FinQuant
A multi-asset workbench for quant PMs · data · trade · backtest · risk loop
What it is
FinQuant is a multi-asset workbench for solo and small-firm quant PMs. One interface, one data pipeline, one backtest engine, covering bonds / equities / FX / derivatives / mixed portfolios — all five major asset classes. Self-hosted: data and strategy code stay on your own machine.
Core modules
- Five asset classes, full loop each: bond / equity / fx / deriv / mixed — each ships its own
/data/trade/backtest/risksub-modules, no cross-asset bleed - Data layer: market / valuation / factor data unified ingestion, multi-source fallback; local ClickHouse / DuckDB storage, zero cloud dependency
- Backtest engine: event-driven + vectorized dual mode, with a 42-case performance baseline; tick / daily / fill-level granularity
- FX wizard: a 5-step GUI wizard — pick pair / pick strategy family (static / momentum / mean reversion) / set params / run backtest / see report — no code required
- Risk: VaR / ES / stress / position limits, configured per asset class
Who it’s for
- Solo quant PMs: you have your own strategy ideas, want a workbench for rapid validation, and refuse to upload strategy code to the cloud
- Small-firm research desks: daily backtest reports / factor decay tracking / IC committee material, with a shared workbench that crosses asset classes
- Quant PM beginners: an FX wizard and starter library — start from config-driven backtests, no need to write an engine from day one
FinQuant does not
- Send real orders (no broker integration) — the workbench stops at backtest and simulation; real execution is your trading system’s job
- Recommend strategies / publish factor rankings / run community leaderboards
- Collect your strategy code, positions, or orders
- Require online activation
Status
POC stage. All five asset classes have working /data + /trade + /backtest + /risk skeletons. FX wizard ships 3 strategy families (static / momentum / mean reversion). 114 smoke tests + 42 performance-baseline tests all green. Next: real FX data source integration + POC demo polish.
Tech stack
Python backend + Rust performance kernel (PyO3 bindings) + ClickHouse / DuckDB data layer + React frontend. uv workspace monorepo.