Survival Analysis for Conditional Trade Target and Stop Probabilities
Summary
This article adapts survival analysis to trade duration, asking how the chance of reaching a target changes for positions that remain open as they age. It represents each trade by duration, outcome, and strategy identifier, then uses Kaplan–Meier survival estimates, Aalen–Johansen cumulative incidence, hazards, quantiles, and conditional probabilities. Wins and losses are competing terminal events; only a position still open at the observation cutoff is censored.
The article contrasts this approach with treating losses as censored, which incorrectly gives stopped trades the future outcomes of surviving trades and inflates estimated target probabilities. It also describes reconstructing trades from MetaTrader deals by position ID, accounting for partial exits and cutoff timing. A demonstration reports different conditional target chances for trades at different ages and shows a large gap between the correct and naive estimates. The panel is descriptive rather than an exit signal: its age-conditioned probabilities omit remaining distance to levels, unrealized profit or loss, and market state.
Key ideas
- A win and a loss are competing terminal events, while an open position at the cutoff is censored.
- Treating stopped trades as censored inflates estimated target incidence.
- Trade records should be reconstructed by position ID to combine entries and partial exits.
- Survival estimates can answer conditional questions about outcomes given trade age.
- The displayed probabilities condition on age alone and do not prescribe exits.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.