Tilt in trading: how to detect it in your own data (and what it costs you)

Tilt is measurable. Three signals in your trade log — entries right after consecutive losses, size jumps, late sessions — and a simple way to price what tilt costs you each month.

"Don't trade on tilt" is advice everyone agrees with and nobody can act on, because tilt does not feel like tilt from the inside. It feels like conviction. The way out is not more willpower; it is a definition you can check in your trade log after the fact, and a rule that fires before the next entry.

A definition you can compute

An entry is a tilt entry if it was taken within 90 minutes after two consecutive losses. That is the rule Sharpen uses by default, and it is deliberately mechanical: no feelings, no interpretation, just timestamps and results.

Two other signals confirm it:

Size jump: the entry is larger than your median size, and the previous trade was a loss. Revenge sizing is the expensive form of tilt.
Late session: the entry is after your usual last trade of the day. The tilt session is the one that "makes it back".

Pricing your tilt

Take your last 100–200 trades (template: free trading journal CSV) and split them in two groups: tilt entries as defined above, and everything else. Compute the average PnL of each group.

Worked example from a sample log: 71 trades, 36 of them within 90 minutes of two losses. The tilt entries averaged −$81 (−$2,910 net over the 36); the other 35 averaged +$27 (+$945 net). The period closed at −$1,965. Remove the tilt entries and the same trader, with the same strategy, is up $945 — the "strategy" was fine, the condition under which it was traded was not.

That number is the whole argument. Not "you were emotional", but "these 36 entries, taken under this condition, cost $2,900". You can run the same calculation in the expectancy calculator (section "cost of tilt").

Rules that actually stop it

Rules work when they are cheap to check and priced. Three that work for most traders:

Stop after 2 losses — for the rest of the session, or for 60 minutes minimum. This one alone removes most tilt entries by construction.
Max size = 1.5× median — until you have three winners in a row. Kills revenge sizing without banning size.
No entries after {your cutoff} — the late-session rule. Your own log tells you the hour; for most day traders it is the last 60 minutes.

Write each rule with its cost attached: "Breaches of rule 1 cost me $1,900 last quarter". A rule with a price sticks; a rule without one gets negotiated with at 15:42.

Why it is hard to see in a spreadsheet

You need timestamps, streaks and sizes on the same row, then a filter that depends on the previous two rows. It is doable in a spreadsheet, and nobody does it every week. Sharpen runs it on every import: the *Trading on tilt* leak shows the affected entries, their average result versus the rest, and the dollar gap. Then the "stop after N losses" rule counts its own breaches — and prices them.