Measuring Portfolio Diversification with Drawdown Compression and Resampling
Summary
The document describes an analyzer for assessing whether several trading systems diversify one another. It groups closed trades by magic number, aggregates profit and loss at a chosen daily, weekly, or monthly interval, and compares the sum of each system’s maximum drawdown with the combined account drawdown. It also reports pairwise Pearson correlations, a heat map, and Monte Carlo drawdown percentiles from randomly reshuffled period results. These measures are intended to show how much losses offset across systems and how sensitive risk is to the historical ordering of returns.
The document recommends using monthly buckets when daily data are sparse and sizing positions with reference to the adverse tail of the resampled drawdowns. Its figures are based on realized closed trades: open-position excursions are excluded, so actual equity drawdown may be larger. Correlations based on short histories are unreliable, and magic number zero may combine manual trades with default-configured systems. The described metrics assess historical behavior; they do not forecast future performance or establish that apparent diversification will persist.
Key ideas
- Comparing combined drawdown with the sum of standalone drawdowns estimates historical drawdown compression across systems.
- Pairwise correlations of period profit and loss help identify systems whose returns tend to move together.
- Monte Carlo reshuffling shows how drawdown can change when the same observed period returns arrive in a different order.
- Sparse daily observations can pull correlations toward zero, making coarser aggregation more suitable for some portfolios.
- Closed trade results omit floating losses, and short overlapping histories provide weak evidence about correlation.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.