Persistent Anomaly Risk Test
Test persistent market anomalies against real-economy risk factors
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 6
- Confidence
- 96%
Begin with a return pattern conventionally described as anomalous, then ask whether its persistence signals compensation for an omitted risk. Disaggregate the asset universe rather than relying on broad averages, identify economic variables that could transmit risk, and test the relationship across business-cycle conditions. For small caps, Vassalou computed default probabilities, sorted the cross-section, and observed how the effect changed as default risk rose or fell. The mechanism distinguishes disappearance from changing magnitude: an effect may become difficult to see across the full universe while remaining concentrated in its most exposed cohort. Behavioral forces can still matter, especially over shorter horizons, but persistent long-run performance should first be tested against measurable economic risk.
Origin
Maria Vassalou developed this research approach while studying the intersection of macroeconomics and finance before applying it to portfolio management. Extracted from Masters in Business.
Core principles
- 01Persistent anomalies may represent unmeasured risk rather than irrationality
- 02Real-economy variables can explain patterns in asset returns
- 03An effects magnitude can change without the effect disappearing
- 04Risk explanations and behavioral explanations can operate over different horizons
How to run it
- 1
Specify the anomaly
Define the claimed excess-return pattern, its comparison group, and the horizon over which it persists.
Pro tip Use a narrow, falsifiable definition instead of a broad market label.
Watch out A loosely defined anomaly can appear or disappear depending on portfolio construction.
- 2
Dissect the universe
Divide the relevant securities into cohorts or quantiles to locate where the effect is strongest.
Pro tip Inspect the extreme cohorts instead of relying only on aggregate results.
Watch out Broad averages can conceal a concentrated effect.
- 3
Propose an economic risk
Identify a real-economy variable that could plausibly create the return premium, such as default risk, GDP growth, investment growth, or productivity.
Pro tip Require a credible transmission mechanism from the economy to cash flows or discount rates.
Watch out Correlation without an economic mechanism is weak evidence.
- 4
Measure exposure
Estimate each assets exposure to the proposed risk and rank the cross-section accordingly.
Pro tip Use measures that can vary through time as economic conditions change.
Watch out Static classifications may miss cyclical changes in risk.
- 5
Test across regimes
Compare the anomalies magnitude during periods when the proposed risk is high and low.
Pro tip Use business-cycle phases as an out-of-sample economic check.
Watch out Do not interpret lower magnitude as proof that the effect vanished.
- 6
Compare competing explanations
Evaluate whether risk-based and behavioral mechanisms explain different horizons or parts of the evidence.
Pro tip Allow multiple mechanisms to coexist when the data support them.
Watch out Forcing a binary verdict may oversimplify a time-dependent effect.
In the wild
Rank small-cap companies by model-implied default probability, form portfolios from the rankings, and compare their returns through expansions and downturns. If the premium broadens and strengthens when default probabilities rise but remains concentrated among the smallest firms in calmer periods, treat it as cyclical compensation for default exposure rather than a vanished anomaly.
→ The investor identifies where the premium resides and when its magnitude should increase.
Compare value and growth portfolio performance with changes in real GDP and sector investment growth. Examine whether muted growth expectations coincide with stronger value performance and whether rapid expansion changes the relative advantage.
→ The style allocation is tied to an explicit macroeconomic mechanism rather than a static label.
Common mistakes
Testing only the aggregate portfolio
The strongest exposure may exist only in an extreme cohort, so an aggregate test can incorrectly suggest that the effect disappeared.
Calling every pattern behavioral
A persistent return premium may compensate investors for an economic risk that the original model omitted.
Demanding one universal explanation
Risk-based mechanisms may dominate over long horizons while behavioral forces drive shorter deviations.
Is it for you?
Best for
It is best for quantitative investors investigating durable patterns such as small-cap, value, or momentum effects.
Not ideal for
It is not ideal for short-lived price moves lacking sufficient historical or fundamental data.
From the transcript
“if an anomaly persist and it doesn't go away, then maybe it's not an anomaly, maybe it's risk based.”
“the small cap effect always exists in the smallest of the small caps and it's related to default risk.”
“I think the risk based explanations need longer time to play out some of this behavioral driver, some more short term drivers.”
From the episode
Maria Vassalou on the Small-Cap Effect
Maria Vassalou