Henry Lin

Posted on Oct 4

Lesson 8: Strategy Batch Comparison

⏱ Duration: 1.5 hours
🎯 Learning Objectives: Master multi-strategy comparison methods
📚 Difficulty: ⭐⭐ Backtesting practical

📖 Course Overview

When you have multiple strategies, how do you quickly find the optimal one? This lesson will teach you to use Freqtrade's batch backtesting feature to efficiently compare the performance of multiple strategies and establish a scientific strategy selection system.

8.1 Batch Backtesting Commands

Basic Batch Backtesting

Method 1: Using `--strategy-list` Parameter

# Backtest multiple strategies at once
freqtrade backtesting \
  -c config.json \
  --strategy-list Strategy001 Strategy002 Strategy003 \
  --timerange 20250701-20250930 \
  --timeframe 15m

Output Features:

Each strategy executes sequentially
Shows summary comparison table at the end
Automatically sorts by key metrics

Method 2: Using Shell Loop

Create batch_backtest.sh:

#!/bin/bash

# Configuration
CONFIG="config.json"
TIMERANGE="20250701-20250930"
TIMEFRAME="15m"
STRATEGIES=(
  "Strategy001"
  "Strategy002"
  "Strategy003"
  "MeanReversionStrategy"
  "MomentumTrendStrategy"
)

echo "Starting batch backtesting of ${#STRATEGIES[@]} strategies..."
echo "Time range: $TIMERANGE"
echo "Timeframe: $TIMEFRAME"
echo "========================================"

for STRATEGY in "${STRATEGIES[@]}"
do
  echo ""
  echo "[$STRATEGY] Backtesting..."
  freqtrade backtesting \
    -c $CONFIG \
    --strategy $STRATEGY \
    --timerange $TIMERANGE \
    --timeframe $TIMEFRAME

  echo "[$STRATEGY] Completed!"
  echo "========================================"
done

echo ""
echo "All strategy backtests completed!"

Run:

chmod +x batch_backtest.sh
./batch_backtest.sh

Batch Backtesting Summary Report

Freqtrade will display a summary comparison table after all strategy backtests are completed:

STRATEGY SUMMARY
┏━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━┓
┃ Strategy                ┃ Trades ┃ Avg Profit ┃ Tot Profit % ┃ Win Rate % ┃ Max Drawdown ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━┩
│ MomentumTrendStrategy   │     68 │       1.52 │        23.45 │       89.7 │         3.21 │
│ Strategy003             │     42 │       1.31 │        18.72 │       85.7 │         4.15 │
│ Strategy001             │    127 │       0.85 │        15.33 │       72.4 │         6.82 │
│ MeanReversionStrategy   │     89 │       0.62 │        11.28 │       78.6 │         5.43 │
│ Strategy002             │     35 │       0.43 │         8.95 │       68.6 │         7.21 │
└─────────────────────────┴────────┴────────────┴──────────────┴────────────┴──────────────┘

Key Information:

Default sorting: By total profit from high to low
Quick identification: See the best strategy at a glance
Multi-dimensional comparison: Display multiple key metrics simultaneously

8.2 Strategy Comparison Table Interpretation

How to Quickly Identify the Best Strategy

1. Profit-First Type

Focus Metrics:

Total Profit % > 15%
Avg Profit % > 0.8%

Case:

MomentumTrendStrategy: Total profit 23.45%, average 1.52% ✅

Suitable For:

Aggressive investors
Pursuing high returns
Can tolerate larger drawdowns

2. Risk-First Type

Focus Metrics:

Max Drawdown < 5%
Sharpe Ratio > 2.5
Win Rate > 80%

Case:

Strategy003: Drawdown 4.15%, win rate 85.7% ✅

Suitable For:

Conservative investors
Focus on capital safety
Cannot tolerate large drawdowns

3. Balanced Type

Focus Metrics:

Total Profit > 12%
Max Drawdown < 7%
Win Rate > 75%
Sharpe Ratio > 2.0

Case:

Strategy003: Profit 18.72%, drawdown 4.15%, win rate 85.7% ✅✅✅

Suitable For:

Most quantitative traders
Pursuing stable returns
Long-term sustainable profits

Multi-dimensional Scoring System

Scoring Formula (100-point scale)

# Weight distribution
profit_weight = 0.30      # 30%
risk_weight = 0.25        # 25%
winrate_weight = 0.20     # 20%
sharpe_weight = 0.15      # 15%
frequency_weight = 0.10   # 10%

total_score = (
    (total_profit% / 30 × 100) × 0.30 +
    ((10 - max_drawdown%) / 10 × 100) × 0.25 +
    (win_rate%) × 0.20 +
    (min(sharpe, 5) / 5 × 100) × 0.15 +
    (trade_frequency_score) × 0.10
)

Trade Frequency Score Standards

Trade Count (30 days)	Score	Description
< 10	60	Too few, insufficient sample
10-30	85	Ideal, low frequency high quality
30-80	100	Perfect, moderate frequency
80-150	80	Acceptable, slightly high frequency
> 150	50	Over-trading

Actual Scoring Case

MomentumTrendStrategy:

Profit score: (23.45 / 30 × 100) × 0.30 = 23.45
Risk score: ((10 - 3.21) / 10 × 100) × 0.25 = 16.98
Win rate score: 89.7 × 0.20 = 17.94
Sharpe score: (3.5 / 5 × 100) × 0.15 = 10.50
Frequency score: (85 / 100) × 0.10 = 8.50

Total score = 77.37 ⭐⭐⭐⭐

Strategy001:

Profit score: (15.33 / 30 × 100) × 0.30 = 15.33
Risk score: ((10 - 6.82) / 10 × 100) × 0.25 = 7.95
Win rate score: 72.4 × 0.20 = 14.48
Sharpe score: (2.1 / 5 × 100) × 0.15 = 6.30
Frequency score: (80 / 100) × 0.10 = 8.00

Total score = 52.06 ⭐⭐⭐

8.3 Strategy Selection Decision Tree

Decision Flow Chart

Start Strategy Selection
    │
    ├─ Step 1: Eliminate Unqualified Strategies
    │   ├─ Total Profit < 5% → ❌ Eliminate
    │   ├─ Max Drawdown > 15% → ❌ Eliminate
    │   ├─ Win Rate < 50% → ❌ Eliminate
    │   └─ Sharpe < 1.0 → ❌ Eliminate
    │
    ├─ Step 2: Classify by Goal
    │   ├─ Pursue High Returns → Sort by total profit
    │   ├─ Pursue Low Risk → Sort by max drawdown
    │   └─ Pursue Balance → Sort by comprehensive score
    │
    ├─ Step 3: Validate Sample Size
    │   ├─ Trade Count < 10 → ⚠️ Insufficient sample, need longer test
    │   ├─ Trade Count 10-50 → ✅ Qualified
    │   └─ Trade Count > 150 → ⚠️ Over-trading, check fees
    │
    └─ Step 4: Final Confirmation
        ├─ Check exit reason distribution
        ├─ Check holding time distribution
        ├─ Conduct out-of-sample testing
        └─ Select Strategy ✅

Python Automated Selection Script

Create strategy_selector.py:

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

"""
Strategy Automatic Selection Tool
Automatically score and rank strategies based on backtest results
"""

strategies = [
    {
        "name": "MomentumTrendStrategy",
        "trades": 68,
        "avg_profit": 1.52,
        "total_profit": 23.45,
        "win_rate": 89.7,
        "max_drawdown": 3.21,
        "sharpe": 3.5
    },
    {
        "name": "Strategy003",
        "trades": 42,
        "avg_profit": 1.31,
        "total_profit": 18.72,
        "win_rate": 85.7,
        "max_drawdown": 4.15,
        "sharpe": 3.2
    },
    {
        "name": "Strategy001",
        "trades": 127,
        "avg_profit": 0.85,
        "total_profit": 15.33,
        "win_rate": 72.4,
        "max_drawdown": 6.82,
        "sharpe": 2.1
    }
]

def calculate_score(strategy):
    """Calculate comprehensive score"""
    # Profit score (30%)
    profit_score = min(strategy["total_profit"] / 30 * 100, 100) * 0.30

    # Risk score (25%)
    risk_score = (10 - min(strategy["max_drawdown"], 10)) / 10 * 100 * 0.25

    # Win rate score (20%)
    winrate_score = strategy["win_rate"] * 0.20

    # Sharpe score (15%)
    sharpe_score = min(strategy["sharpe"] / 5 * 100, 100) * 0.15

    # Trade frequency score (10%)
    trades = strategy["trades"]
    if trades < 10:
        freq_score = 60
    elif 10 <= trades < 30:
        freq_score = 85
    elif 30 <= trades <= 80:
        freq_score = 100
    elif 80 < trades <= 150:
        freq_score = 80
    else:
        freq_score = 50
    freq_score *= 0.10

    total_score = profit_score + risk_score + winrate_score + sharpe_score + freq_score
    return round(total_score, 2)

def filter_strategies(strategies):
    """Filter unqualified strategies"""
    qualified = []
    for s in strategies:
        if (s["total_profit"] >= 5 and
            s["max_drawdown"] <= 15 and
            s["win_rate"] >= 50 and
            s["sharpe"] >= 1.0):
            qualified.append(s)
    return qualified

def rank_strategies(strategies):
    """Strategy ranking"""
    for s in strategies:
        s["score"] = calculate_score(s)

    return sorted(strategies, key=lambda x: x["score"], reverse=True)

# Main program
print("=" * 60)
print("Strategy Automatic Selection Tool")
print("=" * 60)

# Filter
qualified = filter_strategies(strategies)
print(f"\nQualified strategies: {len(qualified)} / {len(strategies)}")

# Rank
ranked = rank_strategies(qualified)

# Output results
print("\nStrategy Ranking (by comprehensive score):")
print("-" * 60)
for i, s in enumerate(ranked, 1):
    print(f"{i}. {s['name']}")
    print(f"   Total Score: {s['score']} | Profit: {s['total_profit']}% | "
          f"Drawdown: {s['max_drawdown']}% | Win Rate: {s['win_rate']}%")
    print()

# Recommendation
print("=" * 60)
print("🏆 Recommended Strategy:", ranked[0]["name"])
print(f"   Comprehensive Score: {ranked[0]['score']}")
print("=" * 60)

Run:

python3 strategy_selector.py

8.4 Strategy Selection in Different Market Conditions

Market Type Identification

1. Bull Market

Features:

Continuous upward trend
Small pullback magnitude
Increased trading volume

Recommended Strategy Types:

✅ Trend following strategies
✅ Momentum strategies
✅ Breakout strategies
❌ Avoid mean reversion strategies

Case Strategies:

MomentumTrendStrategy
BreakoutTrendStrategy
ADXTrendStrategy

2. Bear Market

Features:

Continuous downward trend
Weak rebounds
Decreased trading volume

Recommended Strategy Types:

✅ Conservative strategies
✅ Defensive strategies
✅ Short selling strategies (if supported)
❌ Avoid aggressive momentum-chasing strategies

Countermeasures:

Reduce position size
Raise stop loss standards
Decrease trading frequency
Consider suspending trading

3. Range-Bound Market

Features:

Sideways consolidation
Range-bound fluctuations
Many false breakouts

Recommended Strategy Types:

✅ Mean reversion strategies
✅ Grid trading strategies
✅ RSI overbought/oversold strategies
❌ Avoid trend following strategies

Case Strategies:

MeanReversionStrategy
GridTradingStrategy
Bollinger Bands reversal strategies

Market Condition Comparison Testing

Practical Suggestions:
Divide historical data into different market conditions for testing

# Bull market phase (assume 2025-01-01 to 2025-03-31 is bull market)
freqtrade backtesting \
  -c config.json \
  --strategy MomentumTrendStrategy \
  --timerange 20250101-20250331

# Range-bound market phase (assume 2025-04-01 to 2025-06-30 is range-bound)
freqtrade backtesting \
  -c config.json \
  --strategy MeanReversionStrategy \
  --timerange 20250401-20250630

# Bear market phase (assume 2025-07-01 to 2025-09-30 is bear market)
freqtrade backtesting \
  -c config.json \
  --strategy-list Strategy001 Strategy002 \
  --timerange 20250701-20250930

Strategy Adaptability Matrix

Strategy Type	Bull Market	Bear Market	Range-Bound	Overall Adaptability
Trend Following	⭐⭐⭐⭐⭐	⭐	⭐⭐	⭐⭐⭐
Mean Reversion	⭐⭐	⭐⭐	⭐⭐⭐⭐⭐	⭐⭐⭐
Momentum Strategy	⭐⭐⭐⭐⭐	⭐	⭐	⭐⭐
Breakout Strategy	⭐⭐⭐⭐	⭐⭐	⭐⭐⭐	⭐⭐⭐
Grid Strategy	⭐	⭐	⭐⭐⭐⭐⭐	⭐⭐

💡 Practical Tasks

Task 1: Batch Backtest 5 Strategies

# Ensure you have these strategies
ls user_data/strategies/

# Batch backtesting
freqtrade backtesting \
  -c config.json \
  --strategy-list Strategy001 Strategy002 Strategy003 \
                   MeanReversionStrategy MomentumTrendStrategy \
  --timerange 20250701-20250930 \
  --timeframe 15m

Task 2: Create Strategy Comparison Table

Create Excel or Google Sheets, record the following information:

Strategy Name	Trade Count	Win Rate%	Total Profit%	Avg Profit%	Max Drawdown%	Sharpe	Total Score	Rank
Strategy001	?	?	?	?	?	?	?	?
Strategy002	?	?	?	?	?	?	?	?
Strategy003	?	?	?	?	?	?	?	?
MeanReversion	?	?	?	?	?	?	?	?
MomentumTrend	?	?	?	?	?	?	?	?

Task 3: Select the Best Strategy

Based on the following three goals, select the most suitable strategies respectively:

Pursuing High Returns (Profit-First):
- Selection: ___________
- Reason: ___________
Pursuing Low Risk (Risk-First):
- Selection: ___________
- Reason: ___________
Pursuing Balance (Comprehensive Best):
- Selection: ___________
- Reason: ___________

Task 4: Create Strategy Recommendation List

Recommendation List Template:

🥇 Gold Strategy (Comprehensive Best)

Strategy Name: ___________
Total Profit: ___________%
Max Drawdown: ___________%
Suitable For: ___________
Recommendation Index: ⭐⭐⭐⭐⭐

🥈 Silver Strategy (Alternative)

Strategy Name: ___________
Total Profit: ___________%
Max Drawdown: ___________%
Suitable For: ___________
Recommendation Index: ⭐⭐⭐⭐

🥉 Bronze Strategy (Specific Scenarios)

Strategy Name: ___________
Applicable Market: ___________
Recommendation Index: ⭐⭐⭐

📚 Knowledge Check

Basic Questions

How to backtest multiple strategies at once?
What is the default sorting for strategy comparison tables?
Under what conditions should a strategy be eliminated?

Answers

Use --strategy-list parameter: freqtrade backtesting --strategy-list Strategy001 Strategy002
Sorted by total profit from high to low
Should be eliminated when total profit < 5%, drawdown > 15%, win rate < 50%, Sharpe < 1.0

Advanced Questions

Will a strategy that performs well in bull markets also perform well in bear markets?
How to determine if a strategy is overfitted?
Why need multiple strategies instead of just using the best one?

Thought Questions

If all strategies perform poorly in a certain period, what does it indicate?
Is the top-ranked strategy necessarily the best?
How to build a strategy portfolio to reduce risk?

🔗 Reference Materials

Supporting Documentation

📄 STRATEGY_SELECTION_GUIDE.md - Complete strategy selection guide
📄 TESTING_GUIDE.md - Batch backtesting methods

📌 Key Points Summary

Batch backtesting saves time and effort: Compare multiple strategies at once
Don't just look at total profit: Consider risk, win rate, Sharpe comprehensively
Establish scoring system: Scientifically quantify strategy quality
Vary by market conditions: Select different strategies for different markets
Sample size is important: Strategies with < 10 trades need caution
Filter unqualified strategies: Set minimum standards, strict screening

➡️ Next Lesson Preview

Lesson 9: Time Range Testing

In the next lesson, we will:

Validate strategy stability across different time periods
Identify bull markets, bear markets, range-bound markets
Conduct out-of-sample testing
Avoid overfitting traps

Preparation:

✅ Select 1-2 well-performing strategies
✅ Download at least 6 months of historical data
✅ Understand recent market trend types

🎯 Learning Check Standards:

✅ Can independently batch backtest multiple strategies
✅ Can create strategy comparison tables and scoring
✅ Can select suitable strategies based on goals
✅ Understand the impact of different market conditions on strategies

After completing these tasks, you have mastered the core methods of strategy selection! Ready to move on to time range testing! 🎯

Lesson 8: Strategy Batch Comparison

📖 Course Overview

8.1 Batch Backtesting Commands

Basic Batch Backtesting

Method 1: Using --strategy-list Parameter

Method 2: Using Shell Loop

Batch Backtesting Summary Report

8.2 Strategy Comparison Table Interpretation

How to Quickly Identify the Best Strategy

1. Profit-First Type

2. Risk-First Type

3. Balanced Type

Multi-dimensional Scoring System

Scoring Formula (100-point scale)

Trade Frequency Score Standards

Actual Scoring Case

8.3 Strategy Selection Decision Tree

Decision Flow Chart

Python Automated Selection Script

8.4 Strategy Selection in Different Market Conditions

Market Type Identification

1. Bull Market

2. Bear Market

3. Range-Bound Market

Market Condition Comparison Testing

Strategy Adaptability Matrix

💡 Practical Tasks

Task 1: Batch Backtest 5 Strategies

Task 2: Create Strategy Comparison Table

Task 3: Select the Best Strategy

Task 4: Create Strategy Recommendation List

🥇 Gold Strategy (Comprehensive Best)

🥈 Silver Strategy (Alternative)

🥉 Bronze Strategy (Specific Scenarios)

📚 Knowledge Check

Basic Questions

Answers

Advanced Questions

Thought Questions

🔗 Reference Materials

Supporting Documentation

Recommended Reading

📌 Key Points Summary

➡️ Next Lesson Preview

Method 1: Using `--strategy-list` Parameter