BTC Martingale Backtest Tool

About This Tool

This is a backtesting engine for analyzing Martingale betting strategies on Bitcoin 5-minute OHLCV data. It simulates different entry strategies and measures their profitability, win rates, and risk metrics.

How the Strategy Works

1️⃣ Detect Streaks

Watch for N consecutive candles in the same direction (red or green). When a streak is detected, it signals an entry point.

2️⃣ Place Bet

After N red candles → Bet UP (expect reversal)
After N green candles → Bet DOWN (expect reversal)

3️⃣ Martingale Logic

Win? Pocket profit, reset bet.
Loss? Double bet next time.
Repeat until win or hit max consecutive loss limit.

Example Scenario

Setup: Streak=3, Start Bet=$2, Max Loss=11, Payout=2x

Signal: 3 red candles in a row detected ✓
Action: Bet UP for next candles

Candle 4: DOWN ❌ | Lose $2 | Next bet: $4

Candle 5: DOWN ❌ | Lose $4 | Next bet: $8

Candle 6: UP ✅ | Win $8 profit | Cycle done!

Result: Net profit: $8 - $2 - $4 = $2

How to Use

Prepare CSV Data
Get historical 5-min candle data from Binance in CSV format
Format: open_time, open, high, low, close, volume
Place CSV in backtesting/data/ folder
Default file: backtesting/data/file.csv
Run Backtest
python backtesting/tes.py --streak 3 --start_bet 2 --max_loss 11
Analyze Results
Check win rate, P&L, ruin probability, and loss distribution

Configuration Parameters

Parameter	Default	Description
`--file`	`data/file.csv`	Path to CSV file with OHLCV data
`--streak`	`3`	Number of consecutive candles to detect before entry
`--start_bet`	`2.0`	Initial bet size in USDC
`--max_loss`	`11`	Max consecutive losses before stopping cycle (risk control)
`--payout`	`2.0`	Multiplier for winning bet (2x = earn 1x bet as profit)
`--fee`	`0.0`	Trading fee as decimal (0.01 = 1% per trade)

Interpreting Results

Win Rate

Percentage of trades that are profitable

✅ > 50% = Good

❌ < 50% = Likely lose money

Net P&L

Total profit or loss in USDC

✅ Positive = Strategy works!

❌ Negative = Strategy loses money

Ruin Events

Times when hitting max consecutive loss limit

✅ Low (< 5%) = Safe

❌ High (> 20%) = Risky

Capital Needed

Minimum capital to survive max_loss streak

Example: $2,046 needed to survive 11x consecutive losses

📈 Live Backtest Results - Strategy A (3-Streak, $1 Start Bet)

Data: 846,445 BTC 5-min candles (2017-2025) | Updated: Feb 22, 2026

Fee 0% (Ideal)

Win Rate: 55.9%

Net P&L: $76,621.00

P&L per Cycle: $0.82

Fee 2% (Realistic Polymarket)

Win Rate: 55.9%

Net P&L: $67,074.26

P&L per Cycle: $0.72

Fee Impact

Profit Loss: -$9,546.74

Loss %: -12.5%

Status: ✅ Still Profitable

Win Rate by Fee

Net P&L Comparison

Loss Distribution (Fee 2%)

Ruin Probability vs Max Loss

Metric	Fee 0%	Fee 2%	Impact
Win Rate	55.9%	55.9%	No change
Net P&L	$76,621.00	$67,074.26	-12.5%
P&L per Cycle	$0.8193	$0.7172	-12.5%
Total Cycles	93,517	93,517	No change
Total Trades	167,270	167,270	No change
Max Consecutive Loss	8	8	No change
Ruin Events (8x)	66	66	No change
Capital Required	$255	$255	No change

💡 Key Insights

✅ Edge: 55.9% win rate beats 50% baseline by 11.8%
✅ Fees Matter: Strategy stays highly profitable even with 2% fees
✅ Safe Risk: Only 0.07% chance of hitting 8x consecutive losses
⚠️ Real World: Polymarket spreads may be 2-3%, add 0.5-1% for slippage
💰 Capital: Minimum $255-300 recommended for safe operation
📊 ROI: ~$467/day average on $1 starting bet (1,556% annual if consistent)

Analysis Types

Baseline

Simple "Bet UP every candle" with no streak filter. Used as comparison baseline.

Strategy A

Bet UP after N red candles. Tests if red streaks reverse to green.

Strategy B

Bet DOWN after N green candles. Tests if green streaks reverse to red.

Sweep Analysis

Test win rate for different streak lengths (1-7) to find optimal entry point.

Ruin Analysis

Analyze ruin probability for different max_loss values. Shows probability of hitting limits.

⚠️ Important Caveats

Backtesting ≠ Live Trading - Historical performance doesn't guarantee future results
Fees Matter - Real exchanges charge 0.1%-0.5% per trade. Update --fee accordingly
Slippage Not Included - Assumes instant execution at exact price
Market Assumptions - Assumes random walk, actual crypto has trends and correlations
Payout Assumption - Assumes 2x payout. Verify with your actual contract terms
Risk Management - Even profitable strategies can lose entire account during drawdowns

Usage Examples

Basic run with defaults:

python backtesting/tes.py

Custom parameters:

python backtesting/tes.py \
  --file backtesting/data/file.csv \
  --streak 5 \
  --start_bet 5 \
  --max_loss 8 \
  --payout 2.0 \
  --fee 0.01

Using as Python Module

The backtest engine can be imported and used in other Python projects:

                from backtesting.utils import BacktestEngine, BacktestConfig, load_data, add_direction

# Load and prepare data
df = load_data('data/file.csv')
df = add_direction(df)

# Create engine
engine = BacktestEngine(df)

# Configure
config = BacktestConfig(
    streak_len=3,
    start_bet=2.0,
    max_loss=11,
)

# Run
result = engine.run(config)
print(f"Win rate: {result.win_rate:.1f}%")
print(f"P&L: ${result.balance_change:.2f}")
            

Project Structure

                backtesting/
├── README.md                    # Full documentation
├── tes.py                       # CLI tool (main entrypoint)
├── utils/
│   ├── __init__.py
│   └── backtest_engine.py       # Core reusable module
├── data/
│   └── file.csv                 # Historical OHLCV data
└── output/                      # Output results (if needed)
            

Tech Stack

Python 3.9+ - Core language
Pandas - Data manipulation and analysis
NumPy - Numerical computations

Next Steps

Read the full 📖 README.md for detailed documentation
Prepare your CSV data (Binance 5-min OHLCV format)
Run the backtest with different parameters
Analyze results and compare strategies
Use insights to improve your trading approach