Skip to content

Notebooks

Runnable Marimo notebooks demonstrating QuantWave end-to-end. Each page below is a landing summary; run locally for full interactivity and native Rust performance.

Recommended order

  1. Strategy Backtest — indicator → signal → .bt
  2. Backtest Showcase — sweeps, WFO, Monte Carlo
  3. ML Features → Backtest E2E — feature parity into trades

Backtest & portfolio

Available Notebooks

  • Strategy Backtesting
    Steel-thread example using indicators inside the vectorized backtester with rich signal metadata.

  • Backtest Engine Showcase
    Comprehensive tour of the Polars-native backtester: param sweeps, walk-forward optimization, fast metrics, sizing filters, and cross-sectional panels.

  • Backtest Engine Benchmarks
    Criterion harness comparing quantwave-backtest vs naive row-loop baselines (10K–1M rows, multi-symbol).

  • Portfolio Shared Capital
    Multi-symbol book simulation with one cash pool via .bt.portfolio_backtest().

  • Execution-Aware Research — Canonical execution-realism example
    First-class order types (limit / stop / stop-limit) via .bt.order_backtest(), risk overlays (risk_model=), and benchmark-relative reporting (alpha / beta / Calmar / VaR / CVaR) — all on the batch ↔ streaming parity core.

Indicators & ML

  • Multi-Indicator Analysis Clean chaining of multiple indicators (SMA, EMA, Momentum, SuperTrend, etc.) in one lazy Polars expression.

  • ML Feature Stability & Tiny Model — Canonical example
    Builds feature matrices from the new toolkit, proves batch/streaming parity + no-lookahead, trains a tiny regime+direction model with per-regime metrics.

  • ML Features → Realistic Backtest (E2E) — Primary cross-epic reference (closed epics 4ps + gwx)
    See also the ML Features guide for Polars/streaming patterns.
    End-to-end demonstration of the locked features surface feeding the backtester. Shows batch vs streaming parity with rich metadata preserved all the way into trades.

  • PA Foundation Strategy (MarketStructure + Flags/H&S)
    Production-ready surface for the MQL5 PA toolkit foundation (Parts 21/66/69: swings/bias/flips + geometric). Realistic strategy: bull Flag breakout only on confirmed bullish MarketStructure + regime + ML (hurst) filter, dynamically sized from pole_length_atr rich metadata. Python streaming + Polars Rust paths + backtester sketch. Synthetic + notes for real data. See also the four dedicated PA guides under Native Indicators.

  • PA Flag Breakout Canonical E2E
    Runnable Marimo notebook demonstrating the exact production pattern for the PA geometric tools, complete with sizing via pole_length_atr and confirmed market structure filters.

How to run any notebook locally

# Recommended
pip install "quantwave[all]" marimo polars numpy

# Or from source after building the Python bindings
maturin develop -p quantwave-python --release
pip install marimo polars numpy

marimo edit docs/examples/notebooks/<notebook_name>.py

Why some notebooks show limited content here:
The live documentation site is static (GitHub Pages). Notebooks that depend on QuantWave's native Rust extensions cannot execute inside the browser. The pages above give you context + the exact commands to run the real interactive versions locally.