Testing Whether Factor Returns Decay After Portfolio Rebalancing

Summary

The document examines whether factor portfolios built from recently updated financial statements outperform portfolios using older information. Using factor return data from Professor French’s database, the author compares returns during the first six months after rebalancing with returns during the second six months. The reported pattern runs against the expectation that information is most valuable soon after it becomes available: roughly 80% of the datasets show higher factor returns in the second half, though most differences are not statistically significant. Small-minus-big is singled out for negative first-half and positive second-half returns across the tested datasets.

The author suggests that annual reports arriving during the latter part of the measurement period might help explain the pattern and asks for alternative explanations or relevant research. A response offers the general idea that factor information loses value over time, implying a tradeoff between rebalancing frequency, transaction costs, and the benefit of acting on fresher information. The discussion is exploratory: it supplies no formal test of that explanation, and the observed return differences are mostly insignificant.

Key ideas

• The study compares factor returns in the first and second six-month periods after rebalancing.
• Most tested datasets showed higher returns in the second half, contrary to the author's expectation.
• The reported differences were usually not statistically significant.
• Small-minus-big had negative first-half and positive second-half performance in each tested dataset.
• Factor decay can motivate a rebalancing tradeoff between information freshness and transaction costs.

Tags

# Should factor signals decay?


# Should factor signals decay?












If the factors have information content, I expected factor portfolios constructed using recent financial statements should outperform portfolios using stale data.

To test this, I used the Professor French’s database. I compared the factor returns in the first 6 months after rebalancing against the returns in the second half.

In about 80% of the data sets, the factor returns were higher in the second half of the year. In most cases the difference was not significant. Small minus big stands out because it had negative first half performance and positive second half performance in every data set I tested.

This is the opposite of the result I expected. I expected most to be insignificant or higher returns in the first 6 months after rebalancing.

Maybe it could be explained by new annual reports coming out in the second 6 months (first 6 calendar months). Are there other, better explanations / other papers I should read that may help me understand these results?

My script is on GitHub, here: https://github.com/Charles-Fox1234/FactorResearch

## Answer by user36511 (score -1)

https://quant.stackexchange.com/a/42586

In APT, factor decays. So you have to choose a appropriate rebalancing interval - a trade-off between transaction cost and time value of information.

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.