Skip to content

ROADMAP.md

Latest commit

 

History

History
68 lines (56 loc) · 3.19 KB

ROADMAP.md

File metadata and controls

68 lines (56 loc) · 3.19 KB
Raw

riskkit roadmap

The vision: riskkit should become the default, framework-agnostic risk layer for systematic trading in Python — the dependable answer to "how much do I trade, when do I cut size, and when do I stop?", usable from any backtester or live bot. Backtesters handle signals and execution; riskkit handles survival.

The six core components — position sizing, drawdown control, stops, correlation limits, session/tilt guards, and a pre-trade validator — landed in v0.2. From here the focus is making them effortless to adopt and deeper where it counts.

v0.3 — Adoption & ergonomics

  • RiskManager façade ✅ shipped — wires all six components from a single config; one evaluate() call sizes + validates a trade: 
    from riskkit import RiskManager, RiskConfig, TradeIntent

risk = RiskManager(RiskConfig(
    base_risk_pct=1.0, max_notional_pct=4.0,
    drawdown=dict(tier1_pct=3, halt_pct=10),
    session=dict(max_trades_per_day=5),
))
risk.on_equity(equity)                       # refresh drawdown/session state
decision = risk.evaluate(TradeIntent(...))   # size + validate in one call
if decision.ok:
    place(decision.units, decision.stop)
  • First-class framework adapters:
    • backtesting.py ✅ shipped — the RiskkitStrategy mixin auto-sizes and validates entries via RiskManager (riskkit.adapters.backtesting).
    • freqtrade ✅ shipped — FreqtradeRiskManager drives custom_stake_amount / confirm_trade_entry (riskkit.adapters.freqtrade).
    • vectorbt ✅ shipped — size_signals sizes vectorized entry signals (riskkit.adapters.vectorbt).
  • Config presets ✅ shipped — RiskConfig.conservative() / .balanced() / .aggressive() (and RiskConfig.preset(name)); load from a dict (from_dict) or a YAML file (from_yaml).

v0.4 — Depth & correctness ✅ shipped

  • More sizers ✅ shipped — volatility targeting (volatility_target_size), risk parity (inverse_vol_weights), and full/fractional Kelly (kelly_fraction).
  • More stops ✅ shipped — chandelier, parabolic SAR, and structure-based, added to the StopEngine stack (all tighten-only).
  • Portfolio-level risk ✅ shipped — total open "heat" (max_portfolio_heat_pct), per-sector / asset-class caps (max_exposure_per_sector_pct), and VaR/CVaR (value_at_risk / conditional_value_at_risk).
  • Property-based tests (hypothesis) for the core invariants ✅ shipped — never increase risk after a loss or deeper drawdown, never exceed a configured cap (notional / heat / sector), and stops only ever tighten.

v0.5 — Observability & docs

  • A decision / audit-trail object with optional CSV / pandas export.
  • Deployed docs site ✅ live at https://hasibvortex369.github.io/riskkit/ — recipes and a short "risk 101" guide still to expand.
  • Benchmark notebooks quantifying riskkit's effect (e.g. drawdown reduction) on real strategies.

v1.0 — Stability

  • Frozen, semver-committed public API.
  • Published to PyPI under a clean distribution name.
  • A small cookbook of example repos.

Feedback and contributions are welcome — see CONTRIBUTING.md and open an issue.