Initializing the Differential Sharpe Ratio Recursion

Summary

The document raises an implementation question about the differential Sharpe ratio in an online setting. It contrasts this approach with calculating Sharpe over a sliding time window and notes that the method updates two recursively maintained quantities, commonly denoted A and B, using the current return and their previous values.

Its focus is the startup condition: the equations described in the question define later recursive updates but do not make clear what values to use at the first observation. No proposed initialization, derivation, numerical example, or answer is included, so the document identifies an unresolved methodological detail rather than teaching a complete calculation procedure. Any implementation would need to consult the source definition and state its initial-state convention, since early updates can depend on that choice.

Key ideas

• The differential Sharpe ratio is raised as an online alternative to sliding-window Sharpe calculations.
• Its update uses recursively maintained state variables and the latest return.
• The document asks how to initialize those state variables at the first observation.
• No initialization rule or worked solution is provided.

Tags

# How to calculate "Differential Sharpe ratio"?


# How to calculate "Differential Sharpe ratio"?












Instead of using the "sliding the time window" method of calculating the sharpe ratio under online framework, they've defined "differential sharpe ratio" as such

But under equation 5, you can recursively calculate At and Bt given the current return and previous values of A and B But when t = 1, what are the initial values of At and Bt?

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.