Pukthuanthong · Roll · Subrahmanyam (2019)

PRS Protocol Platform

Factor Identification & Classification · Review of Financial Studies
Three-Condition Test Covariance · Pricing · Sharpe Bound

Is your factor a genuine risk premium,
an anomaly, or just noise?

The PRS Protocol is the academic standard for distinguishing genuine risk factors from characteristics, anomalies, and spurious predictors. Upload monthly returns and get a rigorous three-condition classification in seconds.

Used by researchers worldwide to validate new factors before submission to top finance journals.

C1
Covariance Matrix Relation
Factor must be related to the principal components of asset return covariance matrix via canonical correlation (Connor-Korajczyk 1988)
C2
Cross-Sectional Pricing
Factor must command a significant risk premium in Fama-MacBeth regressions with Shanken (1992) correction on individual securities
C3
Reasonable Sharpe Ratio
Reward-to-risk ratio must not exceed the MacKinlay (1995) bound of 0.60. Factors beating this bound are anomalies, not risk factors

Test Your Factors

Upload monthly returns — supports FF5, JKP, or any wide-format CSV
or
Accepted Formats
Fama-French style (wide, all factors in one file):
dateMkt-RFSMBHMLRMWCMA
199001-7.89-0.80-2.042.45-0.78
1990021.310.56-1.15-0.530.28
..................
✓ Dates: 199001 or 1990-01-31   ✓ Returns: percent (e.g. 1.23) or decimal (e.g. 0.0123) — auto-detected
✓ All numeric columns after date are treated as separate factors
✓ Columns named "RF", "rf", or "risk_free" are excluded automatically

Why Use the PRS Protocol?

With over 400 factors proposed in the literature, distinguishing genuine risk factors from data-mined characteristics is the central challenge in empirical asset pricing. The PRS Protocol provides the only complete three-condition framework grounded in theory.

Unlike simple t-statistic screens, the protocol connects factors to the economic structure of returns through the covariance matrix — ensuring only factors with systematic, pervasive influence on all asset prices can pass.

  • →For researchers: Validate new factors before journal submission. Know whether your factor is a genuine risk premium or an anomaly requiring a behavioral explanation.
  • →For asset managers: Distinguish risk factors (for portfolio construction) from anomalies (for alpha generation). The classification determines the correct theoretical framework for trading.
  • →For allocators: Evaluate factor exposure in smart-beta products. A factor with Sharpe above the MacKinlay bound signals crowding risk and potential for regime-change collapse.
  • →For regulators: Identify which return premiums represent genuine compensation for systematic risk versus behavioral phenomena that may be exploited at investors' expense.

Classification Guide

  • ■ Risk Factor — Passes all three conditions. Related to covariance, priced in cross-section, reasonable Sharpe. Represents compensation for systematic risk.
  • ◆ Anomaly — Sharpe exceeds MacKinlay bound. Too profitable to be risk. Represents a trading opportunity — but may be arbitraged away.
  • △ Weak Signal — Marginally passes covariance condition but premium is statistically weak. May be time-varying or sample-specific.
  • ○ Unpriced / Noise — Fails necessary condition. Not systematically related to return covariation. Likely data mining.
Cite as
Pukthuanthong, K., Roll, R., and Subrahmanyam, A. (2019). A Protocol for Factor Identification. Review of Financial Studies, 32(4), 1674–1707.