Why sample size matters
Every trade adds information about how a strategy behaves. With a very small sample, a handful of unusually good or bad outcomes can dominate the statistics. As the sample grows, metrics such as win rate, average win, average loss and profit factor become more informative.
20 trades: very limited evidence
Twenty trades may be enough to see that a strategy is operating, but usually not enough to draw strong conclusions about its long-term behaviour. One or two outlier trades can materially change the apparent return or win rate.
100 trades: more informative, still context-dependent
A hundred live trades provides a broader sample, but the quality of the evidence still depends on how those trades were generated. If all occurred during one narrow market regime, the record may say little about how the strategy behaves when conditions change.
500+ trades: stronger sample, not a guarantee
Hundreds of trades can make performance statistics more stable and expose more losing sequences. They still cannot guarantee future performance, especially if the strategy changes, market structure changes or many trades are highly correlated.
Trade count and time should be read together
A high-frequency bot can generate hundreds of trades in a few months while a slower strategy may need years. Neither calendar age nor trade count should be used alone. A stronger record combines meaningful duration, sufficient trades and exposure to different market environments.
How long should a trading bot track record be? →
Not all trades are independent
Ten positions opened at nearly the same time on the same market may behave more like one concentrated idea than ten independent observations. Trade count can therefore overstate the effective sample size when positions are strongly correlated.
Which metrics become more useful with a larger sample?
Win rate and profit factor become easier to interpret when they are supported by a substantial number of trades. Drawdown remains essential because a strategy can win frequently while occasionally suffering losses large enough to dominate the result.
Compare return with drawdown →
A practical verification sequence
Start by confirming that the record represents live trading. Then inspect duration, trade count, deposits and withdrawals, maximum observed drawdown, fees and whether leverage or amplification affects the economics of the follower account.
Use the full trading bot verification checklist →
Apply the sample-size question to Sonic AI
For Sonic AI, trade count should be considered alongside the length of the available record, observed drawdown, monthly return distribution and the programme structure. No single metric should carry the assessment.