Replicating Anomalies
Read the paperopens doi.org in a new tab
What they found
The authors attempted the largest replication in finance: 452 anomalies from the published literature, re-tested with consistent methods, value-weighted portfolios, and NYSE breakpoints so that microcaps do not dominate. Under these more realistic procedures, 65% of the anomalies failed to be statistically significant, and even among the survivors, the magnitudes were much smaller than originally reported. Trading frictions and microcap stocks were the main culprits; the strongest survivors were momentum, profitability, investment, and some accrual and valuation measures.
What you can use
- Two-thirds of published anomalies fail replication once you stop letting tiny illiquid stocks drive the results.
- Ask of any factor or screen: does it work in stocks big enough for you to trade? Often the answer is no.
- The survivors (momentum, profitability, investment) are the ones to build around.
Caveats
Their choice of value-weighting and NYSE breakpoints is itself a methodological decision that some anomaly authors dispute. U.S. data 1967 to 2016. A free NBER version exists.
Tags: anomalies, replication, microcaps, factor-zoo
Summaries are our own reading of the paper, not the authors' words. Educational only, not advice. Discuss it in Book Club.