When it comes to stock market analysis, the traditional focus has often been on forecasting returns. Yet, there is a growing interest in a different approach that centres around predicting the probability of stocks outperforming a benchmark. This method leans on a substantial set of firm characteristics as predictors, much like those used in Machine Learning (ML) studies.

Why is this shift significant? Because the probability associated with a stock’s return is insightful in its own right. It captures the information ratio of the stock, a metric that is not just about the expected return but also integrates risk. The information ratio provides a more comprehensive view by comparing the return of an asset against the return of a benchmark, adjusted for the volatility of the excess returns.

The beauty of this probability-based approach is that if you have the ex-ante probability, sorting stocks by this probability is tantamount to sorting by the information ratio. This sorting could be more advantageous than relying solely on expected returns because it accounts for the quality of the return in terms of both return and risk.

In practical applications, however, the performance of this probability-based approach can vary. It is contingent upon how accurately both the probability and the information ratio are estimated. The efficacy of this method hinges on the precision of individual estimates of return and risk. In scenarios where these estimates are precise, the probability-based approach can provide a distinct edge over traditional return forecasts.

Moreover, this approach can be particularly useful for investors who are looking to construct a portfolio with a desired level of risk. Instead of focusing solely on the potential returns, they can consider the likelihood of outperforming a benchmark while also keeping the risk in check.

In essence, while forecasting returns might give you a one-dimensional view of the potential performance of stocks, incorporating probability forecasts and information ratios can offer a more nuanced and risk-aware perspective. As the financial world continues to evolve with more sophisticated models and methods, the probability-based approach is a promising avenue for those looking to make more informed investment decisions.