Choosing Historical Volatility Measures to Match the Price Model

Summary

The document asks why historical volatility estimates can differ when calculated from closing prices, high-low ranges, open-high-low-close data, or returns. Its central guidance is that the measure should match the quantity modeled: use price volatility when the model describes asset prices, and return volatility when it describes returns. High-low range can serve as a volatility proxy, though the answer characterizes that approach as relatively unsophisticated.

The response does not identify one universally preferred estimator or compare methods empirically. It offers a model-alignment principle rather than detailed formulas, assumptions, or selection criteria. In practice, the choice therefore depends on the model and the intended use, and the note alone does not explain how sampling frequency, jumps, or data quality affect the estimates.

Key ideas

• Different volatility calculations measure variation in different representations of price behavior.
• Choose price volatility when the model is formulated in prices and return volatility when it is formulated in returns.
• High-low ranges can approximate volatility but are described as a basic proxy.
• The document gives no universally preferred method or detailed empirical comparison.

Tags

# Why are there so many different ways of calculating historical volatility


# Why are there so many different ways of calculating historical volatility












There appears to be several ways of calculating volatility:

- Price volatility (of which there are several variants): close to close high low range average of open, high, low and close

- Log returns volatility

My question(s) are:

a). Why are there so many different ways of calculating volatility (since the method used sometimes changes the numbers dramatically)?

b). When is a particular method preferred?

## Answer by SRKX (score 4)

https://quant.stackexchange.com/a/2945

The question barely fits the Q&A format.

Volatility is something which is quite abstract.

Basically the type of volatility you use depends on the model you chose to implement.

If you model asset prices, you will use price volatility.

If you model return prices, you will use returns volatility.

High-low price range can be used as a proxy for the volatility, but it's pretty unsophisticated.

In short, there is no real preferred method, the way you compute volatility depends on the model you use.

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.