SAMPLE TRADER REPORT · REAL PUBLIC DATA

A sample trader report, built from real public data

Trader SP1000074 sold winning positions 1.22 times as readily as losing ones: a milder lean than 69% of comparable traders.

This is what the coming trader report looks like, run on one real trader’s history. Your own report would be computed the same way from your own trade export.

• 18 Jul 2016 – 26 Mar 2018
• 46 sell-days
• 29 different stocks
• 107 position-days held

The four counts

Every day this trader held a position, that position was either up (a gain) or down (a loss), and it was either sold that day or kept. That gives four counts. The data for this trader fall in 8 of the 21 calendar months between the first and last date, so the activity came in bursts.

Swipe the table sideways for more columns →

The numbers

PGR 0.5556

Proportion of gains realised. Of all the days a position was up, the share on which it was sold: about 55.6%.

PLR 0.4571

Proportion of losses realised. Of all the days a position was down, the share on which it was sold: about 45.7%.

Ratio 1.22

PGR divided by PLR. Above 1 means winners were sold more readily than losers, a pattern called the disposition effect. Exactly 1 would mean no difference.

Gap 0.0984

PGR minus PLR, with Odean’s (1998) standard error of 0.1026, so t = 0.96. With this few observations, a gap this size could easily be chance (a t below about 2 is usually read that way), so the lean is suggestive rather than clear.

Where this trader stands

Among the 2,262 traders in the same dataset with at least 30 sell-days, this trader’s ratio of 1.22 is higher than that of about 31% of them (the 31st percentile). The middle trader in that group has a ratio of 1.66, and 80% of the group have a ratio above 1.

What this means, and what it doesn’t

• It describes what this trader did in the past: they sold winners more readily than losers. It says nothing about what they, or you, should do next.
• It is not advice. It is not a grade, and it is not a prediction of returns.
• It covers one trader, on one platform, in one market (Chinese A-shares), from 18 Jul 2016 to 26 Mar 2018.
• Selling a winner is not always a mistake, and holding a loser is not always one either. The ratio shows a pattern, not the reason for it.
• The t-value treats every held day as a separate observation. Days of the same position are related, so treat it as a rough guide to how clear the gap is, not a precise test.
• Holding age is not held fixed. Positions are sold less often the longer they are held, so if this trader’s losing positions were older than the winning ones, part of the ratio may be an age pattern rather than a habit. See the study’s robustness update.

How it was computed

• Data: Jin, Li & Zhu (2021), Harvard Dataverse doi:10.7910/DVN/NI6SBJ, file 4053809, CC0. Social-trading portfolios on Xueqiu, 2016–18, pseudonymous IDs.
• How this trader was chosen (rule fixed before any individual result was looked at): among the 2,264 traders with at least 30 sell-days, take the one with the median number of sell-days (46; the lower middle value if the count is even); if several are tied, the lowest ID. 73 traders were tied; SP1000074 has the lowest ID. A sell-day is a date on which the trader closed at least one position.
• Definitions: the same as the flagship study (Odean 1998, as coded in reproduce.py). Each row is one position held on one day; PGR and PLR are shares of those days, so they are daily rates and should only be compared with figures computed the same way.
• Check first: before any per-trader number was used, the same code reproduced the study’s published panel figures: PGR 0.1925, PLR 0.1200, ratio 1.61, averaged over 4,659 traders.
• The script that made this page (samples/sample_trader_report.py) →
• Read the method → · The public script (reproduce.py) →

Every number in this report is printed by sample_trader_report.py, which uses the same definitions as reproduce.py.