kairos-engine 0.3.3

Kairos — adaptive crypto trading engine. Curates the best of the open-source ecosystem and adds a learning layer on top.

Kairos

An event-driven trading engine for Python — live execution, backtesting and paper trading with one API. Built for spot crypto, designed to scale to multi-venue and derivatives.

---

Installation

pip install kairos-engine

Requires Python 3.11 or newer.

Quick start

from kairos import LiveEngine, LiveStrategy, BinanceLive
from kairos.types import ActorConfig


class MeanReversion(LiveStrategy):
    def on_start(self):
        self.sma = self.indicator.sma("close", window=20)
        self.rsi = self.indicator.rsi(window=14)

    async def on_bar(self, bar):
        if self.rsi.value < 30 and bar.close < self.sma.value:
            await self.buy_bracket_pct(
                pct=0.10,          # 10% of USDC balance
                sl_atr_mult=2.0,   # stop-loss = entry - 2×ATR
                tp_atr_mult=3.0,   # take-profit = entry + 3×ATR
            )


engine = LiveEngine()
engine.register_adapter(BinanceLive(api_key="...", api_secret="..."))
engine.add_strategy(
    MeanReversion(ActorConfig(symbol="BTCUSDC", timeframe="1h")),
)
await engine.run()

A ten-minute tutorial lives in docs/getting-started.md.

What's in the box

Area	Capability
Runtime	Async event loop · actor model · market-data cache · warmup gating
Strategies	LiveStrategy base class · 12 built-in indicators · bar + tick hooks
Execution	Market / limit / stop / bracket orders · OCO · atomic entry+SL+TP via BracketManager
Adapters	BinanceLive (Spot, REST + market WS + user-data WS) · PaperAdapter for sims
Backtesting	BacktestEngine with parquet catalog · Sharpe / Sortino / profit factor
Reconciliation	Fill parity tooling (kairos.parity.match_fills) to diff engine outputs
Ops	Structured logging · graceful shutdown · reconnect with exponential backoff

227 tests cover the public surface; CI runs on Python 3.11, 3.12 and 3.13.

Design goals

1. Stable primitives, pluggable policy. The engine handles the boring parts (event loop, reconnects, order state, reconciliation) so strategies stay small and testable.
2. Same API for backtest, paper and live. A strategy you write for backtesting runs unchanged against a paper adapter and then against a live exchange.
3. Composable actors. Risk limits, parameter tuning, notifications and regime detection are actors subscribed to the engine bus. Add your own by subclassing kairos.Actor.
4. Readable over clever. Typed public surface, minimal magic, every decision documented in source.

Example — Actor for risk halt

from kairos import Actor


class DailyLossLimit(Actor):
    def __init__(self, max_loss_pct: float = 3.0):
        super().__init__()
        self._max_loss = max_loss_pct
        self._start_equity: float | None = None

    async def on_event(self, kind: str, event):
        if kind == "equity_update":
            if self._start_equity is None:
                self._start_equity = event.equity
                return
            drawdown = (self._start_equity - event.equity) / self._start_equity
            if drawdown >= self._max_loss / 100:
                await self.publish_signal("risk_halt", reason="daily_loss")

The engine's SignalDispatcher picks up risk_halt and stops new entries until the actor publishes a resume signal.

Status and versioning

Current release: v0.3.1 — production-deployed as the execution core of Trading Autopilot since 2026-04-13, powering DCA strategies on BTCUSDC and ETHUSDC on Binance Spot.

Semantic versioning. Pre-1.0 releases may still contain minor breaking changes; the CHANGELOG calls them out explicitly.

Roadmap

Milestone	Focus
v0.4	Adaptive execution — order-type + timing learned per (symbol × regime × spread)
v0.5	Continual parameter tuning — Bayesian posterior updated per trade
v0.6	Multi-venue — Kraken adapter + venue-registry abstraction
v1.0	Stable public API · long-term support

Details in docs/roadmap.md.

Documentation

• Getting started — first strategy in ten lines
• Architecture — runtime, actors, cache, execution
• Strategy guide — patterns and anti-patterns
• Indicators — full list with parameters
• Backtesting — data catalog, metrics, walkforward
• Exchange setup — Binance API key configuration
• Credits — adopted primitives and license attribution

Contributing

See CONTRIBUTING.md. Issues and pull requests are welcome. Please run pytest before submitting.

Naming

Kairós (καιρός) is the ancient Greek word for "the opportune moment". This package was originally published as autopilot-engine (v0.1.0, March 2026) and renamed at v0.2.0a0. The legacy PyPI name redirects here for deprecation notices.

License

MIT — see LICENSE. Derivatives and commercial use are both permitted; attribution is appreciated.

Citation

@software{kairos2026,
  title  = {Kairos: an event-driven crypto trading engine for Python},
  author = {Rovira, Oscar},
  year   = {2026},
  url    = {https://github.com/Vekkris76/kairos}
}