Building a Review Routine

The short answer

Without a review routine, your record of your own trading is a set of memories — and memories about money are edited by their outcomes. The trades you remember vividly are not a representative sample of the trades you took.

A routine replaces that with a dataset you can actually interrogate.

Two reviews, not one

Conflating these is the most common structural mistake.

Execution review — weekly. Did I do what I said I would do? This has a large enough sample every week, because every trade produces an adherence score whether it won or lost.

Method review — quarterly, or every N trades. Does what I said I would do actually work? This needs a much bigger sample, and reviewing it weekly guarantees you will react to noise.

Mixing them produces a specific failure: a bad week triggers a method change, the method changes before it was ever tested, and after a year you have run six half-tested methods and learned nothing about any of them.

The weekly execution review

Twenty minutes, same time each week, whatever the results.

For each trade:

1. Was it a setup in the written plan? Yes / no.
2. Was size within the risk limit? Yes / no.
3. Was the stop placed at the planned invalidation? Yes / no.
4. Was the exit per plan, or discretionary?
5. Conditions: time of day, session, result of the previous trade, your state.

Then aggregate:

• Adherence rate. Percentage of trades that were fully compliant. This is the primary number, above profit and loss.
• Deviations by type. Not just how many, but which rule broke.
• Conditions of deviations. This is where the clusters from why discipline fails become visible.

If adherence is below roughly 90%, the method is not being tested — you are running something other than the plan, and its results tell you nothing about the plan.

The method review

Only once you have both a decent sample and high adherence. Then look at:

• Number of trades — is it enough to say anything?
• Win rate and average win versus average loss
• Expectancy per trade, net of fees
• Maximum drawdown, and whether you tolerated it
• Whether results cluster in particular conditions

Set the minimum sample before you look. Whatever number you choose — 50, 100 — choosing it in advance is what stops you from stopping at the point the data flatters you.

Recording conditions

Most journals record what happened. Few record the circumstances, which is where the actionable information usually lives.

"Trade 47: long, stopped out, −1%" tells you almost nothing. "Trade 47: long, stopped out, −1%, taken 4 minutes after a losing trade, size 1.8% instead of 1%, no setup in plan" tells you exactly what to fix and exactly when it happens.

Keeping it honest

Two safeguards worth building in:

Timestamp analysis before the outcome. Write your reasoning when you enter, not when you review. Hindsight rewrites reasoning quietly and convincingly.

Score adherence before looking at profit and loss. If you know the result first, outcome bias will colour the adherence score. Fill the process column, then reveal the money column.

The point

Reviewing does not create an edge. What it does is make your feedback loop honest, so that when something is wrong you find out from evidence rather than from a drawdown deep enough to force the issue.

Most traders who improve did not find a better indicator. They built a loop that told them the truth faster.