SYNOPSIS Using stepwise logit regression, Ou and Penman (1989) predicts the sign of future earnings changes and uses these predictions to form a profitable hedge portfolio. Increases in computing power and advances in machine learning allow us to extend Ou and Penman (1989) using more data, computer intensive forecasting algorithms, and modern prediction models. Stepwise logit still provides good predictions and can be used to form a trading strategy that generates small abnormal returns, but random forest significantly improves forecast accuracy and returns. The models identify different variables as being important for prediction in high tech and manufacturing, but this does not lead to better predictions or higher returns. Results confirm Ou and Penman's (1989) finding that financial statement information is useful for investment decisions, and suggest that machine learning techniques can be useful in a variety of accounting contexts.

中文翻译：

---

通过机器学习改善盈利预测和异常回报

概要 使用逐步 logit 回归，Ou 和 Penman (1989) 预测未来收益变化的迹象，并使用这些预测来形成一个有利可图的对冲投资组合。计算能力的提高和机器学习的进步使我们能够使用更多的数据、计算机密集型预测算法和现代预测模型来扩展 Ou 和 Penman (1989)。逐步 logit 仍然提供了很好的预测，可以用来形成一个产生小的异常回报的交易策略，但是随机森林显着提高了预测的准确性和回报。这些模型将不同的变量确定为对高科技和制造业的预测很重要，但这并不会带来更好的预测或更高的回报。结果证实了欧和彭曼'