Pricing Error Reversal: A Diagnosis of Asset Pricing Models

Abstract

We study the trading profitability of pricing errors from six well-known factor models, and find a pricing error reversal pattern. The spread portfolio from sorting the standardized pricing errors earns sizable abnormal profits, suggesting that the models are problematic. We apply the spread portfolio diagnosis further to recently developed machine learning models, and find similar pricing error reversal patterns, raising questions on these major sophisticated models. Additional analysis reveals that the significant trading profits from the pricing error reversal pattern cannot be explained by investor sentiment, limits-to-arbitrage, prospect theory, and expectation extrapolation.

Keywords: Asset pricing tests, factor models, machine learning, pricing errors

JEL Classification: C53, G11, G12, G17