Aggregating Sharpe Ratios Across Portfolio Rebalancing Periods

Summary

The document asks how to assess a dynamic portfolio strategy when a backtest is divided into multiple periods and the strategy is tested with different rebalancing intervals. For each interval, the author has period-level mean return, standard deviation, and Sharpe ratio, and is unsure how to form an overall performance measure. They suspect that simply averaging the individual Sharpe ratios may be inappropriate.

The post provides example period statistics but no answer, proposed aggregation method, or backtest conclusion. Its central issue is that Sharpe ratios are ratios of return and risk, so averaging period-level values need not match the Sharpe ratio calculated from the full sequence of portfolio returns. Valid comparison also depends on consistent return frequency, annualization, treatment of overlapping periods, and the strategy’s changing exposures; these details are not resolved in the document.

Key ideas

• The author compares dynamic portfolio results under several rebalancing intervals.
• Each interval produces multiple period-level return, volatility, and Sharpe statistics.
• A simple average of period Sharpe ratios may not represent the strategy’s overall risk-adjusted performance.
• The post gives example statistics but does not provide an aggregation solution or final interpretation.
• Comparisons require attention to return frequency, annualization, and the dependence between periods.

Tags

# Sharpe ratio for dynamic portfolio


# Sharpe ratio for dynamic portfolio












I want to test the performance of my strategy with different rebalancing period. I'm struggle with calculating the overall performance on backtest results and making final conclusions.

For example, by backtest window is 252 days. I want to measure performance with the set `rebalancing_period = {12,36,63,126}` So after backtest for every rebalancing period `t` I have `252/rebalancing_period[t]` statistics for every period.

```
------------
rebalancing_period = 12
------------
period: 1
p_mu = 0.895
p_std = 0.4
sharpe_ratio = 2.24

period: 2
p_mu = 0.675
p_std = 0.37
sharpe_ratio = 2.01

...
period: 20
p_mu = 0.679
p_std = 0.2
sharpe_ratio = 1.24

------------
rebalancing_period = 36
------------
period: 1
p_mu = 0.596
p_std = 0.21
sharpe_ratio = 1.24

period: 2
p_mu = 0.475
p_std = 0.27
sharpe_ratio = 2.0

...
period: 6
p_mu = 0.345
p_std = 0.15
sharpe_ratio = 1.27

....
```

But I can't understand how can interpret this results. Finding the mean of sharpe ratio over all periods seems weird.

I might be wrong with my initial idea, so I'll appreciate any suggestions and remarks.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.