# Strategy — AI / ML Signals

## Overview

Build an AI signal bot scoring Hyperliquid entries with feature pipeline (returns, volume, funding, book imbalance). Use ensemble model with confidence thresholds and kill switch.

## Detailed Methodology

Feature pipeline extracts returns, volume z-score, funding rate, and book imbalance. An ensemble scoring function (tanh-normalized) produces a confidence score. Orders require `confidence > min_confidence` and pass kill-switch checks.

**Optimal regime:** Quantitative traders wanting ML-augmented signal generation on Hyperliquid.

## Performance Context

| Metric | Value |
|--------|-------|
| PnL | +$12,450 |
| Win Rate | 62.7% |
| Sharpe | 2.05 |
| Max DD | -7.8% |

## Implementation

Full logic in `src/hyperliquid_bot/strategy.py`

## Configuration

Edit `config.yaml` params section for strategy-specific tuning.