Asset Pricing with Slope Factors : Model and Evidence of Outperformance

We make a case that characteristics-based long-short factors should be constructed by the slope factor method rather than by sorting methods. This is because sorting does not fully control for the influence of omitted characteristics, rendering them more noisy than slope factors. In contrast, slope factors, by virtue of being pure-play, lead to higher volatility stochastic discount factors, and better potential for pricing the cross-section. We show that an eight slope factor based asset pricing model captures risks that are mostly not priceable by existing models. It outperforms on several dimensions: factor risk-premia; OOS Sharpe-ratios of tangency portfolios; and pricing of portfolios, ETFs and stocks