SIAM Journal on Financial Mathematics, Volume 11, Issue 3, Page SC-26-SC-37, January 2020.
Deep fundamental factor models are developed to automatically capture nonlinearity and interaction effects in factor modeling. Uncertainty quantification provides interpretability with interval estimation, ranking of factor importance, and estimation of interaction effects. With no hidden layers we recover a linear factor model, and for one or more hidden layers, uncertainty bands for the sensitivity to each input naturally arise from the network weights. Using 3290 assets in the Russell 1000 index over a period of December 1989 to January 2018, we assess a 49 factor model and generate information ratios that are approximately 1.5x greater than the ordinary least squares (OLS) factor model. Furthermore, we compare our deep fundamental factor model with a quadratic LASSO model and demonstrate the superior performance and robustness to outliers. The Python source code, Python notebooks, and the data used for this study are provided at https://github.com/mfrdixon/Deep_Fundamental_Factors.

中文翻译：

---

简短的交流：深层基本因素模型

SIAM金融数学杂志，第11卷，第3期，第SC-26-SC-37页，2020年1月。
开发了深层的基本因子模型，以自动捕获因子建模中的非线性和相互作用效应。不确定性量化提供了区间估计，因子重要性排名和相互作用影响估计的可解释性。在没有隐藏层的情况下，我们恢复了线性因子模型，对于一个或多个隐藏层，对于每个输入的灵敏度的不确定带自然会来自网络权重。使用1989年12月至2018年1月期间Russell 1000指数中的3290项资产，我们评估了49个因子模型，并得出的信息比率大约是普通最小二乘（OLS）因子模型的1.5倍。此外，我们将深层基本因子模型与二次LASSO模型进行了比较，并证明了其对异常值的优越性能和鲁棒性。