Trade journaling · Intermediate

Spotting Patterns Across Your Journal

Cut the same trades five different ways and see what survives. The cut almost nobody makes is usually the loudest.

Last reviewed 15 September 2026 by the MarketPro research desk · Editorial policy

In short

Cut the same set of trades five ways: by setup tag, by session, by instrument, by market condition, and by where the trade fell in that day's sequence. Compare win rate, average R and total R across the buckets. Treat any finding with fewer than thirty trades behind it as a suggestion, demand a mechanism that explains it, then re-test it on trades it has never seen.

What to remember

Cut by trade number within the day. In one 240-trade log the first two trades made +45.4R and everything after gave back 24.3R.
A finding without a mechanism is usually noise with a label attached to it.
Thirty trades in a bucket is the floor for reading it, and a hundred is where you can lean on it.
Never start from the winners and look for what they share. Pick the property first, then measure both outcomes inside it.

Lesson 10 of 14 in Trade journaling and review

PreviousWeekly And Monthly Review Routines NextJournaling Pre-Trade Checklists

A journal starts revealing tendencies somewhere past a hundred entries, and the way to find them is to cut the same set of trades several different ways and see what survives. Five cuts do most of the work: by setup tag, by session, by instrument, by market condition, and by where the trade fell in that day's sequence. The last one is the cut almost nobody makes, and it is frequently the loudest.

Trade one to trade four: where a session goes wrong

Take 240 trades and group them by whether each was the first, second, third, or a later trade that day.

Trade of the day	Count	Win rate	Avg R	Total R
First	96	56%	+0.34	+32.6
Second	71	52%	+0.18	+12.8
Third	41	44%	-0.21	-8.6
Fourth or later	32	38%	-0.49	-15.7

The first two trades of the day produced +45.4R. Everything after them handed back 24.3R, a little over half of it, across 73 trades that consumed real risk and real attention. The decline is monotonic, which is what makes it interesting: this is not one bad bucket, it is a steady fall with the number of decisions already made.

There is a plausible mechanism, and that matters more than the numbers do. The best opportunity of the day is usually the one you were waiting for, while later entries are increasingly ones you went looking for. Add fatigue and the pull of a losing morning and the decline has a reason to exist. A cut with a mechanism is worth acting on. A cut without one usually is not.

The other four cuts, and what each is good for

By setup tag. The highest-value cut, because it maps straight onto a decision: refine, filter, or retire.
By session. London, New York, the overlap and the Asian hours differ in range and in liquidity, which is enough to change whether a level holds. The sessions guide covers what changes and when.
By instrument. Often a proxy for spread and volatility rather than for skill. A pair costing 2.6 pips needs a larger edge than one at 0.8 pips to produce the same net result.
By market condition. Trending against ranging, using whatever definition you can apply consistently. This cut converts a failing setup into a filtered one more often than any other.

10 more sections in MarketPro

Test enough columns and one will always look significant

This is where journal analysis goes wrong, and it is a mathematical fact rather than a discipline failure. Slice 240 trades twenty different ways and one or two of the slices will show a striking difference purely by chance. Day of the week is the classic offender: with five buckets and no real effect at all, the best and worst day will usually sit several percentage points apart.

Three defences, in order of usefulness.

Demand a mechanism. If you cannot say why Tuesdays would differ, the difference is probably noise wearing a label.
Count the bucket, not the total. A finding resting on eighteen trades is a suggestion. Thirty is the floor, a hundred is where you can lean on it.
Re-test forward. Write the hypothesis down, then check it against the next fifty trades, which the pattern has never seen. A real effect survives that. A coincidence usually does not.

The cut people most want to make and should not

Sorting by result and looking for what the winners had in common is the most natural thing to do with a journal and the least reliable. Any group of winners shares dozens of properties, most of them irrelevant, and the losers normally share them too. The comparison only means something when you pick a property first and then measure both outcomes inside it.

The same trap appears as changing a rule after a bad fortnight. Fifteen trades cannot distinguish a broken setup from an ordinary run, and a method with genuinely positive expectancy still produces losing fortnights regularly. Adjustments should be small, rare and made at a scheduled review rather than in the middle of a drawdown, which is also the moment the risk rules matter most.

From a finding to a change in the plan

The sequence that keeps you honest is short. State the pattern in one sentence with its sample size. Name the mechanism. Estimate what acting on it is worth in R per month, so you can tell a 2R improvement from a 15R one. Then apply it as a filter for a defined trial, usually fifty trades or one quarter, with the old behaviour still recorded so the two can be compared.

On the session data above that means not trading after the second entry of the day for a quarter, while logging the trades you would have taken so the counterfactual is measurable. If the skipped trades would have lost, the filter earned its place. If they would have won, you have learned something concrete about the pattern rather than quietly wondering about it for the next two years.

Finish this lesson in MarketPro

The rest of Spotting Patterns Across Your Journal, a quiz that checks it landed, and your place kept in the curriculum.

10 more sections of this lesson
All 320 lessons, with no cap and no paywall
Free on any account, including a free one

In your browserContinue on webor continue on web

Same account as the app · No install · Free to start

Or scan with your phone camera to get the MarketPro app.

Download on theApp Store Get it onGoogle Play

Free account, no card needed.

Not investment advice. Past performance is not indicative of future results.

Run the numbers while this is fresh

Work it through on your own position size. Nothing leaves your browser and there is no sign-up: Drawdown calculator, Risk of ruin calculator.

Written and reviewed by the MarketPro research desk

MarketPro is a trading-signal and trading-education app operated by Harajuku Holdings LTD (Cyprus). Our signal desk publishes and tracks every trade idea in-app, and every page here is checked against the app's live behaviour before publishing. Last reviewed 15 September 2026. How we produce signals and content · About MarketPro

FAQ

Spotting Patterns Across Your Journal: questions

How many trades do I need before journal patterns are reliable?

About thirty in a single bucket before its numbers are worth discussing, and around a hundred before acting with confidence. The total size of the journal matters less than the size of the group you are reading, so a 500-trade log cut into fifteen buckets is still making claims from small samples.

Why do most journal patterns disappear later?

Because slicing the same trades many different ways produces apparent differences by chance. With five buckets and no real effect at all, the best and worst will still look several points apart. Requiring a mechanism that explains the pattern, and re-testing it on the next fifty trades, removes most of the false ones.

Is it worth tracking the time of day I trade?

Yes, and it is one of the more informative cuts, because liquidity, typical range and the participants all change through the day. Record the session as well as the clock time, since the session is what explains the behaviour. Also record whether a trade was your first, second or fifth of the day, which is a separate effect and often a larger one.

Trade · Learn · Earn

The whole academy is free in MarketPro

All 320 lessons, 12 guided paths, quizzes and a daily streak. Free on every account, in the app or your browser, with vetted free signals alongside it.

In your browserContinue on webor continue on web Download on theApp Store Get it onGoogle Play

Same account as the app · No install · Free to start

Not investment advice. Past performance is not indicative of future results.