📊 EMA Crossover Strategy Backtest System

Complete, production-ready backtesting system for EMA crossover strategy with StochRSI confirmation, €500 max loss/€2000 max gain risk management, and comprehensive trade analysis.

---

🎯 Strategy Overview

Entry Rules

1. 50EMA crosses above 200EMA (uptrend confirmation)
2. Wait 3 candles (filter false breakouts)
3. StochRSI crosses above 0.20 (momentum confirmation)
   • Execute LONG position when all conditions align

Exit Rules (Any of the following)

1. Max Loss: -€500 per trade (hard stop loss)
2. Max Gain: +€2000 per trade (profit target)
3. StochRSI falls below 0.20 (momentum reversal)
4. 50EMA crosses below 200EMA (trend reversal)
5. Daily ATR exhausted (volatility spike >20% above 20-bar MA)

Account & Risk

• Initial Capital: €1,000,000
• Position Size: 2% risk per trade
• Commission: 0.1%
• Margin: 10% (90:1 leverage)
• Stocks: 30 major stocks from NYSE, NASDAQ, Euronext

---

📁 Files & Structure

backtest_system/
├── ema_strategy_complete.py      # Main backtest engine
├── backtest_analyzer.py           # Results analysis & visualization
├── run_backtest.py                # Automated pipeline (use this!)
├── QUICKSTART.py                  # Setup guide & examples
├── requirements.txt               # Python dependencies
├── README.md                      # This file
└── results/                       # Output directory (created after run)
    ├── all_trades_summary.csv
    ├── trades_SYMBOL_*.csv
    ├── backtest_detailed_report.xlsx
    ├── backtest_charts/
    │   ├── 01_pnl_distribution.png
    │   ├── 02_cumulative_pnl.png
    │   ├── 03_win_loss_distribution.png
    │   ├── 04_exit_reason_pnl.png
    │   ├── 05_symbol_performance.png
    │   ├── 06_duration_vs_pnl.png
    │   └── 07_monthly_pnl.png
    └── backtest_report.html

---

🚀 Quick Start

1. Installation

# Clone or download this directory
cd backtest_system

# Create virtual environment (recommended)
python -m venv backtest_env
source backtest_env/bin/activate  # On Windows: backtest_env\Scripts\activate

# Install dependencies
pip install -r requirements.txt

2. Run Backtest (Simplest Method)

# Run complete pipeline: backtest → analysis → report
python run_backtest.py

This single command:

1. ✅ Installs dependencies
2. ✅ Runs backtest on 15 stocks
3. ✅ Analyzes all trades
4. ✅ Creates visualizations
5. ✅ Generates HTML report

3. View Results

results/
├── all_trades_summary.csv        ← Open in Excel/Sheets
├── backtest_detailed_report.xlsx  ← Multi-sheet analysis
└── backtest_charts/              ← 7 visualization charts

---

🎨 Customization

Change Parameters

Edit the beginning of ema_strategy_complete.py:

# Line ~330 - STRATEGY PARAMETERS
stocks = [
    'JPM', 'AAPL', 'MSFT',  # Your stocks here
    # Add or remove as needed
]

# Change date range
results_df = run_backtest(stocks, 
                         start_date='2023-01-01',
                         end_date='2025-12-20')

Adjust Strategy Settings

In ema_strategy_complete.py, modify the EMACrossoverStrategy.params:

class EMACrossoverStrategy(bt.Strategy):
    params = (
        ('short_ema', 50),            # Change 50EMA period
        ('long_ema', 200),            # Change 200EMA period
        ('max_loss_eur', 500),        # Change max loss
        ('max_gain_eur', 2000),       # Change max gain
        ('account_size', 1000000.0),  # Change starting capital
    )

Test Different Timeframes

Change yfinance interval:

# Line ~163 - Change from:
data = yf.download(symbol, start=start_date, end=end_date, 
                  auto_adjust=False, progress=False)

# To:
data = yf.download(symbol, start=start_date, end=end_date,
                  interval='1h',  # Hourly (limited to 730 bars)
                  auto_adjust=False, progress=False)

# Or for 15-min:
data = yf.download(symbol, start=start_date, end=end_date,
                  interval='15m',  # 15-minute (limited to 60 days)
                  auto_adjust=False, progress=False)

---

📊 Understanding the Output

all_trades_summary.csv

Column	Meaning
Symbol	Stock ticker
Entry_Time	When position was entered
Exit_Time	When position was exited
Entry_Price	Price at entry
Exit_Price	Price at exit
Shares	Number of shares traded
Position_Value_EUR	Market value (shares × entry_price)
PnL_EUR	Profit/loss in EUR
Max_Win	Maximum profit if positive
Max_Loss	Maximum loss if negative
Margin_Used_EUR	Capital tied up (10% of position)
At_Risk_Capital_EUR	€500 per trade
Candles_In_Trade	How many bars held
Exit_Reason	Why trade exited
Return_Percent	Return on position

backtest_detailed_report.xlsx

Sheet 1: All Trades

• Complete trade log with all columns from CSV

Sheet 2: Summary

• Win rate, average PnL, best/worst trades
• Total stats across all symbols

Sheet 3: By Symbol

• Performance breakdown for each stock
• Win rates and total P&L per symbol

Sheet 4: By Exit Reason

• Analysis of each exit condition
• Which exits are most profitable

Charts Generated

Chart	Purpose
01_pnl_distribution.png	Histogram showing profit distribution
02_cumulative_pnl.png	Running total profit over time
03_win_loss_distribution.png	Wins vs losses comparison
04_exit_reason_pnl.png	Profit by exit reason
05_symbol_performance.png	Performance per stock
06_duration_vs_pnl.png	Trade length impact on profit
07_monthly_pnl.png	Monthly aggregated results

---

🔧 Troubleshooting

No trades found

Problem: Backtest runs but generates 0 trades

Solutions:

1. Lower STOCHRSI_THRESHOLD from 0.20 to 0.15
2. Reduce BARS_AFTER_CROSSOVER from 3 to 1
3. Use older date range (earlier start date)
4. Check stock symbols are correct

Memory issues

Problem: Python runs out of memory on all 30 stocks

Solution:

# Test on smaller list first
results_df = run_backtest(stocks[:10])  # Just first 10
# Or:
results_df = run_backtest(['AAPL', 'MSFT', 'GOOGL'])  # Specific stocks

yfinance download fails

Problem: "Unable to download data" error

Solutions:

1. Check internet connection
2. Verify ticker symbols (e.g., 'ASML.AS' not just 'ASML')
3. Use simpler symbols first: AAPL, MSFT, GOOGL
4. Try again later (rate limits)

Slow performance

Problem: Backtest takes hours

Solutions:

1. Reduce stock list (start with 5-10 symbols)
2. Use shorter date range
3. Run on hourly/daily instead of 15-min (less data)
4. Close other applications

---

📈 Performance Metrics Explained

Win Rate

• Definition: % of trades that were profitable
• Good: >50%
• Excellent: >60%
• Formula: (Winning Trades / Total Trades) × 100

Profit Factor

• Definition: Total wins divided by total losses
• Good: >1.0 (profitable)
• Excellent: >2.0 (very profitable)
• Formula: (Sum of Wins) / Abs(Sum of Losses)

Sharpe Ratio

• Definition: Risk-adjusted return
• Good: >1.0
• Excellent: >2.0
• Higher is better

Drawdown

• Definition: Largest peak-to-trough decline
• Meaning: Worst case scenario
• Lower is better

Average Trade

• Definition: Mean profit per trade
• Formula: Total PnL / Number of Trades

---

💡 Strategy Improvements

To Increase Win Rate:

1. Raise max loss threshold (€500 → €300)
2. Lower entry threshold (StochRSI 0.20 → 0.30)
3. Add more entry filters

To Increase Average Win:

1. Raise max gain target (€2000 → €3000)
2. Trail stops instead of hard exits
3. Add profit-taking at partial targets

To Reduce Losses:

1. Lower position size (2% → 1%)
2. Add volatility filter (ATR-based)
3. Skip trades in low-liquidity stocks

To Improve Consistency:

1. Add market regime filter (VIX, trend)
2. Optimize EMA periods (test 40/180, 45/200, etc.)
3. Add position sizing by volatility

---

🔗 Integration with Live Trading

This backtest system can be extended to live trading:

# Switch from yfinance (historical) to broker API:
import ccxt  # For crypto
from alpaca_trade_api import REST  # For stocks

# Modify data feed in ema_strategy_complete.py
# Add broker integration
# Use IB (Interactive Brokers) or other broker APIs

See Backtrader documentation for live trading integration.

---

📚 Resources

• Backtrader Docs: https://www.backtrader.com/
• yfinance: https://github.com/ranaroussi/yfinance
• Pandas Docs: https://pandas.pydata.org/docs/
• Matplotlib: https://matplotlib.org/

---

⚖️ Disclaimer

This is for educational and backtesting purposes only.

• Past performance does not guarantee future results
• Backtesting results may not reflect real trading conditions
• Commission, slippage, and gaps are simplified
• Always test thoroughly before live trading
• Risk management is YOUR responsibility

---

📝 License

Free to use and modify for personal use.

---

🎯 Next Steps

1. Run backtest: python run_backtest.py
2. Review results: Open results/all_trades_summary.csv
3. Analyze charts: View results/backtest_charts/
4. Optimize parameters: Edit ema_strategy_complete.py
5. Test different stocks: Modify stock list
6. Try new timeframes: Change interval parameter

---

Questions? Refer to QUICKSTART.py for more examples and configuration details.