Trade Loss Attribution: Was It the Read, the Instrument, or the Exit?

Trade loss attribution is the step between losing money and learning something, and most traders skip straight past it to a conclusion that feels explanatory: I was on the wrong side. That diagnosis is almost always available after the fact, and it is almost never actionable. A loss has three separable causes — the read, the instrument, or the exit — and only one of them is usually the real one.

Asked in the right order, three questions separate them in under a minute.

The three questions, in order

1. Did the move I predicted actually happen?          → if yes, the READ was not the problem
2. Would my contract have paid on a NORMAL move
   for this symbol?                                    → if no, it was the INSTRUMENT
3. Was there a point where I was up and did not
   take it?                                            → if yes, it was the EXIT

The order is not arbitrary. Question 1 comes first because if the move happened, every directional lesson you were about to draw is void. Question 2 comes second because it is arithmetic you could have checked before entry. Question 3 comes last because it only becomes the binding cause once the first two are clean.

Why "I traded counter-trend" is the wrong lesson

It is the most common self-diagnosis in a trading journal and it fails on two grounds.

It is tautological. After a loss, "I was on the wrong side" is a restatement of the outcome, not a cause of it. It cannot be wrong, which is exactly what makes it useless.

It has been measured. The rule it implies — don't fight the day's direction — was tested across 38,897 signals and 2,703 sessions, using five separate causal proxies for direction at three timeframes. The pre-declared contrast came back +0.027 R with a Holm-corrected p of 0.34: not supported. The trend-alignment interaction ran +0.054 at one minute and −0.012 at fifteen — it does not even hold its sign.

And +0.027 R is not merely insignificant, it is smaller than the cost of trading it: the round-trip option spread runs 0.06 to 0.27 R. Even had the effect been real, expressing it would have cost more than it paid.

A worked case: yes / no / yes

A TSLA 0DTE put, 31 August 2026. The thesis was price ran up all morning, expect a pullback. The trade lost 100%. Run the three questions.

Question 1 — did the move happen? Yes. Price fell, promptly, in the direction predicted. Whatever went wrong, direction did not.

Question 2 — would the contract have paid on a normal move? No. TSLA's median session gives 1.39% of downside. The contract needed 2.26% — roughly three times the move that actually arrived, and above the symbol's typical day. Its unconditional base rate for a ≥2% down-move is 36%, and measured from the open at that: the entry was mid-morning near the highs, so the true conditional probability was worse still.

Question 3 — was there a point of being up? Yes. At the pullback low the contract was worth about $1.13 against $0.60 paid — up 89% — and it was carried to expiry.

The verdict is instrument and exit, not direction. The actionable version reads: the strike I chose could only be used as a fast scalp, and I held it like a position. That is a rule you can apply tomorrow. "I traded counter-trend" is not.

What each verdict actually changes

If it was the instrument, the fix is a fifteen-second calculation before the click: (spot − (strike − premium)) ÷ spot for a put, then check that percentage against the symbol's median day. Our guide to 0DTE strike selection has the base-rate table for thirteen symbols.

If it was the exit, the fix is mechanical and must be decided before entry, not during:

• a premium stop — for a long option the stop is a premium level, not a price level; −50% is a reasonable default, and the point is that it exists in advance
• a time stop — if this has not worked within N minutes, the read was wrong — honoured independently of price
• two contracts minimum. One contract forces an all-or-nothing decision. Two give you a mechanism: bank one, run one.

If it was the read, resist doing anything at all on a single sample. Look at the last twenty. If the read fails more often than it lands, that is a signal-quality problem, and it will not be fixed by a filter bolted on after a bad afternoon.

Where this belongs in a journal

Attribution is worth almost nothing as a one-off and a great deal as a column. Record the verdict on every loss and after thirty trades the distribution tells you which problem you actually have — and each has a different fix, applied in a different place.

A trader convinced their problem is direction, whose journal says instrument eleven times out of fifteen, has been researching the wrong thing for months. That is the value of attribution: it points at the drawer where the fix actually lives.

The takeaway

Before you write a lesson in your journal, answer three questions in this order: did the move happen, would the contract have paid on a normal move, and was there a moment you were up.

Most losses that feel directional are not. They are an instrument that could not pay an ordinary day, or an exit that was never written down — and both of those are fixed before the trade, not after it.