One of the most common sources of weak conclusions in a trading journal is ignoring sample size. A strategy with a 100% win rate over three trades does not carry the same evidence as one with 58% over two hundred.
There is no universal magic number
The amount of data required depends on strategy variability, trade frequency, process stability and the precision you need. Avoid absolute rules such as “30 trades are always enough”.
Use minimum thresholds for comparisons
When comparing strategies, timeframes or sessions, define a reasonable minimum number of trades before a group becomes eligible. Smaller groups can still be displayed, but they should carry a clear caution.
Do not look only at the average
The smaller the sample, the more a metric can change when one or two new trades are added. Watch how results stabilize as the history grows and avoid interpreting decimals with more precision than the data supports.
Every filter reduces the sample
Filtering simultaneously by market + strategy + timeframe + direction can create groups of one or two trades. Detail increases while available evidence decreases.
Collect data without constantly changing the rules
If the strategy changes after every small block, different versions of the system become mixed together and it becomes harder to know what you are evaluating. Keep rules sufficiently stable while collecting data.