Research and projects

Market studies.
Factor research.

Explore published market and factor studies, compare their results and follow the full methods and evidence.

Published research

Two views of published aggregate results, with their sample definitions and evidence limits. Read the full studies for methods and research records.

Interpretation text reviewed October 30, 2026. Each study states its sample dates separately.

01 · Retrospective comparison

18-year A-share market style analysis

The historical study forms five groups each month. The chart shows synthetic long-short returns; each factor’s score direction needs a separate check.

Coverage varies by factor. Historical formulas and point-in-time availability remain incompletely verified. Current effects require later samples.

Methods and sources

Yearly returns use the source period_return field; first and last years may be partial. The full study retains factor directions, definition checks and original sources. Synthetic long-short returns and account returns use different conventions.

02 · Retrospective comparison

PB and ROE rankings

PB is price-to-book and ROE is return on equity. All five rankings share an eligible pool; the PB-only comparison also requires computable ROE.

The entire period informed exploration. Historical filing availability and execution remain incompletely verified; present applicability needs later samples.

Methods and sources

Historical statement revisions and publication availability have not been independently checked filing by filing. Execution uses daily approximations without full market impact or real-fill replay.

How the projects work together
  1. Prepare dataThe market data platform collects, checks and publishes versioned datasets.
  2. Test ideasResearch projects consume the data, the shared framework provides portfolio and risk tools, and the runtime executes jobs.
  3. Publish resultsThe research site, factor observatory and market reports each present their reviewed material.

Strategy details and complete experiment records remain in private research. Public pages describe their samples, methods and limitations. Demonstration data, historical backtests and operational results should be read separately.

Data and backtesting

Factors and models

Factors and models

Classic factor research

money-trees

Factor computation, validation, mining and rolling backtests across Alpha101, Alpha191, Alpha158 and Alpha360.

Factors and models

Deep learning for market data

quant-deep-learning

Studies deep-learning models for order books, minute data and L2 event streams, currently focusing on next-day cross-sectional prediction for A-shares.

Research and reports

Research and reports

Market report platform

quant-intel-platform

Collects market data and news, validates research artifacts, and generates and delivers reports. The separate market report website publishes public reports.

Trading and experiments

Trading and experiments

Trading research workbench

quant-trading-workbench

A web dashboard for A-share, Hong Kong and US research snapshots, intraday ideas and simulated portfolios. Static demo data may lag; see the project’s scope notes.

Trading and experiments

A-share animal and plant indices

a-share-zoo-garden

Builds thematic indices from animal and plant terms in historical company names, publishing constituents, methods and comparisons with the CSI 300.

Trading and experiments

Language-model stock ranking

ai-stock-picker

Uses a language model to rank an existing shortlist, validates that selections stay within the pool, and records inputs, results and model calls.

More code and experiments: GitHub repositories ↗