Quant / 01

Factor library

A library of computable factors drawn from research papers and quant communities. Each entry records its formula, economic rationale, data requirements and information availability for use in subsequent research. The library contains 576 registered entries in 33 modules, covering intraday and daily data, with a separate macro module.

Registered entries · 33 modules: 576
Methods & tools: Python · NumPy / pandas · Causal features · DEAP / TCN

Research visuals

Factor library overview: registered entries, default inputs and macro factors · The top panels summarise all 576 registered entries, factor families selected for the default genetic-programming inputs, and the macro module. The middle groups the 33 modules by theme; the bottom shows the five macro categories. — **Factor library overview: registered entries, default inputs and macro factors**
The top panels summarise all 576 registered entries, factor families selected for the default genetic-programming inputs, and the macro module. The middle groups the 33 modules by theme; the bottom shows the five macro categories.
View full image ↗

Intraday factor seed library · This intraday snapshot contains 336 outputs across six groups: price and trend, volume, volatility, microstructure, statistics and context. Complete-linkage clustering at an absolute correlation threshold of |ρ| ≥ 0.90 selects 230 representatives to show redundancy among signals. The full candidate library is retained. — **Intraday factor seed library**
This intraday snapshot contains 336 outputs across six groups: price and trend, volume, volatility, microstructure, statistics and context. Complete-linkage clustering at an absolute correlation threshold of |ρ| ≥ 0.90 selects 230 representatives to show redundancy among signals. The full candidate library is retained.
View full image ↗

Microstructure factor seed library · The microstructure snapshot contains 68 outputs covering order books, trading activity, L1/L5 and LLOB. Complete-linkage clustering on Spearman correlations, at |ρ| ≥ 0.90, selects 39 representatives to show redundancy. The full candidate set is retained. — **Microstructure factor seed library**
The microstructure snapshot contains 68 outputs covering order books, trading activity, L1/L5 and LLOB. Complete-linkage clustering on Spearman correlations, at |ρ| ≥ 0.90, selects 39 representatives to show redundancy. The full candidate set is retained.
View full image ↗