Standardized Earnings Surprise (SES) is a classic measure of the post-earnings price drift effect (PEAD). The PEAD effect refers to the fact that when a company's actual earnings exceed (or fall short of) analysts' expectations, the stock price will continue to adjust upward (or downward) in the weeks or even months after the earnings announcement. SES uses the standard deviation of earnings surprises over the past eight quarters to standardize the current quarter's earnings surprises, which helps identify earnings surprises that are truly above or below historical levels. Generally speaking, a higher SES value indicates a positive price drift effect, while a lower SES value indicates a negative price drift effect. This factor can be used as an earnings quality signal and combined with other factors in a quantitative stock selection model to improve the return of the portfolio.

More detailed explanation:

• Earnings Surprise: Using the difference between the earnings per share of the current quarter and the same period four quarters ago can effectively eliminate the seasonal impact of corporate earnings, thereby more accurately reflecting the actual performance of the current period.
• Standardization: Standardizing earnings surprises by dividing by the standard deviation of earnings surprises over the past eight quarters makes earnings surprises comparable across companies and over time. This helps identify companies that have significant deviations from historical levels.
• PEAD Effect: The phenomenon of price drift after earnings announcements is related to cognitive biases in behavioral finance, such as the anchoring effect. Investors tend to underreact to earnings surprises, resulting in gradual price adjustments. The SES factor can be used to capture investment opportunities resulting from this underreaction.
• Practical Application: In practical applications, investors can sort stocks according to the size of SES, select high SES stocks for investment, or use SES as an important factor in quantitative portfolio models.