Exiting trades too early is usually described as a personality flaw: "I can't let winners run." That sentence cannot be practised. A relief exit can. It is an exit whose job was to change how you felt, not to satisfy a written rule. On replay you can tag it at the click, then — without using the later tape to pardon yourself — count what the next five bars did.
The point is not to prove that holding would have been better. Sometimes the relief exit "saves" you. The point is to stop mixing two ledgers.
What a relief exit looks like on an NSE replay
A rule exit can be pointed at on the page you printed this morning: target, trail trigger, time stop, structure break, or session square-off. A relief exit shows up as a sentence you would not put on that page:
- "It's green; I should take it before it vanishes."
- "The 09:40 wiggle is too much; I am out."
- "Nifty stalled, so my large-cap long must be finished."
- "I will scratch to reset, then find a better entry."
Scratching to a written time-stop is a rule. Scratching because your stomach dropped on a practice chart is relief. Partial exits need the same honesty: if the scale was not on the card, it is a relief slice.
You will also see relief in the last hour. Intraday cash into 15:30 has a real clock. Flattening at a pre-written 15:10 is a rule. Flattening at 14:20 because a small green open felt fragile is not.
The REX protocol (Relief Exit Index)
Run this on hidden-date, candle-by-candle replay. Frozen rules. Hypothetical 1R defined before the first entry. Think in R, not in rupee theatre — R-multiples keep the relief story from hiding in ticks.
Session rules
- Product declared: cash intraday or delivery. Not both.
- Exit laws written: at least one profit-taking law and one invalidation law. If your only exit is "when I feel done," you cannot detect relief. You are relief.
- No blending a new trail because this trade is green. Trails belong on a trail card or they belong in the relief column.
At every exit, before the next reveal
Tag exactly one:
| Tag | Meaning | Allowed evidence |
|---|---|---|
exit-rule |
A pre-written law fired | You can quote the sentence |
exit-relief |
You flattened to feel better | You can quote the feeling-sentence |
exit-mixed |
A rule was nearby and you used it as cover | Both sentences exist; default to mixed |
Then record, still before you moralise:
mfe_next5— maximum favourable excursion, in R, over the next five bars after exit (mark N/A if replay ends)mae_next5— maximum adverse excursion over those five barsprocess: 0-2— did you follow the tag honestly, not whether the exit was "right"
mfe_next5 is a measurement, not a verdict. A relief exit followed by +0.8R of leftover trend does not prove you should have held. A relief exit followed by a collapse does not prove fear is wisdom. You are building a tag frequency, not a guru backtest.
Block size
Twenty exits. Entries that hit the stop without you touching them are exit-rule (stop law). Do not skip logging those; otherwise the sample becomes "only the exits I chose," which is where relief hides.
After twenty:
REX = count(exit-relief) / count(all exits)
REX is a rate in your log. It is not a market statistic. Do not compare it to anyone else's number. Track it across two windows: 09:15–11:00 and 13:30–15:30. Opening relief and last-hour relief are different animals.
What to do with the number (narrow)
If REX is high in the open window, the usual defect is a missing hold rule, not a missing pep talk. Write one hold sentence that can fail a stranger test: "After entry, no discretionary scratch until either (a) stop, (b) target, or (c) a close back inside the opening range." Then run another twenty. If REX does not move, the sentence is still a toast.
If REX is high only after the trade is green, you do not have an exit problem so much as a give-back problem. Tag protect-green and treat those rows with the same seriousness as emotion logging.
If REX is low but process scores are also low, you may be holding through invalidation and calling it "not relief." That is hope. Different article, different counter.
Failure modes
- MFE worship. You start holding because last week's
mfe_next5was large. That is outcome-chasing with extra columns. Freeze the exit laws for the block. - Target-as-relief. You set a 0.3R target so you can always tag
exit-rule. The rule is then a costume for relief. Targets have to be written before the session, not when the trade is +0.2R. - Index alibi. "Bank Nifty turned" is not an exit law unless your page says the stock trade dies on an index close. Otherwise it is a story.
- Review rewrite. In the evening you change
exit-relieftoexit-rulebecause the next bars were ugly. The tag is timestamped at the click or it is fiction. - Sample shopping. You only run REX on trend days you already like. Randomised starts exist to stop that.
Process metrics still outrank the R column while the sample is small. Twenty exits tell you about your hands. They do not tell you about expectancy. Practice fills ignore brokerage and taxes; leftover MFE on a sim chart is not a live rupee.
Replay Trader's AI review can list exits that cluster at the first green bar. Treat that as a prompt to re-read your tags, not as a grade.
Tag the feeling-exit. Then count. Do not pardon it with the next candle. Join the waitlist.
Related reading
- Practising Partial Exits Without Regret Math
- Trailing Stops: The Hardest Thing to Practise
- Thinking in R Instead of Rupees
- Logging Emotions Without Turning It Into a Diary
Educational note
Educational practice content only. Not investment advice. No guaranteed returns. Practice ≠ live profitability.