform worse when they take long posi-
tions. The gross and net profits are −676 and −1,014 TWD respectively. With short positions,
their average payoff is 160 TWD and − 196 TWD after transaction costs. This suggests that when
individual day traders are shorting, they are more cautious and less overconfident than when they
are taking long positions in general.
For institutional day traders, on the contrary, the average payoff for long-initiated round-trip
trade is much higher than which of the short-initiated, with a difference of 12,274 (12,339) TWD
including (excluding) transaction costs. With more transactions in each round-trip trade as shown
in Table 2, the magnitude of gains and losses is larger for institutional day traders. Overall, the
crucial insight to be garnered from Table 3 is the poor average performance of round-trip trades.
It bodes the unpromising performance of the day traders that we analyze in the next section.
4.3. The overconfident individual day traders
With the payoff of every round-trip day trade, we can aggregate trades at investor level over our
sample period to gauge the profitability. Following Jordan and Diltz (2003), Table 4 reports the
distribution of the gross and net profits for day traders with 50 thousand TWD in each bin. Like
previous section, we also present the performance when long-initiated and short-initiated day
trades are aggregated separately. In total, we have 3,470 individual day traders who submit day-
trade orders at least on five trading days during our sample period. Similar to previous section,
the numbers of day traders taking long and short positions are similar. This implies that most of
the day traders trade in both directions.
If day traders are rational, they are supposed to be the most informed or with best information
processing technology since they are willing to bear the risk of forced liquidation at the end of
the day. They believe that the benefit of extra leverage, with 50% reduction of initial margin
by submitting day-trade order, outweighs the risk of premature liquidation. Therefore, following
10
Odean (1999), our first null hypothesis is that the net profit of day traders should be positive.
However, the results in Table 4 clearly reject the null hypothesis. For all individual day trades,
on average, they make a significantly negative profit of 61.5 thousand TWD after transaction
costs. While the finding based on the sample average might be driven by outliers, the significantly
negative median alleviates the concern. Besides, only less than 20% of the individual day traders
make positive net profit. With no other reasons to trade like the tax filing or liquidity demand, our
evidence strongly corroborates that the individual day traders are overconfident.
The second question to ask is that whether the day traders are systematically misinterpreting
information available to them. In other words, we would like to study whether trading decision
based on the very information is wrong. Therefore, the second null hypothesis is that on average
the day traders should have positive gross profits.
Again, the first column in Table 4 shows that the hypothesis is rejected. The day traders suffer
an average of 26.7 thousand TWD losses before transaction costs. The magnitude is smaller for
median but still significant. The ratio of day traders making positive gross profit is only around
29%. Overall, the evidence indicates that the day traders have biased information which leads
to negative gross profits. Meanwhile, consistent with the previous section, we also find that the
long-initiated trades perform worse than short-initiated trades.
The day traders in our sample voluntarily restrict the freedom of liquidation timing to the day
of transaction. The information story cannot reconcile the ex ante nature of day trading and the
poor average performance we document ex post. Therefore, a conclusion to be drawn from Table 4
is that the individual day traders are both overconfident and also biased in their information. The
result is compatible with the findings in Odean (1999) and Barber, Lee, Liu, and Odean (2009)
for regular individual equity traders.
4.4. The overconfident institutional day traders
Now we turn to the performance of the institutional day traders. Table 5 shows that the overall
performance for the 42 traders are also poor. The average net profit of is −89.1 thousand TWD
11
while the average gross profit is 4.2 thousand TWD. The larger difference between the profit
and net profit, compared with individual day traders, is due to the higher number of transactions
per institutional investor. However, due to the small number of observations, both profit and
net profit are not statistically significant. When focus on the direction of trades separately, the
institutional day traders perform better when taking long positions as we found in the previous
section. However, probing further into the distribution, we see that the performance varies widely.
The average performance is mainly driven by few outliers. Besides the small sample, the wide
variation of payoff also contributes to the low statistical power.
4.5. Trading is only hazardous to the overconfident day traders
Having established that the individual day traders are overconfident in Section 4.3, we would like
to further ask whether those day traders lose more by trading more as documented in Barber and
Odean (2000) for US equity stock traders. In our context, essentially, we are interested in whether
day traders suffer more losses by trading more futures contracts.7
Table 6 Panel A is analogous to Figure 1 in Barber and Odean (2000). We sort the total con-
tracts traded into quintile and report the averages within each quintile for profit, profit per contract,
net profit, net profit per contract, and number of round-trip trades. The losses are increasing with
the number of contract traded for the first 4 quintile, from 16.4 thousand TWD to 62.6 thousand
TWD. For the most frequent trading quintile, those day traders manage to make a total profit of
31.2 thousand TWD on average. However, the gain for the most aggressive day traders is all con-
sumed by the transaction costs. The negative net profits are monotonically increasing from 18.3
thousand TWD to 108.2 thousand TWD. The takeaway here is similar to the Barber and Odean
(2000) that trading is hazardous to day traders’ wealth.
The relation between number of contracts traded and the performance is quite different when
we focus on the profit per contract.8 The magnitude of negative profit per contact, both before
7Due to the small number of the institutional day traders, the rest of the study will only focus on the individualday traders.
8If we assume that the margin deposited for each trade remains at the same level for all time, the profit per contract
12
or after the transaction costs, reduces as the number of contracts traded increases. Overall, two
core findings to be learned here. First, the day traders make less loss per contract when they
trade more. Second, the day traders belonging to most frequent trading quintile could make more
money by trading more. To further explore the relation, we thus sort day traders into quintile
by the total profit or average profit instead of the contract number. Table 6 Panel B and Panel C
report the results respectively.
Panel B shows that only the fifth quintile day traders earn significant amount of profit. The
average gross profit within the quintile is 267 thousand TWD, compared with a miserable −324.9
thousand TWD for the first quintile. After the transaction costs, the fifth quintile day traders still
make a 188.5 thousand TWD profit. This is consistent with Barber, Lee, Liu, and Odean (2011)
who find some day traders in Taiwan equity market have superior trading skills. However, the
transaction costs roughly account for 30% of the grass profit. The profit per contract monotoni-
cally increases from a negative −504 TWD to 308 TWD while net profit per contract increases
from a negative 593 TWD to 218 TWD. The similar pattern could be found in the Panel C as well.
One of the most interesting findings for the paper is the pattern of contract number in Panel
B and C. We use Figure 1 and Figure 2 to convey the most essential point. When sorted by total
profit, we see an U-shape of the contract number traded. This implies that the most winners and
and losers both trade a lot. Trading is hazardous to the wealth od losers but not to that of the
winners. Mechanically, investors need to trades more to incur large gains or losses. However, the
U-shape is not due to this simple mechanism. When we sort the day traders by the average profit
per contract, Figure 2 shows that the U-shape largely preserves.
To formally test the relation between profit and number of contracts traded among winners and
losers, we adopt the quantile regression technique. The advantages of using quantile regression
are twofold. First, the test is less affected by the outliers. Second, any quantile can be estimated
with higher statistical power compared with running least square regressions within subgroups.9
Empirically, we run 19 different quantile regressions of the total profit or total net profit on a
serves as a reasonable proxy for the return.9Koenker (2005) provides a detail instruction of quantile regression.
13
constant term and the contract number traded, starting the 5th percentile to 95th percentile.
For brevity, we only show the coefficients in interest and the confidence intervals in Figure 3
and Figure 4 for the total profit and the total net profit respectively. Both figures concur with the
U-shape pattern we found in Figure 1 and Figure 2. For day traders below the 15th percentile
in profit or net profit, the more they trade, the more losses they suffer. However, for day traders
above the 60th percentile, the more they trade, they more profit they can make. For example,
the day traders in the 95th percentile (the most winning day traders) make extra 400 TWD gross
profit by trading one more contact. On the contrary, the day traders in the 5th percentile (the most
losing day traders) make extra 800 TWD gross loss by taking one more position. Collectively, the
documented relation between profit and contract number does not fully corroborate that trading is
hazardous to the investors’ wealth. Instead, trading is only hazardous to the overconfident losers’
wealth, but not to the informed or skillful winners.
4.6. Excessive day trading is related to past trading volume but not profit
In this subsection, we try to answer whether the day trading activity and the profits are related to
the investors’ past trading record. In other words, we would like to address the question that the
overconfident day trading is due to learning as shown in Gervais and Odean (2001).
We first identify 2,641 day traders out of 3,470 with trading record from January 2000 to
September 2007. Panel A of Table 8 reports the performance of individual day traders in this
subsample. Interestingly, the day traders with trading record perform worse than those without
trading record. The average gross (net) profit is −38.4 (−73.4) thousand TWD, which are both
larger than those reported in Table 4. These experienced investors seem to gain no superior skills
from their past trading.
Next, we regress the number of round-trip day trades or total net profits of day traders on
the number of contracts traded, number of trading days, and gross profits earned during January
2000 to September 2007. The number of contracts traded and the number of trading days serve
as proxy for the trading experience while the gross profits help investors to know their “type”
14
in the context of Gervais and Odean (2001). The independent variables are standardized for
the convenience of interpretation and the standard errors are adjusted for heteroscedasticity by
Davidson and MacKinnon (1993).
Table 7 shows that both past number of contracts traded and number of trading days are
strongly correlated with the day-trade activity. One standard deviation increase in the number of
contracts traded during January 2000 to September 2007 leads to 61 more round-trip day trades.
Similarly, one standard deviation increase in the of number of trading days leads to 25 more
round-trip day trades. The more experienced individual investors exhibit more aggressive day
trading behavior. However, the day trading activity is only weakly related to the profits earned in
the past. The higher past profit alone does not contribute to higher day trading activity.
Finally, day trading profit is correlated with number of contracts traded. One standard devia-
tion increase in the past number of contracts traded would lead to 270 thousand TWD day-trade
profit. But the day-trade profit is unrelated to the past trading profits. The overconfident day
traders do not to learn their types from their past performance. It could also be interpreted that
they do not gain trading skills or better information processing technology. Overall, the evidence
suggests that the trading experience plays a limited role to help individual investors to earn profits
by taking advantage of the extra leverage brought by the new day trading policy.
4.7. The complete day-trader sample and the subsample convey qualita-
tively the same results
The analysis in the previous sections requires at least five executed day trades for investors to be
included in the sample. We relax the ad hoc threshold to examine whether our results are robust
to other choices of threshold. We re-do the analysis in Table 4 but only report the main statistics
for brevity. Panel B of Table 8 indicates that the performance of all individual day traders, who at
least conduct one day trade, is as poor as that shown Table 4. On average, they make significantly
losses of 48.8 thousand TWD after transaction costs. Before transaction costs, the day traders
lose 26.7 thousand TWD on average. Panel C reports the results when we increase the threshold
15
to ten day trades. The sample size reduces but it yields qualitatively similar results. The pattern
of day traders losing more in long positions than in short postilions also preserves.
To sum up, the documented overconfidence of the individual futures day traders in Taiwan
appears to be structurally stable and is not an artifact of the sample.
5. Concluding remarks
In this paper, we take advantage of a new day-trading policy implementation in Taiwan futures
market to study the performance of day traders. The policy provides us a clear-cut way to identify
day traders ex ante. Essentially, an investor commits to be a day trader by flagging her orders
as day-trade orders. Those positions need to be closed on the same day. This setup provides us
an ideal laboratory to address the following questions: Are the day traders overconfident in the
precision of information? Do they have biased interpretation of information? Do they lose more
money by trading more contracts? In other words, is trading hazardous to their wealth? Is there
any difference in the relation between trading frequency and performance among winning and
losing day traders? Finally, do the day traders’ past trading experience and performance affect
their day trading behavior?
Several features motivate us to answer these questions by studying the day traders in Taiwan
futures market. First, the investors who voluntarily take liquidation risk are supposed to be either
the most informed or the most confident that they are informed. Second, the maximum investment
horizon for the investors is restricted to one day by default. The day traders’ positions can be
observed without potential bias from disposition effect. Third, the futures day traders can easily
capitalize their negative information by taking short position. Finally, they are unlikely to trade
for consumption, liquidity, portfolio re-balancing, diversification, hedging, and tax-motivation.
The results show that the 3,470 individual day traders make substantially negative gross profits
and net profits. This implies that the individual day traders are not only overconfident in precision
of information but also having biased interpretation of information. The result for institutional
16
day traders is similar but statistically insignificant due to a much smaller number of observations.
We also find that the numbers of short position and long position are similar for the individual
day traders. To our surprise, the short-initiated round-trip trades perform better than the long-
initiated round-trip trades for individual investors. Beside, the result does not fully support that
the investors lose more by trading more. Trading is only hazardous to the overconfident losers, but
not to the winners. Last, we find that the more experienced day traders exhibit more aggressive
day trading behavior. But the overconfident day traders do not learn their types from their past
performance. The trading experience thus plays a limited role to help individual investors to earn
profits by taking advantage of the extra leverage brought by the new day trading policy.
17
References
Barber, B., Lee, Y.-T., Liu, Y.-J., Odean, T., 2009. Just how much do individual investors lose by
trading?. Review of Financial Studies 22, 609–632.
Barber, B., Lee, Y.-T., Liu, Y.-J., Odean, T., 2011. The cross-section of speculator skill evidence
from day trading. Unpublished working paper. University of California at Davis and Peking
University and University of California, Berkeley.
Barber, B., Odean, T., 1999. The courage of misguided convictions: The trading behavior of
individual investors. Financial Analysts Journal pp. 41–55.
Barber, B., Odean, T., 2001. Boys will be boys: Gender, overconfidence, and common stock
investment. Quarterly Journal of Economics 116.
Barber, B. M., Odean, T., 2000. Trading is hazardous to your wealth: The common stock invest-
ment performance of individual investors. Journal of Finance 55, 773–806.
Boehmer, E., Kelley, E., 2009. Institutional investors and the informational efficiency of prices.
Review of Financial Studies 9, 3563–3594.
Davidson, R., MacKinnon, J. G., 1993. Estimation and Inference in Econometrics. Oxford Uni-
versity Press, New York.
Feng, L., Seasholes, M., 2005. Do investor sophistication and trading experience eliminate be-
havioral biases in financial markets?. Review of Finance 9, 305–351.
Garvey, R., Murphy, A., 2004. Are professional traders too slow to realize their losses?. Financial
Analysts Journal 60, 35–43.
Garvey, R., Murphy, A., 2005. Entry, exit and trading profits: A look at the trading strategies of a
proprietary trading team. Journal of Empirical Finance 12, 629–649.
Gervais, S., Odean, T., 2001. Learning to be overconfident. Review of Financial Studies 14, 1–27.
Harris, J., Schultz, P., 1998. The trading profits of soes bandits. Journal of Financial Economics
50, 39–62.
18
Jordan, D., Diltz, D., 2003. The profitability of day traders. Financial Analysts Journal 59, 85–95.
Koenker, R., 2005. Quantile Regression. Cambridge University Press, Cambridge, UK.
Linnainmaa, J., 2005. The individual day trader. Unpublished working paper. University of
Chicago.
Odean, T., 1999. Do investors trade too much?. American Economic Review 89, 1279–1298.
19
Tabl
e1
All
Inde
xFu
ture
sDes
crip
tive
Stat
istic
s
Tota
lIn
vest
orTy
pePr
oduc
tTyp
eTr
ansa
ctio
nsIn
divi
dual
Inst
itutio
nal
TX
FM
XF
EX
FFX
FO
ther
s20
07/1
01,
870,
994
1,38
1,27
648
9,71
81,
078,
166
506,
157
129,
863
141,
018
1579
020
07/1
12,
470,
271
1,77
9,26
069
1,01
11,
454,
938
711,
988
139,
628
139,
699
2401
820
07/1
22,
716,
747
1,80
4,61
991
2,12
81,
525,
523
936,
893
123,
451
113,
037
1784
320
08/0
14,
176,
970
2,54
4,21
81,
632,
752
2,23
5,00
71,
361,
298
298,
008
225,
626
5703
120
08/0
21,
791,
517
1,27
0,27
352
1,24
492
5,13
654
8,78
012
9,29
911
0,56
377
739
2008
/03
3,40
3,47
32,
515,
959
887,
514
1,83
9,39
61,
131,
589
197,
695
201,
394
3339
920
08/0
42,
886,
335
2,17
4,86
971
1,46
61,
510,
854
955,
650
175,
457
205,
264
3911
020
08/0
52,
864,
748
2,09
5,69
276
9,05
61,
522,
378
934,
713
168,
154
189,
498
5000
520
08/0
62,
970,
428
2,12
5,78
784
4,64
11,
607,
198
1,02
7,27
114
6,66
915
4,92
534
365
2008
/07
3,27
2,21
72,
374,
412
897,
805
1,78
5,64
91,
158,
925
137,
089
151,
129
3942
520
08/0
83,
349,
750
2,55
2,30
379
7,44
71,
798,
758
1,20
3,78
217
4,04
113
2,16
841
001
2008
/09
3,64
0,04
92,
628,
370
1,01
1,67
91,
978,
549
1,30
2,01
716
8,38
714
7,84
443
252
Tota
l35
,413
,499
25,2
47,0
3810
,166
,461
19,2
61,5
5211
,779
,063
1,98
7,74
11,
912,
165
472,
978
Ave
rage
2,95
1,12
52,
103,
920
847,
205
1,60
5,12
998
1,58
916
5,64
515
9,34
739
,415
Rat
io10
0.00
0%71
.292
%28
.708
%54
.390
%33
.262
%5.
613%
5.40
0%1.
336%
Not
es.T
able
1re
port
sth
esu
mm
ary
stat
istic
sof
allt
rade
sin
the
Taiw
anfu
ture
sm
arke
tfro
mO
ctob
er8t
h,20
07to
Sept
embe
r30t
h,20
08.
The
num
ber
ofto
tal
tran
sact
ion,
the
num
ber
oftr
ansa
ctio
nsby
dom
estic
indi
vidu
alan
din
stitu
tiona
lda
ytr
ader
s,an
dth
enu
mbe
rof
tran
sact
ions
inth
eTa
iwan
Stoc
kE
xcha
nge
Inde
xFu
ture
s(T
XF)
,the
min
i-Ta
iwan
Stoc
kE
xcha
nge
Inde
xFu
ture
s(M
XF)
,the
Ele
ctro
nic
Sect
orFu
ture
s(E
XF)
,th
eFi
nanc
eSe
ctor
Futu
res
(FX
F),
and
the
othe
rty
pes
offu
ture
sar
ere
port
edre
spec
tivel
y.T
henu
mbe
rof
tran
sact
ions
are
repo
rted
onea
chm
onth
whi
leth
eto
tala
ndav
erag
eov
erth
em
onth
sar
eal
sopr
esen
ted.
20
Tabl
e2
Day
-tra
deIn
dex
Futu
resD
escr
iptiv
eSt
atis
tics
Num
bero
fCon
trac
tsN
umbe
rofT
rans
actio
nsIn
vest
or/D
ayIn
vest
orD
ate
Indi
vidu
alIn
stitu
tiona
lIn
divi
dual
Inst
itutio
nal
Indi
vidu
alIn
stitu
tiona
lIn
divi
dual
Inst
itutio
nal
2007
/10
38,4
541,
180
26,3
5661
82,
907
4074
27
2007
/11
115,
502
5,18
276
,243
3,09
47,
339
8512
5511
2007
/12
99,0
433,
602
69,7
802,
516
7,71
574
1368
1220
08/0
114
9,06
867
895
,680
514
8,12
954
1485
1320
08/0
267
,692
491
49,6
7532
75,
703
5812
9218
2008
/03
140,
022
1,62
396
,775
1,03
89,
218
134
1609
1720
08/0
411
7,43
81,
102
82,3
8380
79,
485
115
1585
1520
08/0
511
4,62
86,
378
80,2
273,
222
9,02
810
515
7315
2008
/06
127,
341
4,02
689
,410
2,35
59,
120
6515
7111
2008
/07
142,
063
1,61
098
,211
1,25
88,
527
6416
4018
2008
/08
150,
233
13,2
8611
0,23
37,
304
10,6
0513
416
7116
2008
/09
160,
646
4,81
211
0,74
63,
733
9,86
810
816
8317
Tota
l1,
422,
130
43,9
7098
5,71
926
,786
97,6
441,
036
17,4
7417
0A
vera
ge11
8,51
13,
664
82,1
432,
232
8,13
786
1,45
614
Max
160,
646
13,2
8611
0,74
67,
304
10,6
0513
41,
683
18M
in38
,454
491
26,3
5632
72,
907
4074
27
Not
es.
Tabl
e2
repo
rts
the
sum
mar
yst
atis
tics
ofth
eda
ytr
ades
for
four
Taiw
anin
dex
futu
res,
incl
udin
gT
FX,M
FX,E
FX,a
ndFX
F,fr
omO
ctob
er8,
2007
toSe
ptem
ber3
0,20
08.T
henu
mbe
rofc
ontr
acts
,num
bero
ftra
nsac
tions
,num
bero
finv
esto
r/da
y,an
dth
enu
mbe
rof
inve
stor
spe
rm
onth
for
dom
estic
indi
vidu
alan
din
stitu
tiona
lday
trad
ers
are
repo
rted
sepa
rate
ly.
The
iden
tify
ofda
ytr
ader
sin
each
mon
thco
uld
bela
rgel
yov
erla
pped
over
mon
ths.
Into
tal,
we
have
3,47
0di
stin
ctin
divi
dual
day
trad
ers
and
42in
stitu
tiona
lday
trad
ers.
21
Table 3 Profit and Net Profit for Round-trip Trades
Panel A: All round-trip tradesIndividual Institutional
(thousands TWD) Profit Net Profit Profit Net ProfitMean ***-0.266 ***-0.613 0.043 -0.905p-value (0.000) (0.000) (0.994) (0.869)Median ***0.300 ***0.189 ***0.400 ***0.284p-value (0.000) (0.000) (0.000) (0.000)Max 2,298 2,065 12,160 12,086Min -5,180 -5,235 -10,464 -10,653number of obs 348,063 348,063 4,134 4,134
Panel B: Long-initiated round-trip tradesIndividual Institutional
(thousands TWD) Profit Net Profit Profit Net ProfitMean ***-0.676 ***-1.014 6.824 5.841p-value (0.000) (0.000) (0.374) (0.444)Median ***0.300 0.148 ***0.450 *0.294p-value (0.000) (0.392) (0.000) (0.051)Max 1,908 1,799 12,160 12,086Min -5,180 -5,235 -2,769 -2,833number of obs 177,415 177,415 1,862 1,862
Panel C: Short-initiated round-trip tradesIndividual Institutional
(thousands TWD) Profit Net Profit Profit Net ProfitMean ***0.160 ***-0.196 -5.515 -6.433p-value (0.005) (0.001) (0.475) (0.409)Median ***0.350 ***0.193 ***0.400 ***0.248p-value (0.000) (0.000) (0.000) (0.000)Max 2,298 2,065 10,385 10,314Min -2,596 -2,689 -10,464 -10,653number of obs 170,648 170,648 2,272 2,272
Notes. Table 3 reports the profit and net profit of all, long-initiated, and short-initiated round-triptrades. Net profit is the gross profit minus the transaction fee and tax. The profit and net profit fordomestic individual day traders and institutional day traders are reported separately.
22
Table 4 Distribution of Profit and Net Profit for Domestic Individual Investors
All Trades Long-intiated Short-intiatedProfit Net Profit Profit Net Profit Profit Net Profit
(thousands TWD)below −500 78 105 44 54 24 28−500 to −450 16 20 7 13 4 4−450 to −400 25 17 13 15 4 7−400 to −350 26 30 21 27 6 10−350 to −300 29 36 29 33 11 13−300 to −250 48 59 37 34 9 14−250 to −200 71 92 48 61 27 31−200 to −150 107 154 90 96 42 56−150 to −100 228 269 186 207 94 115−100 to −50 485 540 381 444 239 288−50 to 0 1361 1473 1,623 1,722 1,479 1,6880 to 50 697 473 792 608 1,093 87250 to 100 133 80 70 56 140 92100 to 150 33 27 28 17 51 32150 to 200 27 23 17 12 19 13200 to 250 19 10 12 12 19 11250 to 300 12 6 6 5 8 9300 to 350 7 7 6 5 5 6350 to 400 9 5 8 1 8 3400 to 450 6 5 1 1 4 5450 to 500 5 2 2 1 5 3above 500 48 37 22 19 35 26
Mean *−26.654 ***−61.470 ***−34.813 ***−52.253 8.229 *−10.040p-value (0.071) (0.000) (0.000) (0.000) (0.281) (0.095)Median ***−18.675 ***−29.996 ***−13.000 ***−19.219 ***−2.950 ***−6.709p-value (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)Max 29,171 23,121 14,547 11,580 14,623 11,541Min −20,991 −23,384 −19,201 −20,982 −2,993 −3,351Observations 3,470 3,470 3,443 3,443 3,326 3,326
Notes. Table 4 reports the total profit and net profit of all, long-initiated, and short-initiated tradesof domestic individual investors. The profit and net profit are reported in the bin of 50 thousandTWD.
23
Table 5 Distribution of Profit and Net Profit for Institutional Day Traders
All Trades Long-intiated Short-intiatedProfit Net Profit Profit Net Profit Profit Net Profit
(thousands TWD)below −500 1 1 0 0 1 1−500 to −450 0 0 0 0 0 0−450 to −400 0 0 0 0 0 0−400 to −350 1 1 0 0 0 0−350 to −300 0 0 0 0 0 0−300 to −250 0 1 0 0 0 0−250 to −200 0 0 0 1 0 0−200 to −150 1 3 3 2 1 1−150 to −100 0 1 0 2 0 1−100 to −50 5 5 4 2 3 3−50 to 0 20 18 20 21 17 190 to 50 7 6 8 9 12 1150 to 100 1 1 2 0 3 1100 to 150 1 0 0 0 0 0150 to 200 0 1 0 1 0 0200 to 250 1 0 0 1 1 2250 to 300 0 0 0 0 0 0300 to 350 0 0 2 0 0 1350 to 400 0 0 0 0 1 0400 to 450 0 1 0 0 1 0450 to 500 0 0 0 0 0 0above 500 4 3 2 2 2 2
Mean 4.218 −89.058 309.921 265.272 −298.324 −348.015p-value (0.988) (0.785) (0.234) (0.265) (0.557) (0.513)Median −11.875 **−15.264 −5.500 *−7.937 −0.850 −4.481p-value (0.100) (0.041) (0.223) (0.054) (0.703) (0.504)Max 5,240 5,069 10,460 9,569 5,156 5,072Min −9,962 −11,956 −187 −204 −20,422 −21,525Observations 42 42 41 41 42 42
Notes. Table 5 reports the total profit and net profit of all, long-initiated, and short-initiated tradesof institutional day traders. The profit and net profit are reported in the bin of 50 thousand TWD.
24
Table 6 Quintile of Contract Number, Profit, Net Profit, Profit per Contract, Net Profit perContract, and Round-trip Trade Number
Panel A: Sorted by number of contract1st Q 2nd Q 3rd Q 4th Q 5th Q
Number of Contract 24.14 59.40 120.02 251.22 1613.61Profit (thousands TWD) -16.43 -33.90 -51.54 -62.58 31.17Profit per Contract (TWD) -680.71 -570.63 -429.38 -249.10 19.32Net Profit (thousands TWD) -18.26 -38.45 -60.63 -81.78 -108.23Net Profit per Contract (TWD) -756.20 -647.35 -505.16 -325.54 -67.07Round-trip Trade Number 10.83 24.25 47.70 94.89 323.87
Panel B: Sorted by total net profit1st Q 2nd Q 3rd Q 4th Q 5th Q
Number of Contract 644.82 236.01 210.05 110.96 866.56Profit (thousands TWD) -324.90 -57.35 -16.56 -1.46 267.01Profit per Contract (TWD) -503.86 -243.02 -78.85 -13.19 308.12Net Profit (thousands TWD) -382.11 -74.81 -30.50 -8.43 188.51Net Profit per Contract (TWD) -592.59 -316.98 -145.21 -75.98 217.54Avg Round-trip Trade Number 148.47 90.36 72.02 47.22 143.46
Panel C: Sorted by net profit per contract1st Q 2nd Q 3rd Q 4th Q 5th Q
Number of Contract 312.22 206.57 261.64 409.50 878.47Profit (thousands TWD) -277.40 -86.84 -37.02 0.35 267.64Profit per Contract (TWD) -888.48 -420.36 -141.49 0.84 304.66Net Profit (thousands TWD) -307.39 -104.40 -57.62 -26.40 188.47Net Profit per Contract (TWD) -984.54 -505.38 -220.24 -64.47 214.54Avg Round-trip Trade Number 43.33 67.42 90.65 151.26 148.87
Notes. Table 6 reports the average values of the total profit, total net profit, profit per contract,net profit per contract, and total round-trip trades in each quintile groups for domestic individualinvestors. Panel A reports the results sorted by the number of total contracts traded by individualday traders during the sample period. Panel B shows the results sorted by the total profit. ThePanel C shows the results sorted by the profit per contract.
25
Tabl
e7
Ove
rcon
fiden
tDay
Trad
ing
isR
elat
edto
Past
Trad
ing
Act
ivity
butn
otPr
ofits
Num
bero
frou
nd-t
rip
trad
esTo
taln
etpr
ofits
Num
bero
fcon
trac
ts**
*61.
0**
*60.
8*2
7033
9.1
*270
355.
6(0
.000
)(0
.000
)(0
.093
)(0
.093
)
Num
bero
ftra
ding
days
***2
5.3
***2
4.9
1072
2.2
9646
.7(0
.004
)(0
.004
)(0
.561
)(0
.611
)
Past
Profi
ts7.
4**
2.9
6.0
1974
2.5
-224
.219
202.
2(0
.683
)(0
.011
)(0
.671
)(0
.886
)(0
.997
)(0
.889
)
Num
bero
fobs
erva
tions
2,64
12,
641
2,64
12,
641
2,64
12,
641
2,64
12,
641
2,64
12,
641
Adj
-R-S
quar
ed4.
93%
0.85
%0.
07%
4.94
%0.
89%
15.3
6%0.
02%
0.08
%15
.36%
0.10
%
Not
es.T
able
7re
port
sth
ere
gres
sion
resu
ltsw
ithnu
mbe
rofr
ound
-tri
pda
ytr
ades
orto
taln
etda
y-tr
ade
profi
tsas
the
depe
nden
tvar
iabl
e.T
hein
depe
nden
tva
riab
les
incl
udes
num
ber
ofco
ntra
cts
trad
ed,
num
ber
oftr
adin
g,an
dda
ysto
tal
gros
spr
ofits
from
2000
/01/
01to
2007
/10/
07.
The
inde
pend
ent
vari
able
sar
est
anda
rdiz
ed.
The
obse
rvat
ions
cont
ains
2,64
1in
divi
dual
day
trad
ers
who
have
trad
ing
reco
rdfr
om20
00/0
1/01
to20
07/1
0/07
.The
stan
dard
erro
ris
adju
sted
forh
eter
osce
dast
icity
byD
avid
son
and
Mac
Kin
non
(199
3).
26
644.8
236.0 210.1111.0
866.6
-503.9
-243.0
-78.9-13.2
308.1
-592.6
-317.0
-145.2-76.0
217.5
Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5
Average Profit and Contract Numbers Sorted by Total
Net Profit
Contract Number Average Profit Average Net Profit
Figure 1 The Average Total Contract Number, Total Profit, and Total Net Profit withineach Quintile sorted by the Total Profit
312.2206.6
261.6
409.5
878.5
-888.5
-420.4
-141.5
0.8
304.7
-984.5
-505.4
-220.2
-64.5
214.5
Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5
Average Profit and Contract Numbers
Sorted by Average Net Profit
Contract Number Average Profit Average Net Profit
Figure 2 The Average Total Contract Number, Total Profit, and Total Net Profit withineach Quintile sorted by the Average Profit per Contract
27
Table 8 Robustness Check: the Distribution of Profit and Net Profit for Domestic IndividualInvestors
Panel A: With Trading RecordAll Trades Long-intiated Short-intiated
Profit Net Profit Profit Net Profit Profit Net ProfitMean **-38.394 ***-73.364 ***-43.558 ***-61.200 5.078 **-13.194p-value (0.012) (0.000) (0.000) (0.000) (0.509) (0.028)Median ***-20.750 ***-32.639 ***-14.875 ***-20.724 ***-3.200 ***-6.998p-value (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)Max 23,718 15,028 11,114 7,229 12,605 8,984Min -20,991 -23,384 -19,201 -20,982 -2,634 -3,182Obs 2,641 2,641 2,622 2,622 2,523 2,523
Panel B: Threshold ZeroAll Trades Long-intiated Short-intiated
Profit Net Profit Profit Net Profit Profit Net ProfitMean **-23.462 ***-48.786 ***-29.032 ***-42.413 4.450 **-10.107p-value (0.026) (0.000) (0.000) (0.000) (0.457) (0.032)Median ***-10.150 ***-16.674 ***-7.600 ***-11.419 ***-2.450 ***-5.050p-value (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)Max 29,171 23,121 14,547 11,580 14,623 11,541Min -20,991 -23,384 -19,201 -20,982 -2,993 -3,351Obs 4,867 4,867 4,585 4,585 4,252 4,252
Panel C: Threshold TenAll Trades Long-intiated Short-intiated
Profit Net Profit Profit Net Profit Profit Net ProfitMean -28.192 ***-71.849 ***-39.332 ***-61.025 11.227 -11.251p-value (0.143) (0.000) (0.001) (0.000) (0.247) (0.140)Median ***-24.950 ***-39.899 ***-18.175 ***-25.857 ***-3.550 ***-9.293p-value (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)Max 29,171 23,121 14,547 11,580 14,623 11,541Min -20,991 -23,384 -19,201 -20,982 -2,993 -3,351Obs 2,653 2,653 2,644 2,644 2,601 2,601
Notes. Table 8 reports the total profit and net profit of all, long-initiated, and short-initiatedtrades of domestic individual investors. Panel A reports the results of all individual day traderswith trading record from 2000/01/01 to 2007/10/07. Panel B reports the results of individual daytraders with at least one day trade. Panel C reports the results of individual day traders with atleast ten day trades.
28
-1200
-1000
-800
-600
-400
-200
0
200
400
600
800
1000
5th 10th 15th 20th 25th 30th 35th 40th 45th 50th 55th 60th 65th 70th 75th 80th 85th 90th 95th
Profit on Contract Number Upper Confidence Interval Lower Confidence Interval
Note:
Figure 3 The Coefficients of Quantile Regression of the Profit on Number of Contractstraded and the 95%Confidence Intervals
-1200
-1000
-800
-600
-400
-200
0
200
400
600
800
5th 10th 15th 20th 25th 30th 35th 40th 45th 50th 55th 60th 65th 70th 75th 80th 85th 90th 95th
Net Profit on Contract Number Upper Confidence Interval Lower Confidence Interval
Figure 4 The Coefficients of Quantile Regression of the Net Profit on Number of Contractstraded and the 95%Confidence Intervals
29