A hundred recorded trades can be a useful review milestone. It is not a statistical certificate, and reaching that count is never a reason to keep risking money after a loss limit or execution problem appears.

The questions are what you are estimating, how variable the outcomes are, and whether the trades represent the conditions in which you intend to use the strategy.

What 100 Wins and Losses Can Tell You

Suppose a fixed strategy records 55 winners from 100 trades. The observed win rate is 55%, not proof that the underlying win probability is 55%. Under a simplified model of independent trades with a constant win probability, a 95% Wilson interval is approximately 45.2% to 64.4%. The NIST confidence-interval guide explains why a proportion estimate needs an interval rather than a single confident number.

Real trades may violate those assumptions. Ten positions opened together in correlated markets do not provide the same information as ten independent observations. A hundred trades in one unusually strong trend may say little about a sideways market.

Win rate also leaves out win size, loss size and costs. An interval for the win rate is not an interval for profitability.

Collect a Consistent Record

Write down the entry and exit rules before collecting the sample. Record all eligible trades, including skipped signals, rejected orders and execution errors. Decide whether a scale-in and partial exits count as one trade or several, then keep that convention consistent.

If you change the stop, filter or entry logic, record a new strategy version. Combining the favourable parts of several versions can make an unrepeatable result look convincing. Position-size changes also distort dollar averages; keep both net dollar results and results relative to each trade's planned risk.

Review Before the Milestone When Necessary

Set a money limit and review conditions before any live experiment. Pause for an unexpected execution issue, an account-rule breach, a loss-budget limit or evidence that the test no longer follows its assumptions. Do not treat avoidable losses as required tuition.

Simulation and historical replay can help test the workflow first. They cannot establish live fills, queue position or the effect of trading real money. Keep simulated and live results separately labelled.

Test Whether the Result Depends on a Few Trades

At a scheduled review, calculate net expectancy, profit factor and equity drawdown. Then inspect the distribution:

  • How much of the total came from the largest one or two winners?
  • Were losses clustered in one session or market condition?
  • Does a separate period that was not used to tune the strategy show similar behaviour?
  • Would realistic fees and worse fills reverse the result?

Do not delete outliers simply to improve the numbers. Investigate whether they are data errors or real outcomes the strategy must accommodate.

Frequently Asked Questions

Is 200 trades enough? More relevant observations can reduce uncertainty, but no count removes selection bias, dependence or changes in market behaviour. Specify the question and required precision rather than assuming a larger round number settles it.

Should I take extra trades to reach 100 sooner? No. Trading outside the stated setup changes the experiment. Record eligible signals over time, or use historical data while preserving a separate test period.

Can I combine paper and live trades? Keep separate summaries. Compare them to identify differences in fills, costs and execution decisions; do not present the combined result as a live track record.

Risk Disclaimer: Trading involves substantial risk of loss. Past performance is not indicative of future results. See our full risk disclaimer.