Statistical Hypothesis Tests for Trading Results and Execution Data
Summary
The article introduces null and alternative hypotheses, significance levels, and the risks of false rejection or acceptance. It illustrates nonparametric testing with trading examples: a Wilcoxon signed-rank test compares results before and after changing entry prices, a Mann–Whitney U test compares order execution times across brokers, and Spearman rank correlation tests whether two strategies’ stop-loss series move together.
The examples report that the entry-price change did not significantly affect the relative yield among completed trades, although missed entries reduced absolute profitability. Broker execution-time samples are reported as different, while the strategy stop-loss samples show differing correlation results. These findings depend on the selected samples and outlier treatment; the article does not establish that the results generalize to other periods, brokers, or strategies. It also cautions that correlation alone does not decide whether a strategy should be removed from a portfolio.
Key ideas
- A hypothesis test compares a null claim with an alternative and balances significance against type I and type II errors.
- The Wilcoxon signed-rank test is used to compare paired trade profit samples after an entry-price adjustment.
- The Mann–Whitney U test compares independent broker execution-time samples without assuming a known distribution.
- Spearman rank correlation can assess relationships between strategies’ stop-loss series, but portfolio choices still require judgment.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.