The Weekly Review Loop

A journal nobody reads is a diary. The review is what converts it into improvement, and it works because it ends in exactly one decision — not a list of resolutions, not a redesigned strategy, one change, tested for a week.

The review agenda

One hour, same time each week, same order. The order matters: process before results, so that the numbers are read against how you traded rather than the other way round.

Step 6 is the point of the other five. A review that ends with "be more patient" has not produced a change — it has produced a mood. A change is something you could hand to somebody else: a rule added, a rule removed, an hour dropped, a setup retired.

Process error or normal loss

The central skill of a review, and the one that stops people either changing a working system or ignoring a real problem. Judge the process, not the outcome.

The top-right cell is the one that does damage, because nothing about it feels like a problem. A rule broken on a trade that won is a rule that will be broken again, and the next time the market will not cooperate. Count those separately and review them as errors, whatever the P/L said.

Sorting the losses

1. Was the setup valid by the written rules? If not, it is a process error regardless of result.
2. Was the size correct? Recompute it from the journal's balance, risk and stop columns.
3. Was the stop where the rules put it, and did it stay there?
4. Was the exit by rule, or by feel? This is where the planned-versus-realised R gap comes from.
5. If all four are clean, it was a normal loss. File it and change nothing.

One change, one week

Change one thing. Not two. The reason is not discipline for its own sake — it is that two simultaneous changes cannot be attributed, so the following week's result tells you nothing about either.

• Write it in the version log from Track 6, with the reason and the evidence.
• State what you expect it to do, in a number: "this should cut the planned-versus-realised gap from 0.4R to under 0.2R".
• Give it enough trades, not enough days. One week of a swing strategy is three trades, which decides nothing.
• Change nothing else while it runs.

Which change to pick

When the review surfaces several candidates, take them in this order:

1. A rule broken repeatedly. Fix the execution before touching the strategy — a strategy you are not following has not been tested.
2. A measurable gap, like exits consistently short of target. These have an obvious intervention and a number to check it against.
3. A condition that keeps losing — an hour, a pair, a market state. Standing aside is usually the cheapest change available.
4. A strategy parameter. Last, and only with 40-plus trades of evidence, and with Track 6's plateau warning in mind.

Almost every review's answer is in the first three. Parameters are the most interesting to change and the least likely to be the problem, which is exactly why the order is written down.

Tracking whether the change worked

A change with no follow-up is indistinguishable from a whim. The next review opens by checking the last one.

1. Keep if the number moved as expected and nothing else got worse.
2. Revert if it did not. A reverted change is information, not failure, and the log keeps it so you do not retry it in four months.
3. Extend if the sample is still too small — most changes need 30-plus trades before the verdict means anything.

What the loop produces

One change a week, each with a reason and a measured verdict, is about forty evaluated changes a year. Most will be reverted. The ones that stay are the ones with evidence behind them, and that accumulation is what a trading edge actually looks like from the inside — not a discovery, but a slow filtering.

It is also the honest answer to how long this takes. There is no version of this loop that produces a finished trader in three months, and any method promising one is selling the absence of the loop rather than a faster version of it.