Hold-time asymmetry is the pattern of holding losing trades materially longer than winning trades. Two divergent psychological pulls produce the same data signature:
- Cut winners short — the gain feels real, the trader 'protects' it by exiting prematurely. Avg winner hold-time drops below the strategy's natural lifespan.
- Hold losers long — the loss isn't real until you close it, so the trader hopes for recovery. Avg loser hold-time bloats past anything the original setup justified.
The detector takes the average hold-time on winners and on losers separately, then computes the ratio (avg loser hold / avg winner hold). A ratio above the threshold is the conventional behavioural-finance marker for meaningful loss aversion.
Uses ``close_time - open_time`` as the hold-time proxy. Open trades (no close) are excluded; positions that straddle the weekend get their full elapsed hold counted, which is the right answer for swing traders and benign noise for day traders (whose positions never straddle the weekend in the first place).
Across the analysis window: 12 winners averaging 18 minutes hold-time, 14 losers averaging 56 minutes. Ratio: 56 / 18 ≈ 3.1×.
Severity bands on the asymmetry ratio: - Mild — losers held meaningfully longer than winners but the gap is still in single-digit-multiples of the winner average. Note in the journal; a single bad week can produce this. - Moderate — losers held 2-3× the winner average. The pattern is established. Time-stop discipline becomes mandatory, not optional. - Severe — losers held >3× the winner average. The classic capitulation footprint: a few hopium-driven losers per month bleed weeks of disciplined gains.
The Onyx-Engine assigns the ratio to one of these bands; the cutoff thresholds are TradeOnyx-internal calibration.
Why average hold matters more than maximum. A single 8-hour loser among forty 20-minute winners would NOT trip a max-hold rule but WILL ruin the average. The asymmetry ratio captures the cumulative behavioural drag — every long-held loser eats two or three winners' worth of opportunity cost.
The fix is mechanical, not motivational. Define a hard time-stop on losing trades equal to the average hold-time on your winners. If a loser hasn't recovered within that window, close it — the losing-trade hopium is what bloats hold-time without changing outcomes. Pair this with the Stop-Loss-Drift detector when both fire together; the time-stop is the temporal counterpart to the price-stop discipline.
Tier: Pro. Exit-quality axis. Pairs with Hold-Time-Quality (Phase A, free tier) which uses MFE/MAE for the same axis — this detector reads TIME instead, giving a different lens on the same behavioural state.
How to read the card:
1. Hero (left) — the asymmetry ratio with severity verdict. Single number commit-to-memory. 2. Two MicroStats — average hold-time on winners (green) vs losers (red), side-by-side in minutes. The visual gap IS the diagnosis. 3. Hint line — actionable: the time-stop rule keyed to the winner average.
Re-look frequency: weekly. The hold-time signature stabilises after ~10–15 winners + 10–15 losers; daily noise isn't useful here.
Tier: Pro.