Plugin Architecture : extend, don't fork.
StratCraft plugins follow a clean interface contract. Add new data sources, optimization algorithms, or execution adapters by implementing a standard plugin interface. The core engine remains unchanged.
Plugin Ecosystem · open sourceConnect
Connect
Everything
Extend StratCraft with exchange connectivity, advanced optimization, and systematic quant patterns: without touching the core engine.
CCXT Integration : one API, every venue.
Exchange connectivity through the CCXT library.
Use CCXT-supported venue adapters where the installed plugin and venue capabilities permit. Availability does not imply live deployment approval.
Install →
FrankWolfe Optimization
Convex optimization for portfolio weighting.
The FrankWolfe algorithm supports constrained convex optimization research. Its outputs remain candidates for evaluation, not production portfolio approval.
Install →
Simons-Style Systematic
Renaissance-inspired systematic quant patterns.
Explore systematic research patterns such as signal normalization, correlation filtering, ensemble weighting, and regime-aware position sizing.
Install →
The plugin contract
: five methods.
Implement one Python class against the typed Plugin protocol. The core registers it on startup; you get a slot in the Network graph for free.
# Implement one class. Register it. Done. from stratcraft.plugin import Plugin, PluginKind, Bars, Signal @Plugin.register(kind=PluginKind.SignalSource) class MySignal(Plugin): name = "my-signal" version = "0.1.0" def setup(self, cfg: dict) -> None: self.window = cfg.get("window", 20) def on_bar(self, bars: Bars) -> Signal | None: zscore = (bars.close[-1] - bars.close[-self.window:].mean()) return Signal(strength=zscore, ttl=bars.bar_seconds) def teardown(self) -> None: ...
Community Plugins plugins.
All plugins are open source. Browse the community repository, fork existing plugins, or contribute your own. The plugin ecosystem grows with every contributor.
DS
v 1.0.2Polygon Aggregates
by @tarunkrishna
Polygon.io 1m/5m/1d aggregates with on-disk Parquet cache. Resilient to rate limits.
DATAEQUITIES
★ ·↓ ·MIT
scp install polygon-aggregatesInstall →EM
v 0.9.1ML Ensemble Stacker
by @priyaq
Stack LightGBM + linear + simple-NN models with rolling cross-validation and per-bar refresh.
SIGNALSML
★ ·↓ ·MIT
scp install ml-ensembleInstall →RX
v 1.8.3DD-Aware Sizing
by @oolong-quant
Drawdown-aware position sizing that scales exposure inversely to rolling 60-day equity DD.
RISKSIZING
★ ·↓ ·MIT
scp install dd-awareInstall →TR
v 0.8.4TradeRepublic Bridge
by @helmrich
EU broker connector for TradeRepublic with order routing and tax-lot reporting.
EXECUTIONEU BROKER
★ ·↓ ·MIT
scp install tr-bridgeInstall →NB
v 0.2.1News-Bias Sentiment
by @ana-c
Local-only sentiment scorer over Bloomberg-style headlines using a 7B distilled model.
SIGNALSNLP
★ ·↓ ·MIT
scp install news-biasInstall →BL
v 1.7.3Black-Litterman View
by @wei-xu
Bayesian view-merging on top of MeanVar: confidence-weighted overlays without rewriting your prior.
OPTIMIZEBAYES
★ ·↓ ·MIT
scp install bl-viewInstall →Build a plugin in an afternoon.
The SDK ships with type stubs, a local registry, a hot-reload dev server, and golden test fixtures for every plugin kind.
Start Your Signal Factory
Free tier includes the C++ backtest engine, regime detection, and YFinance + Dukascopy data: everything you need to start building at scale.