Out-of-sample total R² for test-asset portfolio.

This table reports the OOS total R2 (%) for test-asset portfolios using observable Fama–French (FF) factor models and conditional autoencoder (CA) models (CA0 through CA3). It presents the result of applying latent factors (K) from 1 to 6 to each model. Panels A and B illustrate the results from equal-weight and value-weight portfolios, respectively.本表格报告了基于可观测的法玛-弗伦奇（Fama-French, FF）因子模型与条件自编码器（conditional autoencoder, CA）模型（CA0至CA3），对测试资产投资组合计算得到的样本外（out-of-sample, OOS）总R²（以百分比计）。本表格同时展示了为各模型应用取值范围为1至6的潜在因子（latent factors, K）后的计算结果。面板A与面板B分别对应等权重投资组合与市值加权投资组合的实验结果。