DEV Community

Propfirmkey
Propfirmkey

Posted on

Automating Trade Analysis with Python: A Practical Guide

#python #automation #trading #datascience

Manual trade review is tedious. Let's automate the analysis part so you can focus on improving your strategy.

The Problem

After a week of trading, you have dozens of trades to review. Manually calculating metrics, identifying patterns, and generating reports takes hours.

Solution: A Python Analysis Pipeline

Step 1: Import Trade Data

Most platforms export to CSV:

import pandas as pd
import numpy as np

def load_trades(filepath):
    df = pd.read_csv(filepath)
    df['date'] = pd.to_datetime(df['date'])
    df['pnl'] = df['exit_price'] - df['entry_price']
    df['pnl'] *= np.where(df['direction'] == 'short', -1, 1)
    df['pnl'] *= df['size']
    return df
Enter fullscreen mode Exit fullscreen mode

Step 2: Core Metrics 

def calculate_metrics(df):
    metrics = {}
    metrics['total_trades'] = len(df)
    metrics['win_rate'] = (df['pnl'] > 0).mean() * 100
    metrics['avg_win'] = df[df['pnl'] > 0]['pnl'].mean()
    metrics['avg_loss'] = df[df['pnl'] < 0]['pnl'].mean()
    metrics['profit_factor'] = (
        df[df['pnl'] > 0]['pnl'].sum() /
        abs(df[df['pnl'] < 0]['pnl'].sum())
    )
    metrics['max_drawdown'] = calculate_max_drawdown(df)
    metrics['sharpe'] = df['pnl'].mean() / df['pnl'].std() * np.sqrt(252)
    return metrics

def calculate_max_drawdown(df):
    cumulative = df['pnl'].cumsum()
    peak = cumulative.cummax()
    drawdown = (cumulative - peak)
    return drawdown.min()
Enter fullscreen mode Exit fullscreen mode

Step 3: Pattern Detection 

def find_patterns(df):
    # Best performing hours
    df['hour'] = df['date'].dt.hour
    hourly = df.groupby('hour')['pnl'].agg(['mean', 'count'])
    best_hour = hourly['mean'].idxmax()

    # Consecutive losses
    df['is_loss'] = df['pnl'] < 0
    df['loss_streak'] = (df['is_loss'] != df['is_loss'].shift()).cumsum()
    max_streak = df[df['is_loss']].groupby('loss_streak').size().max()

    # Performance by instrument
    by_instrument = df.groupby('instrument')['pnl'].agg(['sum', 'mean', 'count'])

    return {
        'best_hour': best_hour,
        'max_loss_streak': max_streak,
        'by_instrument': by_instrument
    }
Enter fullscreen mode Exit fullscreen mode

Step 4: Weekly Report 

def generate_report(df, metrics, patterns):
    report = f"""
    WEEKLY TRADING REPORT
    {'='*40}
    Total Trades: {metrics['total_trades']}
    Win Rate: {metrics['win_rate']:.1f}%
    Profit Factor: {metrics['profit_factor']:.2f}
    Max Drawdown: ${metrics['max_drawdown']:.2f}
    Sharpe Ratio: {metrics['sharpe']:.2f}

    Best Trading Hour: {patterns['best_hour']}:00
    Max Loss Streak: {patterns['max_loss_streak']}

    BY INSTRUMENT:
    {patterns['by_instrument'].to_string()}
    """
    return report
Enter fullscreen mode Exit fullscreen mode

Going Further

This basic pipeline can be extended with:

  • Matplotlib/Plotly charts for equity curves
  • Monte Carlo simulations for risk assessment
  • Machine learning for setup classification
  • Automated email reports via cron

If you're trading with a prop firm, these metrics help you stay within their risk parameters. Understanding your own patterns is key to passing evaluations — I cover what each firm expects at propfirmkey.com.

What's your trade analysis workflow? Share in the comments.

Top comments (0)