Skip to content
All library documents

Why High-Minus-Low Portfolios Reveal Cross-Sectional Effects

Article Quant Q&A · Author: town
Read the original

Summary

High-minus-low analysis compares returns from portfolios formed at opposite ends of a chosen stock characteristic. The spread tests whether the characteristic is associated with differences in average returns over time, with extreme groups making the contrast in the sorting variable largest. Researchers commonly assess whether the spread’s time-series mean differs from zero.

The document emphasizes that portfolio sorts are nonparametric: they do not require a specified functional form for the relationship between characteristics and returns. This makes them useful for detecting nonlinear patterns that a parametric regression might miss. It also notes that portfolio analysis can show how characteristics vary across the sorted groups. The discussion is conceptual and cites an asset-pricing textbook; it gives no empirical example or estimate. A high-minus-low spread is evidence of an association under the chosen sort, but the document does not establish causality or address implementation details such as transaction costs.

Key ideas

  • High-minus-low analysis compares average returns across portfolios sorted on a characteristic.
  • The difference portfolio tests whether the return spread is statistically distinguishable from zero.
  • Extreme portfolios maximize the contrast in the sorting variable.
  • Portfolio sorts are nonparametric and can reveal nonlinear cross-sectional relationships.
  • Portfolio analysis can also describe how characteristics vary across sorted groups.

Tags

EquitiesFactor investingStatistics
Full text
# What is the benefit of High-minus-Low as in Fama French model?


# What is the benefit of High-minus-Low as in Fama French model?












Can anyon explain the concept of using High-minus-Low in finance literature.

## Answer by skoestlmeier (score 2, accepted)

https://quant.stackexchange.com/a/43766

"High-minus-Low" refers to portfolio analysis, which is one of the most commonly used statistical methodologies in empirical asset pricing. There are several benefits of this technique in comparison to regression-models presented in Bali/Engle/Murray (2016), p. 33:

> While the most common application of portfolio analysis is to examine future return predictability, the portfolio methodology can also be employed to understand variation in the characteristics of the entities (stocks) across the different portfolios. Perhaps the most important benefit of portfolio analysis is that it is a nonparametric technique. This means that it does not make any assumptions about the nature of the cross-sectional relations between the variables under investigation. In fact, portfolio analysis can be helpful in uncovering nonlinear relations between variables that are quite difficult to detect using parametric techniques.

The main statistical argument is, that we frequently want to test whether the time-series mean for each of the portfolios differs from some null hypothesis mean value (wich is often assumed to be zero). Most importantly, we want to examine whether the time-series mean of the difference portfolio is statistically distinguishable from zero. As commented, we commonly use the most extreme portfolios, to test if a certain cross-sectional relation between stocks exists, because the difference for the control-variable (sorting-variable) is highest for these extreme portfolios.

Reference

Bali, Turan G., Robert F. Engle, and Scott Murray (2016): Empirical asset pricing: the cross section of stock returns. John Wiley & Sons

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.