Why You Break Your Own Trading Rules (and the Fix That Sticks)

You did not break the rule. That is the first thing to get straight, because it explains why the usual fix never works.

At the moment you clicked, you had a reason. The setup was close enough. The level was obvious. You were already down and this one was clean. Rule-breaking almost never feels like rule-breaking in real time. It feels like an exception, and it is only in the review, hours later, with the outcome known, that it turns into a violation.

Which means the advice to be more disciplined next time is aimed at a moment that never presents itself as a choice. You cannot exercise willpower against a decision you did not experience as one.

So this post does something different. It treats a broken rule as a diagnostic, sorts the four reasons a rule fails, and gives the fix that matches each one. Only the last of the four is a discipline problem. The other three are design problems, and design problems are far easier to solve than character.

The Four Reasons a Rule Fails

Every rule you repeatedly break falls into one of these. Not two of them, usually one, and the fixes are close to opposites.

1. The rule is not decidable at the moment you act. It can only be judged afterwards, so following it in real time was never actually available to you.

2. Breaking it is free. Nothing records the break. No cost, no count, no evidence, so nothing accumulates that could change your behaviour.

3. The rule does not pay. You are overriding it because your instinct is right and the rule is wrong. Your data would say so if you asked it.

4. You were not really choosing. Post-loss, late session, out of time, third trade of a bad morning. The rule is fine, the state was not.

Cause 1: The Rule Cannot Be Decided Before You Act

Read your trading plan and mark every rule that requires knowing the outcome. Most plans are more than half made of these.

The test is one question: at the instant you have to act, and without knowing how the trade turns out, can this be answered yes or no? Not by a panel of traders after the fact. By you, in the moment, with only what is on the screen.

The right column is charmless and that is the entire point. A rule you can argue with is not a rule, it is a preference wearing a rule's clothes, and preferences bend at exactly the moment you need them not to.

If most of your plan lives in the left column, you do not have a discipline problem yet. You have a specification problem, and the score you would get from measuring adherence to unspecified rules would be meaningless anyway. The mechanics of the rewrite, and how to turn the results into a single number, are worked through in the discipline score post.

Cause 2: Breaking the Rule Costs You Nothing

Assume the rule is now decidable. You still break it, and here is the uncomfortable mechanism: nothing happens when you do.

There is no fine. There is no log. In four days you will not remember it, and if the trade won you will remember it as good reading rather than a violation. A rule with no record attached is a rule you break for free, and behaviour that is free tends to continue.

The fix is not motivation. It is a field.

When the condition is a field you fill in on every trade, with a required answer defined in advance, three things change. You have to look at the condition while logging, so it stops being background noise. The break becomes a data point rather than a memory. And at 30 trades you can count violations instead of estimating them.

That count is the part that does the work, because memory is unreliable in one specific direction. Before you look at your own number, write down your guess. Most traders can correctly name the rule they break most and are badly wrong about the frequency. Guessing 3 and finding 11 is a different experience from being told to try harder, and it is the only version of this that reliably changes behaviour.

Be clear about what a journal can and cannot do here, because the honest limit matters. A journal does not block the trade. Nothing in TradingSFX stops you saving a trade that violated a rule, and any tool promising to enforce your plan at the broker is describing something else. What the field does is make the break countable and visible next to its outcome. Enforcement is still yours. Evidence is what you are buying.

The second half of the fix is how the rule is phrased. Compare:

• I will not trade after two losses.
• If I take a second loss in a session, then I close the platform for the day.

The second names the trigger and the response together, so the decision is made once, in advance, rather than negotiated in the moment when you are least equipped to win the negotiation. Psychologists call this an implementation intention, a concept introduced by Peter Gollwitzer in 1999. Traders arrive at it independently and call it a hard stop.

Write your rules in that shape. If X, then Y. It costs nothing and it converts a judgement call into a lookup.

Cause 3: The Rule Does Not Actually Pay

This is the possibility almost nobody entertains, and it is common enough to check before you spend another month grinding on willpower.

You keep overriding the rule because some part of you has noticed it does not help. Rules get inherited from courses, mentors and Discord servers, and they are rarely tested against the person who ends up carrying them. A rule tuned to someone else's session, instrument and holding period can be actively wrong for yours.

The test is a two-set comparison, and it needs your journal rather than your opinion:

1. Filter to trades where the rule was respected. Compute expectancy.
2. Filter to trades where it was violated. Compute expectancy.
3. Compare.

Three outcomes, three different actions:

• Respected clearly outperforms. The rule is real and every break is costing you money. Go back to cause 2 and make the break visible.
• No meaningful difference. The rule is folklore. Delete it. This feels like giving up and is the opposite: a plan with four rules you keep beats a plan with nine you do not, and every dead rule you carry dilutes the attention available for the live ones.
• Violated outperforms. Usually the rule is drawn too tightly rather than being wrong outright. Widen it, then measure again.

Two guard rails. Give it at least 30 recorded checks before acting, because a rule with six is a rumour. And remember the violated set is not a random sample: those trades cluster on the days you were tilted or tired, so part of any gap is the state you were in rather than the rule. Treat a small difference as noise.

If expectancy is not a number you compute today, the formula and three worked examples take about ten minutes to run on a sample you already have.

Cause 4: You Were Not Really Choosing

Now the genuine discipline case, and it is narrower than people assume once the first three are cleared.

The signature is clustering. Violations are not spread evenly across your trades, they bunch. The way to see it is to stop looking at the rule and start looking at when the breaks happened. For every violated trade, record three things:

• Position after a loss. Was the previous trade a loss? Were the previous two?
• Time inside the session. First hour, or the last thirty minutes before you had to leave?
• Trade number that day. Was it your first, or your fourth?

Then compare each against your baseline. If a fifth of your trades are third-or-later trades of a day but half your violations are, the rule is not the problem. The fourth trade of the day is the problem.

Post-loss clustering has its own name and its own pattern in the data, including the size creep and time compression that come with it, all of which is countable: how to spot revenge trading in your journal covers the four numbers that expose it. End-of-session clustering usually means you are trading a window you should not be in at all, which shows up cleanly in a performance breakdown by hour.

The fix for state is never to try harder inside the state. It is a circuit breaker set outside it, phrased in the decidable form from cause 1:

• If I take two losses in a session, then I stop for the day.
• If I have taken three trades today, then the platform closes regardless of what I see.
• If a trade closes at a loss, then I wait 30 minutes before the next entry.

These are unglamorous, and they are the only category of rule that works while you are compromised, because they do not require you to make a good judgement at the moment your judgement is worst.

What This Costs, Worked Openly

Some arithmetic, because rule-breaking gets dismissed as a minor tax until you price it. Everything below is an illustration built from stated inputs, not a study.

Take 100 trades, risking 1R each, with one rule tracked. It was respected on 80 trades and broken on 20, so one trade in five, which most traders would call decent adherence.

The respected set: 36 wins averaging 2.1R gives 75.6R, and 44 losses averaging 0.95R gives -41.8R, so +33.8R. Expectancy is 33.8 divided by 80, or +0.42R per trade.

The violated set: 6 wins averaging 1.4R gives 8.4R, and 14 losses averaging 1.05R gives -14.7R, so -6.3R. Expectancy is -0.32R per trade.

Two things worth sitting with.

The direct cost is 6.3R. On a 10,000 account risking 1%, one R is 100, so those 20 trades removed 630 from a sample where the strategy itself was working.

The hidden cost is the drag on the headline number. The strategy produces 0.42R per trade when run as designed. What actually landed in the account is 27.5R over 100 trades, or 0.275R. A third of the edge went to a rule broken one time in five.

And notice what the win rate did while all that happened: 45% respected, 42% overall. Three points. If you were watching win rate for a sign that something was leaking, you would have seen nothing, which is a recurring theme with win rate as a headline metric.

The 30-Day Protocol

Do not define twelve rules on a Sunday evening. That version is abandoned by week three, every time.

Pick one rule. The one you already know you break, the one written in your journal three separate times.

Make it decidable. Rewrite it until it can be answered in the moment, without the outcome. Use the table above as a model.

Phrase it as if-then. Trigger and response in the same sentence.

Tag it on every trade for 30 days. Every trade, including the ones where you broke it. Especially those, since a blank field is how the count quietly lies to you.

Then check three things, in this order. Was it decidable in practice, or did you find yourself arguing with it? What is the actual violation count against the number you guessed? And do the respected trades outperform the violated ones by enough to matter?

You now know which of the four causes you are in, and the fix follows from the answer rather than from resolve.

If you are on a funded account, run this before the clock matters. Evaluations end on rule breaches far more often than on bad strategies, and what to track in a prop firm journal covers which conditions are non-negotiable there.

In TradingSFX

The mechanism this post describes maps onto one feature. A confluence is a field you define and fill in per trade, and it becomes a tracked rule the moment you set a required answer on it. Yes/no for conditions, a numeric minimum for thresholds, an allowed list for text.

That single definition then does three jobs, which is what makes the diagnostic above cheap enough to run:

• The count. The Discipline Score card shows adherence per rule, sorted worst-first, plus the trend across your trade sequence. That is cause 2 answered.
• The comparison. The same tag is a filter on the analysis pages, so respected and violated become two sets you can read expectancy off directly rather than rebuilding by hand. That is cause 3 answered.
• The clustering. Filtered trades can be read against time of day and sequence, which is where cause 4 shows itself.

Two details that matter in practice. Rules can be scoped to one workspace, so a condition that belongs to your breakout strategy does not appear on the form when you are logging mean-reversion trades and does not get scored against it. And the automatic post-trade verdict on Pro and above reads your Coach Profile, so when a trade repeats a weakness that is already established in your history, it says so rather than reviewing the trade in isolation.

On plans: Basic is free forever at 10 trades a month and includes one custom confluence, which is exactly enough to run the 30-day protocol on a single rule. Pro at 19.99 a month raises that to five and unlocks the per-symbol and per-strategy analysis pages where the respected-versus-violated comparison lives. Premium at 29.99 removes the confluence limit.

The Part Worth Keeping

You are probably not undisciplined. You are more likely carrying rules that were never checkable, breaking them at no cost, and holding at least one that does not earn its place.

Fix the specification, make the break visible, delete what does not pay, and put a circuit breaker around the states where you are not really choosing. What is left after that is the actual discipline problem, and it is a much smaller thing than the one you have been fighting.

Turn your first rule into a tracked condition and find out how often you really break it.

Published August 19, 2026. No competitor pricing or feature claims are made in this post, so nothing required price verification. The implementation intentions concept is attributed to Peter Gollwitzer (1999); no effect sizes or study results are quoted. All arithmetic is worked openly from the stated inputs and no statistic here comes from a survey or study.