Crypto Backtest Bot

OI + CVD + Long/Short + Funding + Liquidations — Aggregated Multi-Exchange

A Python backtesting engine that generates trading signals from aggregated derivatives data across 10+ exchanges (Binance, Bybit, OKX, Deribit, BitMEX, etc.) via the Coinalyze API.

---

Quick Start (< 2 minutes)

# 1. Clone / copy the project
cd crypto_backtest

# 2. Install dependencies
pip install -r requirements.txt

# 3. Test immediately with sample data (no API key needed)
python run.py --sample

# 4. For real data: get free API key → https://coinalyze.net/account/api-key/
export COINALYZE_API_KEY="your_key_here"
python run.py --symbol BTC --interval 4hour --days 90

---

Architecture

┌──────────────────────────────────────────────────────────────┐
│                     DATA LAYER (Coinalyze API)               │
│  Aggregated across: Binance, Bybit, OKX, Deribit, BitMEX,   │
│  Bitfinex, Huobi, Kraken, Hyperliquid, Gate.io, WOO X       │
├──────────────────────────────────────────────────────────────┤
│  OHLCV + Buy/Sell Vol  │  Open Interest  │  Long/Short Ratio │
│  Funding Rate          │  Liquidations   │  (all aggregated) │
└────────────┬─────────────────────┬───────────────────────────┘
             │                     │
             ▼                     ▼
┌──────────────────────┐  ┌──────────────────────┐
│   SIGNAL ENGINE      │  │   CVD CALCULATION    │
│                      │  │                      │
│  1. OI Momentum      │  │  buy_vol - sell_vol  │
│  2. CVD Divergence   │  │  cumulative sum      │
│  3. L/S Contrarian   │  │  price divergence    │
│  4. Funding Contrarian│  │  detection           │
│  5. Liq Cascade      │  │                      │
└────────────┬─────────┘  └──────────┬───────────┘
             │                        │
             ▼                        ▼
┌──────────────────────────────────────────────────────────────┐
│              COMPOSITE SCORE (-1 to +1)                      │
│  Weighted sum → EMA smoothed → threshold crossing → SIGNAL   │
└────────────┬─────────────────────────────────────────────────┘
             │
             ▼
┌──────────────────────────────────────────────────────────────┐
│                    BACKTESTING ENGINE                         │
│  • Bar-by-bar simulation with SL/TP                          │
│  • Commission (0.06%) + slippage (0.02%)                     │
│  • Full metrics: Sharpe, Sortino, Max DD, PF, Win Rate       │
│  • HTML report + trade log CSV + metrics JSON                │
└──────────────────────────────────────────────────────────────┘

---

Data Sources — Why Coinalyze

Feature	Coinalyze	Bybit Only	CoinGlass
Price	Free	API cost	Free
Aggregated	✅ 10+ exchanges	❌ Single	✅
OI History	✅ Free	✅ Limited	💰 Paid API
CVD (Buy/Sell Vol)	✅ Free	❌ Manual calc	💰 Paid
Long/Short Ratio	✅ Free	✅	💰 Paid
Funding Rate	✅ Free	✅	💰 Paid
Liquidations	✅ Free	❌ Streams only	💰 Paid
API Rate Limit	40/min	120/min	Varies
Setup Time	1 min	5 min	10+ min

Coinalyze wins for backtesting: single API gives you everything aggregated, free, with historical data.

Symbol Format

BTCUSDT_PERP.A    ← Aggregated across ALL exchanges (recommended)
BTCUSDT_PERP.6    ← Bybit only
BTCUSDT_PERP.1    ← Binance only

Exchange codes: 0=BitMEX, 1=Binance, 2=Bitfinex, 4=Deribit, 6=Bybit, 7=Huobi, B=OKX

Data Retention

• Intraday (1min–12h): ~1500-2000 bars (~250 days for 4h)
• Daily: Unlimited history

---

Signal Logic

1. Open Interest Momentum (weight: 25%)

OI Change	Price Change	Signal	Interpretation
↑ Spike (>5%)	↑ Up	+0.7 (Bullish)	New money entering trend
↑ Spike (>5%)	↓ Down	-0.3 (Bearish)	New shorts opening
↓ Drop (<-3%)	↑ Up	-0.2 (Weak)	Short squeeze, no conviction
↓ Drop (<-3%)	↓ Down	+0.3 (Bullish)	Capitulation → reversal

2. CVD Divergence (weight: 25%)

• Bullish divergence: Price down + CVD up → accumulation
• Bearish divergence: Price up + CVD down → distribution
• Uses normalized rate-of-change for scale-independent comparison

3. Long/Short Ratio — Contrarian (weight: 20%)

• Crowded longs (>65%) → bearish signal
• Crowded shorts (<35%) → bullish signal
• Smoothed over 6 bars to avoid noise

4. Funding Rate — Contrarian (weight: 15%)

• Extreme positive (>0.05%) → overheated longs → bearish
• Extreme negative (<-0.03%) → overheated shorts → bullish

5. Liquidation Cascade (weight: 15%)

• Short liquidation spike → forced buying → bullish
• Long liquidation spike → forced selling → bearish
• Triggered when liquidation volume > 2x rolling average

Composite Score

Score = Σ(signal_i × weight_i)    range: [-1, +1]
Smoothed = EMA(Score, span=3)
Entry: |Smoothed| > threshold → Long if positive, Short if negative
Exit:  Signal reverses, flattens, or hits SL/TP

---

Usage

Basic Backtest

# BTC, 4-hour candles, 90 days
python run.py

# ETH, 1-hour candles, 60 days
python run.py --symbol ETH --interval 1hour --days 60

# SOL, daily candles, 180 days
python run.py --symbol SOL --interval daily --days 180

Parameter Optimization

python run.py --optimize
# Tests 100 combinations of entry threshold, stop loss, take profit
# Outputs ranked results by Sharpe ratio

Sample Data (No API Key)

python run.py --sample --bars 1000

Outputs

results/
├── backtest_BTC_4hour.html      ← Interactive HTML report with charts
├── trades_BTC_4hour.csv         ← Full trade log
├── metrics_BTC_4hour.json       ← Performance metrics
└── optimization_BTC.csv         ← Parameter sweep results

---

Configuration

Edit config.py to tune:

# Symbols
SYMBOLS = {
    "BTC": {"perp": "BTCUSDT_PERP.A", "spot": "BTCUSDT.C"},
    "ETH": {"perp": "ETHUSDT_PERP.A", "spot": "ETHUSDT.C"},
    "SOL": {"perp": "SOLUSDT_PERP.A", "spot": "SOLUSDT.C"},
}

# Strategy Parameters
STRATEGY_PARAMS = {
    "oi_spike_threshold": 0.05,     # 5% OI increase
    "cvd_lookback": 12,             # CVD trend window
    "ls_extreme_long": 0.65,        # Crowded long threshold
    "funding_extreme_pos": 0.0005,  # Overheated funding
    "entry_threshold": 0.15,        # Composite score entry
    "stop_loss_pct": 0.03,          # 3% stop loss
    "take_profit_pct": 0.06,        # 6% take profit
}

---

Adding More Coins

Add any Coinalyze-supported symbol to config.py:

SYMBOLS["DOGE"] = {
    "perp": "DOGEUSDT_PERP.A",
    "spot": "DOGEUSDT.C",
}

Then run: python run.py --symbol DOGE

To find available symbols, run:

from data_fetcher import CoinalyzeClient
client = CoinalyzeClient()
markets = client.get_future_markets()
# Filter for aggregated perpetuals
perps = [m for m in markets if m['is_perpetual'] and m['symbol'].endswith('.A')]

---

Project Structure

crypto_backtest/
├── config.py           # All settings, API keys, strategy params
├── data_fetcher.py     # Coinalyze API client + caching + sample data
├── strategy.py         # 5 sub-signals + composite scoring
├── backtester.py       # Position management, SL/TP, metrics
├── run.py              # CLI entry point, optimizer, HTML report
├── requirements.txt    # pandas, numpy, requests
└── README.md           # This file

---

Next Steps / Extending

1. Live trading: Connect signals to Bybit API (pybit library) for order execution
2. Multi-timeframe: Combine 1h signals with 4h confirmation
3. Walk-forward optimization: Rolling window parameter tuning
4. Additional signals: Whale wallet tracking, exchange inflows/outflows, options gamma
5. Alert system: Telegram/Discord webhook when composite score crosses threshold