By Volodymyr Kravchenko Updated
Risk · Stress test · Free
The Turkey Test
A quiet record is not safety. Test the day you have not seen.
A turkey is fed for 1,000 days and gains more confidence every day, until day 1,001. A track record only shows what has happened to you. Test your own trades against a loss you have not had: smaller risk, a real stop on every trade, and no raising size because things feel safe.
The idea
The story in one minute 1,000 good days, then the slaughter.
Each morning the turkey is fed, and its data gets better: a thousand days, not one bad one. Its confidence peaks the day before Thanksgiving, which is also the day of greatest danger. Nothing in the old data could show the new event.
Source of the story: Nassim Nicholas Taleb, The Black Swan (2007), after Bertrand Russell's chicken. Fooled by Randomness (2001) is the earlier book on luck, survivorship and "alternative histories". This page is our own summary.
Where traders meet the turkey High win rate, hidden tail, growing size.
- Averaging down, martingale, no stop, grids: many small wins, then one loss that erases them.
- A quiet backtest: data without a crash, a gap or a news spike is a thousand feeding days.
- Rising size after a hot streak: confidence grows, risk grows, and the first real shock lands on the largest size.
- A funded account: the firm's static daily and max loss turn one gap into the end of the account.
The '100 times' trap Positive expectancy is necessary, not sufficient.
"If I take this trade 100 times, am I in profit?" is the right first question: it is expectancy. The missing part is "and will I still be here for the 100th?". Example (arithmetic): win 1R in 9 trades of 10, lose 8R in 1: expectancy +0.1R. At 2% risk one 8R loss is 16% of the account; at 1% it is 8%. The average is positive and the account can still be gone before the average has time to work.
Stress test
Could your record be hiding a turkey?
Uses the results in R from your Trading Journal on this device, or numbers you paste. Runs in your browser.
Your trades in R
What to do
Eight rules to survive day 1,001 Cap the worst case, not the average.
- Cap the worst outcome. A real stop at the broker on every trade. No averaging down, no martingale.
- Size for the worst case. Risk so that the typical longest losing streak, plus one gap of 2 to 3R, does not touch your max loss.
- Treat news separately. Smaller size or flat around core releases; check your firm's window in the checker.
- Stress the data. Take your worst day, multiply it by two or three, and see whether the account survives.
- Judge by drawdown and the largest loss, not by win rate. Log the biggest loss in R in the journal.
- Do not raise size from comfort. A long calm streak is day 500 of feeding: change size by schedule and numbers, not mood.
- Prefer small, known losses with room to run (see 1:2, 1:3, 1:4 targets) to selling insurance for small gains. It is a style with its own cost (long losing streaks), not a guarantee.
- Ask about other histories. Your result is one path of many. Resample it, as the test above does, before you trust it.
Frequently asked questions
What is the turkey problem?
A turkey is fed every day for 1,000 days and each day raises its confidence that people mean it well. On day 1,001, before Thanksgiving, it is slaughtered. Its data was perfect and its conclusion was wrong. In trading: a long record without a big loss does not show that a big loss cannot come.
Where does the story come from?
Nassim Nicholas Taleb tells it in The Black Swan (2007); the idea goes back to Bertrand Russell's chicken. Taleb's earlier book Fooled by Randomness (2001) covers luck against skill and survivorship. This page is our own summary and teaching tool, not a copy of either book, and not an endorsement of any course.
My expectancy is positive. Why is that not enough?
Expectancy is an average over many repeats, and it assumes you survive to repeat. A strategy that wins 1R nine times in ten and loses 8R once has +0.1R per trade, yet one 8R loss at 2% risk is 16% of the account. The path matters, not only the average.
What does the stress test do?
It takes the results of your own trades in R, draws from them at random 4,000 times to build possible paths, and reports the typical worst drawdown, how often the firm's max loss would have been touched and the longest losing streak. Then it adds one shock loss you choose. It is a simulation of the past, not a forecast.
Is my data sent anywhere?
No. The test runs in your browser and reads the journal on this device, or the numbers you paste. Nothing is uploaded.
This content is provided for educational purposes only and is not investment advice. Trading involves risk, and past performance does not guarantee future results. Affiliate relationships may exist. Always conduct your own independent research before making financial decisions.