How trading behavioural coaching works, step by step

It works in three passes: measure the behaviour from your own trade history, name the pattern in your own numbers, then change one thing and measure again. Everything beyond that is decoration.

The short version

Behavioural coaching is a measurement loop, not a conversation. Pass one gets enough data to see a pattern. Pass two states the pattern as a number from your own trades, because a number is harder to argue with than an impression. Pass three changes exactly one behaviour and re-measures over a comparable sample. If a coach cannot show you which number they are moving, there is no loop.

Pass one — the data

Nothing works without a trade record that contains, for every position:

• entry and exit price, and the times of both
• position size, in money or in risk units
• the stop you set, and every change you made to it
• what you said you would do before you entered

The last one is the one people skip, and it is the one that makes the rest meaningful. Without a stated intention there is no deviation to measure — only outcomes, and outcomes are noisy enough to justify anything after the fact.

If you are pulling this from a broker automatically, the first four come for free and the fifth does not. Write the intention before the entry, in one line. "Long, breakout retest, stop under the level, out at the prior high." That sentence is the entire input for half of what follows.

Pass two — the pattern, in your own numbers

Here is a worked example of the shape this takes.

A trader believes they size consistently. Their record says their average position after a winning trade is 1.0 R, and after a losing trade 1.8 R.

They are not sizing consistently. They are running a strategy that increases exposure after adverse outcomes — the opposite of what they would design on purpose.

The arithmetic matters more than the disapproval. Suppose the underlying method wins 40% of the time at 1.5 to 1. Sized flat, that is a small positive expectancy. Sized at 1.8 times after every loss, the losing streaks are amplified while the winning streaks are not, and the same method now needs a materially higher win rate to produce the same result. Nothing about the analysis changed; only the sizing did.

This is why the number has to come from the trader's own record rather than a textbook. "Sizing up after losses is bad" is a sentence anyone can nod at and ignore. "Your average size after a loss is 1.8 times your average after a win, across 114 trades" is not.

Other patterns that show up the same way:

• average hold time, winners versus losers. Longer on losers is the hold-and-hope signature, and it is extremely common.
• stop movements, counted. Not whether you move stops — how many times, in which direction, and whether the trade was in loss at the moment.
• trades taken within N minutes of a loss. The tilt window is usually shorter and sharper than people expect.
• the gap between stated intention and executed trade, counted as a rate.

Pass three — one change, then measure

One. Not three.

Two changes at once cannot be attributed. If you cap size after losses and start taking only A-setups, and the next sixty trades improve, you have learned nothing about which one did it — and you will keep doing both forever, including the one that did nothing.

A change is specified as a rule, not an intention. "Be more disciplined about size" is not a rule. "Maximum size after any loss is the same as maximum size after any win, no exceptions, checked before entry" is.

Then it has to run long enough to mean something. Which brings up the part almost everyone gets wrong.

How long before the numbers mean anything

Sample size, not calendar time.

Thirty trades is not a result. At a 40% win rate, thirty trades will contain stretches that look like a broken system and stretches that look like a solved one, and both are noise. A behavioural change needs to be measured against a comparable sample before and after — a hundred trades on each side is a reasonable floor, and even that is not much.

For a trader taking two positions a week, that is a year. This is uncomfortable and it is also true. Any coach reporting progress after ten trades is reading variance and calling it improvement.

The practical consequence: measure the behaviour, not the P&L. Whether you sized correctly after a loss is knowable on trade three. Whether it made money is not knowable for months. The behaviour is the thing under your control and the thing worth reporting weekly.

What this does not solve

It assumes an edge exists. The loop measures execution against a plan. If the plan has negative expectancy, perfect execution converges on losing money steadily. Check the strategy first — expectancy over a real sample — before spending months on behaviour.

It cannot see what is not in the record. The trade you did not take because you were afraid leaves no row. The position you sized down because your rent was due leaves no row either. Both matter and neither is measurable this way.

It does not survive a broken data source. A record with missing stops, or entries logged after the fact from memory, produces confident numbers about nothing. Automatic capture from the broker is worth more than any analysis run on top of hand-typed data.

> Indikora runs the measurement half of this loop automatically — reading a connected trade history for these patterns and reporting them with your own numbers. The argument about whether last Tuesday was different is still yours.