Portfolio Formation Timing in Fama–French Factor Replication

Summary

This question examines whether daily returns for replicated Fama–French five-factor portfolios should begin on the portfolio formation date or after a one-day delay. The author describes applying a one-day lag to portfolios formed at June-end, which treats the positions as tradable at the close on July 1, and finding higher correlations with the official factor series when the lag is removed. The comparison is the document’s only reported evidence; it does not provide data, detailed implementation choices, or an explanation for the correlation difference.

The central issue is the gap between a factor’s accounting-based portfolio formation and a feasible trade. The author asks whether same-day returns reflect the official construction convention and whether researchers commonly use that convention in academic work. The document poses these questions but includes no answers, so it does not establish the Fama–French timing rule or a general academic practice. Replicators should investigate the source portfolio definitions and align formation, execution, and return measurement before interpreting correlations.

Key ideas

• The author compares factor replications with and without a one-day return delay.
• Removing the delay produces higher reported correlations with the official factor returns.
• The question highlights a potential mismatch between portfolio formation timing and executable trade prices.
• The document provides no resolution on the official convention or academic practice.

Tags

# Does fama-french factors apply 1-day delay between their portfolio formation and the trade? If so, why?


# Does fama-french factors apply 1-day delay between their portfolio formation and the trade? If so, why?












I'm currently replicating fama-french 5 factors using price and financial data.

I applied 1-day delay in calculating daily portfolio returns of factors, which assumes June-end portfolio to be traded on close price(precisely, total return index) of July 1st.

However, correlations between my replicated factors and official fama-french factor returns are higher when I remove the delay.

So I'm guessing that fama-french use 0-day delay in their portfolio return calculation.

I'm confusing because it seems not a good assumption of real investment or trading scenario for we usually can't trade a security on close price after the market closed.

Q1. Is there any good reason for using 0-day delay in making factors ?

Q2. Is it more common to use 0-day delay in academic ?

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.