tidy-finance/factor-library

该数据集包含50个金融风险因子在多种方法论规范下的多空投资组合月度回报，涵盖了1960年至2024年的美国股票数据。每个排序变量在所有有效的预处理选择（样本排除、滞后约定、断点定义、加权方案和再平衡频率）的组合下进行评估，每个变量产生数百种规范路径。排序变量的选择遵循Jensen、Kelly和Pedersen（2023）研究的特征横截面。数据集由150个Parquet文件组成，包含6列数据。

The dataset contains monthly long-short portfolio returns for 50 sorting variables commonly used in empirical asset pricing. Each sorting variable is evaluated across all valid combinations of preprocessing choices (sample exclusions, lagging conventions, breakpoint definitions, weighting schemes, and rebalancing frequencies), yielding hundreds of specification paths per variable. The selection of sorting variables follows the cross-section of characteristics studied by Jensen, Kelly, and Pedersen (2023). The dataset consists of 150 Parquet files with 6 columns.