Annualizing Sharpe Ratios for Intraday Strategies

Summary

The document raises a practical question about annualizing a Sharpe ratio calculated from a strategy that trades on five-minute or hourly bars. It contrasts the familiar daily scaling factor based on 252 trading days with scaling by the square root of the total number of bars, and asks which period count should be used.

It provides no answer, supporting analysis, or evidence, so it does not establish that either scaling choice is appropriate. Annualization depends on matching the Sharpe calculation’s return interval to the number of such intervals in a year, while the usual square-root scaling relies on assumptions about returns, including limited serial dependence. The document also does not clarify whether the underlying return series is actually intraday or daily, leaving the calculation method unresolved.

Key ideas

• Sharpe annualization must use a period count that matches the interval of the returns being measured.
• The document contrasts daily annualization with scaling by the number of intraday bars.
• It leaves the correct scaling choice unanswered and provides no empirical evidence.
• Serial dependence can limit the validity of standard square-root-of-time scaling.

Tags

# Annualizing Sharpe Ratio using small time frames


# Annualizing Sharpe Ratio using small time frames












I have coded a strategy that works 5m time frame. I know you multiply it by 252 but i am using 5m or sometimes 1h time frame. Which number do i have to chose to multiply? There are 72576 five minutes in a year. Do i have to chose this number? Here how i calculate sharpe ratio

```
daily = sm['cash'].pct_change()
sharpe_ratio = daily.mean() / daily.std()
a_sharpe_ratio = (252 ** 0.5) * sharpe_ratio
entire_sharpe_ratio = (total_bars ** 0.5) * sharpe_ratio
```

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.