Modeling Trading Loss Clusters with Bernoulli Trials
Summary
The article examines whether trading outcomes can be modeled as independent Bernoulli trials, with each trade classified as a win or loss. It defines clusters as consecutive outcomes of the same kind and uses probability theory to reason about how often long winning or losing runs may appear. This framing highlights why an average win rate alone does not describe drawdown risk: outcomes with the same overall win-loss ratio can produce very different loss sequences and recovery paths, especially under geometric position sizing.
The discussion applies this perspective to high-win-rate scalping systems, arguing that a favorable hit rate and large stop relative to target do not by themselves establish profitability. It presents expected waiting times for runs under selected probabilities and discusses synthetic series as a way to assess plausible loss clusters and short backtests. The analysis depends on the Bernoulli assumption of a stable win probability and independence between trades. The author recognizes that real markets may violate those assumptions, so the model is a risk lens rather than proof that trading systems are random or that money management alone determines results.
Key ideas
- A Bernoulli model treats each trade as an independent win or loss with fixed probabilities.
- Loss-cluster lengths reveal risk that an aggregate win rate does not capture.
- Different sequences with the same win-loss ratio can produce different drawdowns under position sizing rules.
- A high win rate and a large stop relative to target do not alone establish a profitable scalping system.
- Bernoulli-based estimates depend on independence and stable probabilities, assumptions that may fail in financial data.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.