Building an Idiosyncratic Volatility Factor Portfolio

Summary

The document asks how to construct a factor from stocks ranked by idiosyncratic volatility after accounting for the Fama–French market, size, and value factors. The proposed long–short return subtracts the returns of low-volatility stocks from those of high-volatility stocks, with either percentile cutoffs or a median split as possible ways to form the groups.

The reply distinguishes academic factor construction from a simpler industry analysis. For an academic-style portfolio, it recommends forming six portfolios by combining size groups with high and low values of the signal, then averaging the small- and big-stock portfolio returns. For practical work, average excess returns across quintiles or deciles may be adequate. The response does not provide empirical evidence that this particular idiosyncratic-volatility factor earns a premium, nor does it specify a complete sorting or rebalancing protocol. Researchers would need to define the volatility estimate, sample, timing, and weighting choices to implement and evaluate the factor consistently.

Key ideas

• A high-minus-low idiosyncratic-volatility return can be used as a candidate factor.
• Academic factor construction can combine size groups with high and low signal groups.
• The suggested academic procedure averages returns from the small and big portfolios.
• Quintile or decile excess returns may be sufficient for an industry analysis.
• Portfolio formation details beyond the broad sorting approach are not specified.

Tags

# Constructing Idiosyncratic Risk Factor


# Constructing Idiosyncratic Risk Factor












I am studying idiosyncratic volatility. After applying the Fama Frech 3 Factor model with its Marktet, SMB and HML factors I want to build a factor based on idiosyncratic volatility.

Can I just build a portfolio including the highest idiosyncratic volatility assets and one with the lowest idiosyncratic volatility and subtract those from each other ?

My factor would then be:

IVOL factor = excess return of highest IVOL portfolios - excess returns of lowest IVOL portfolios

I could either used the 20% highest/lowest or just build the median and then assign them to one of the portfolios.

I am not sure if this approach is correct.

Best wishes

## Answer by stevew (score 1)

https://quant.stackexchange.com/a/60298

Are you doing this in an academic context? If so, the standard factor portfolio procedure is the Fama-French construct, which involves building 6 portfolios (split by size into 3/4/3 buckets and high/low based on your score) then average of the small and big. If you're doing this in an industry/work context, calculating average quintile/decile excess returns will suffice.

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.