Detector Rule Engine: Define Strategies in TypeScript

What It Is

The Detector rule engine lets you define trading strategies as structured TypeScript objects instead of imperative code. Three sections: inputs, rules, thresholds.

Signals from the rule engine feed directly into Advisor's 8-stage AI pipeline. Better rule definitions = better AI decisions.

How It Works

Inputs — what data to use

inputs: {
  candles: { symbol: 'BTCUSDT', timeframe: '5m' },
  orderflow: { symbol: 'BTCUSDT', depth: 10 },
}

Data is fetched from Provider automatically.

Rules — what to look for

rules: [
  {
    id: 'ema_trend_aligned',
    description: 'Price above 20 EMA on 5m',
    condition: ({ candles }) => candles.close > candles.ema(20),
    weight: 2.0,
    direction: 'long',
  },
  {
    id: 'orderflow_imbalance',
    description: 'Buy flow exceeds sell by > 30%',
    condition: ({ orderflow }) => orderflow.buyRatio > 0.65,
    weight: 1.5,
    direction: 'long',
  },
  {
    id: 'volume_spike',
    description: 'Volume 2x above 20-period average',
    condition: ({ candles }) => candles.volume > candles.avgVolume(20) * 2,
    weight: 1.0,
    direction: 'long',
  },
]

Each fired rule adds its weight to the total score.

Thresholds — when to signal

thresholds: {
  long: { min: 3.5 },
  short: { min: 3.5 },
}

First two rules fire = score 3.5 (at threshold). All three = 4.5.

What You Get

• Full attribution — see exactly which rules fired for each candle and why
• Standardized scoring — all strategies use the same scale, making comparison easy
• Instance isolation — one broken rule doesn't affect other strategies
• Plain TypeScript — typed, autocompleted, versioned in git

Multiple Strategies

Run as many Detector instances as you want:

• Different strategies for different market regimes
• Different instances for different assets
• A/B test rule weights on the same symbol

How It Feeds AI

The signal's rule attribution passes to Advisor's pipeline. The LLM receives structured rule data instead of raw indicators — producing more grounded decisions.

Better rules → better signals → better AI decisions.

Get Started

• Detector API reference — full config schema
• Detector installation guide — set up your first instance
• See the advisor-llm-hybrid example in the repo for a production-ready config