跳转至

Shadow Factor

shadow-factor is the factor engine in the Shadow stack.

It turns financial and market data into factor-ready outputs, and provides a factor DSL for generating new factors from existing fields and operators.

What It Is For

Use shadow-factor when you want to:

  • query factor panels as final dataframes
  • treat source data as factor inputs instead of manually preprocessing wide tables
  • define new factors through a compact DSL rather than custom imperative code
  • register reusable factors for repeated research or production use
  • keep factor logic readable, reviewable, and composable

Stable Public API

The current stable public API is intentionally small:

  • panel(...) — evaluate one or more factors into a dataframe panel
  • register_factor(...) — register a named factor formula for reuse
  • list_factors(...) — inspect registered factors and metadata

Supported Factor Surface

当前内置支持面包括：

  • 100 个标准目录因子，可通过 list_factors(...) 发现
  • 343 个基础字段因子，覆盖日频行情字段和财报字段
  • 317 个财报字段家族，按利润表、资产负债表、现金流量表组织，并自动展开原值、TTM、同比、环比等变体

完整列表见：

Core Ideas

Data as a factor

In shadow-factor, base fields are not just raw columns to export and manually transform later.

They are the starting point of factor construction.

That means you can treat financial fields such as net_profit, operating_revenue, or total_assets as building blocks for higher-level factors.

Factor DSL as a generator

Instead of writing one-off calculation scripts, you can define factor logic as formula strings.

Examples:

"TTM(net_profit)"
"YoY(TTM(net_profit))"
"SafeDiv(TTM(net_profit), total_assets)"

This makes factors easier to register, compare, version, and reuse.