The mistake: reviewing on a calendar
The standard advice is to review daily. It sounds disciplined and it produces two predictable failures.
Conclusions from noise. A day is five to twenty trades. At that size everything looks like a pattern — three losses on one instrument, two good trades in the morning — and none of it survives contact with a larger sample. A trader who adjusts after every session is adjusting to randomness, continuously.
Review as self-punishment. A daily post-mortem after a losing day is rarely analysis. It is rumination with a spreadsheet open, and it makes the next session worse rather than better.
The fix is to match the question to the sample size it needs, and to ask each question only when the data exists to answer it.
Three intervals, three different questions
After each session — process only, five minutes.
Not results. One question: did I do what I said I would? Note the trades that broke a rule, and the ones you did not take that you should have. No conclusions about the method, because there is no sample.
Weekly — behaviour, twenty minutes.
Now there are enough days to see whether a pattern exists: were the busy days worse, did entries speed up after losses, did size drift. This is where the discipline check and the behavioural analyzers belong, because they need days rather than trades.
Still not a place for method changes.
Monthly or per hundred trades — the method, an hour.
Which setups made money, which lost, is the edge distinguishable from luck at your own variability. This is the only interval where changing something is defensible, and even here the honest precondition is a sample large enough to support the change — the number for your own history comes from luck or skill.
What a review is for
Not to find out whether you made money — you already know that. The three things worth extracting:
Did the process hold? Independent of results. A losing week with a followed plan is a better week than a winning one without.
What is the most expensive repeated behaviour? Not every flaw. The one with the largest number attached — that is what the analyzers are for.
What is the single change for next month? One. A review producing five changes produces none, because nothing can be attributed afterwards.
The trap of reviewing only losses
Losing trades feel like the material. They are half of it, and often the less useful half.
A winning trade taken outside your rules is a more dangerous event than a losing trade taken inside them, because it rewards the wrong behaviour and you will repeat it. Reviews that only examine losses systematically miss the trades that are teaching you bad habits at a profit.
The same applies to size: your best trade of the month deserves the same question as your worst — was it the plan working, or was it the plan being ignored in a direction that happened to pay.
During a drawdown
Reviewing more often during a losing run is the natural instinct and it is the wrong one. The sample has not grown; only the anxiety has.
Keep the interval, reduce the size, and let the record accumulate. If the run is longer than your win rate ordinarily produces, that is a specific arithmetic question with an answer — see losing streaks — and it is not answered by looking at the same twenty trades again.