Stop-loss drift is the behavioural pattern of widening or removing the SL during a losing trade — either literally (dragging the line on the broker UI as the loss runs) or implicitly (cancelling the SL and 'holding for a recovery'). It is the leading single cause of account-blowing drawdowns in self-directed trading.
The Onyx detector doesn't have an audit log of SL changes — broker exports only carry the FINAL `sl` price. So it infers drift from the relationship between the recorded SL and the actual close. For a long trade with `sl=99` and `open=100`: if the trade closed at 97.5, it bled 1.5 points past the original SL. That overshoot is then expressed in R-units (multiples of the original 1R risk distance, `|open - sl|`) so the metric is comparable across symbols and timeframes.
Overshoots below 0.3R we treat as ordinary slippage; beyond that, the pattern is behavioural.
20 trades with a recorded SL. A handful overshot meaningfully past their SL. The worst was a TSLA long that closed 2.5R past its original SL.
Severity bands on the drift-rate: - Mild — occasional drift; could be 1-2 emotional moments per month. Not yet structural but worth a journal note when it happens. - Moderate — the pattern is established. Roughly one in three losers is now drifting; the average overshoot compounds quickly into the equity curve. - Severe — drift is the dominant risk-management mode. Stop journaling and start trading paper until the drift rate falls back materially over the next 30 days — or the next overshoot is the one that ends the account.
The Onyx-Engine assigns your drift rate to one of these bands; the cutoff thresholds are TradeOnyx-internal calibration.
The detector flags overshoot, not slippage. A small overshoot on a fast-moving instrument is the broker filling beyond the SL price — that's market mechanics, not behaviour. The detector's overshoot threshold is the conservative dividing line between the two.
Three causes, one fix. Trader didn't believe the setup at entry; trader believed the setup but couldn't accept the loss; trader's SL was a bluff to themselves. All three converge on the same fix: write the SL in the journal entry BEFORE entering the trade and treat the broker SL as a hardware-level enforcement of that journal entry.
Tier: Pro. Risk-management discipline axis.
How to read the card:
1. Hero (left) — the drift rate as a percentage with a severity verdict. The single number to commit to memory week-over-week. 2. Three MicroStats — Trades with SL (the denominator the rate reads against), Drift count (absolute number of overshooting trades), Avg overshoot (mean R past SL across all drift trades). 3. Worst-case callout (red box) — the single worst overshoot with the symbol and the R-distance. This is the trade you don't want to repeat. Use it as the anchor when re-reading your journal entry for that trade.
The Stop-Loss-Drift → Position-Sizing pairing. Drift and oversized positions tend to fire together. Drift means the trader couldn't accept the loss at the planned SL; oversized positions mean the planned SL was too painful to honour. Both signal the same root cause — risk-per-trade is set above the trader's true tolerance — and the fix is mechanical: cut declared risk-per-trade by 30 %, run for 30 days, re-check.
Re-look frequency: weekly. Daily fluctuations are noise; the behavioural pattern only stabilises after ~10–15 trades.
Tier: Pro.