DEV Community

Henry Lin
Henry Lin

Posted on

Lesson 16: Real-time Signal Monitoring

#cryptocurrency #tutorial #testing #monitoring

Lesson 16: Real-time Signal Monitoring

โฑ Duration: 2 hours
๐ŸŽฏ Learning Objectives: Learn to obtain real-time trading signals
๐Ÿ“š Difficulty: โญโญ Real-time Signals

๐Ÿ“– Course Overview

Backtesting is just the first step; the real test is live trading. This lesson will teach you how to start Freqtrade's paper trading mode (Dry-run), monitor trading signals in real-time, and prepare for live trading.

16.1 Dry-run Mode Introduction

What is Dry-run?

Definition: Paper trading mode that uses real market data but does not execute actual trades.

Differences from Backtesting:

Feature Backtesting Paper Trading (Dry-run)
Data Historical data Real-time data
Execution Fast (minutes) Real-time (continuous)
Purpose Verify strategy historical performance Verify strategy real-time performance
Slippage Cannot simulate Partially simulated
Order Execution Assume immediate fill Simulate real delays
Risk Zero risk Zero risk

Why Need Dry-run?

Reason 1: Verify Strategy Real-time Performance

Backtesting performance: +25% (historical data)
Dry-run performance: +18% (real-time data) โœ… Basically consistent, can go live

Or:
Backtesting performance: +25%
Dry-run performance: -5% (real-time data) โŒ Too big difference, strategy has issues
Enter fullscreen mode Exit fullscreen mode

Reason 2: Familiarize with Live Environment

  • Understand signal generation timing
  • Familiarize with order execution process
  • Test notification and monitoring systems

Reason 3: Discover Potential Issues

  • Strategy code errors
  • Configuration file problems
  • Network connection issues
  • Data delay issues

Dry-run Limitations

โš ๏ธ Cannot Fully Simulate:

  • Slippage
  • Market depth impact
  • Liquidity issues in extreme market conditions
  • Exchange system failures

๐Ÿ’ก Recommendation: Dry-run for 1-2 weeks before considering live trading

16.2 Starting Paper Trading

Preparation

1. Check Configuration File

Ensure config.json is correctly configured:

{
  "max_open_trades": 3,
  "stake_currency": "USDT",
  "stake_amount": 100,
  "dry_run": true,  // โš ๏ธ Ensure this is true
  "dry_run_wallet": 1000,  // Paper wallet amount

  "exchange": {
    "name": "binance",
    "key": "",  // Dry-run doesn't need API key
    "secret": "",
    "ccxt_config": {},
    "ccxt_async_config": {},
    "pair_whitelist": [
      "BTC/USDT",
      "ETH/USDT",
      "BNB/USDT"
    ]
  },

  "strategy": "Strategy001",
  "timeframe": "5m"
}
Enter fullscreen mode Exit fullscreen mode

Key Configuration Notes:

  • dry_run: true: Enable simulation mode
  • dry_run_wallet: Initial paper account amount
  • No need to fill real API key/secret

2. Select Strategy 

# View available strategies
freqtrade list-strategies -c config.json

# Output example
Strategy001
Strategy002
MomentumTrendStrategy
MeanReversionStrategy
Enter fullscreen mode Exit fullscreen mode

Startup Commands

Basic Startup 

# Activate environment
conda activate freqtrade

# Start Dry-run
freqtrade trade -c config.json --strategy Strategy001
Enter fullscreen mode Exit fullscreen mode

Output Example:

2025-09-30 10:00:00 - freqtrade - INFO - Starting freqtrade in Dry-run mode
2025-09-30 10:00:00 - freqtrade - INFO - Using strategy: Strategy001
2025-09-30 10:00:00 - freqtrade - INFO - Timeframe: 5m
2025-09-30 10:00:00 - freqtrade - INFO - Dry-run mode enabled
2025-09-30 10:00:00 - freqtrade - INFO - Starting worker threads
2025-09-30 10:00:05 - freqtrade - INFO - Bot started. Press Ctrl+C to stop.
Enter fullscreen mode Exit fullscreen mode

Background Running 

# Run in background using nohup
nohup freqtrade trade -c config.json --strategy Strategy001 > freqtrade.log 2>&1 &

# View logs
tail -f freqtrade.log

# View processes
ps aux | grep freqtrade

# Stop
pkill -f freqtrade
Enter fullscreen mode Exit fullscreen mode

Using systemd (Linux Recommended)

Create service file /etc/systemd/system/freqtrade.service:

[Unit]
Description=Freqtrade Trading Bot
After=network.target

[Service]
Type=simple
User=your_username
WorkingDirectory=/home/your_username/freqtrade
ExecStart=/home/your_username/anaconda3/envs/freqtrade/bin/freqtrade trade -c config.json --strategy Strategy001
Restart=always
RestartSec=10

[Install]
WantedBy=multi-user.target
Enter fullscreen mode Exit fullscreen mode

Start service:

# Reload configuration
sudo systemctl daemon-reload

# Start service
sudo systemctl start freqtrade

# Start on boot
sudo systemctl enable freqtrade

# View status
sudo systemctl status freqtrade

# View logs
sudo journalctl -u freqtrade -f
Enter fullscreen mode Exit fullscreen mode

16.3 Real-time Signal Viewing

1. Terminal Output

Dry-run outputs signals in real-time when running:

2025-09-30 10:05:00 - freqtrade - INFO - New candle for BTC/USDT - 5m
2025-09-30 10:05:02 - freqtrade - INFO - Found BUY signal for BTC/USDT
2025-09-30 10:05:03 - freqtrade - INFO - Buy signal: BTC/USDT at 43500.00 USDT

2025-09-30 10:05:05 - freqtrade - INFO - Opened order #1
โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”“
โ”ƒ ID        โ”ƒ Pair        โ”ƒ Open Rate  โ”ƒ Amount   โ”ƒ Open Date  โ”ƒ
โ”กโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ฉ
โ”‚ 1         โ”‚ BTC/USDT    โ”‚ 43500.00   โ”‚ 0.00230  โ”‚ 10:05:03   โ”‚
โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜
Enter fullscreen mode Exit fullscreen mode

2. Log Files

View detailed logs:

# Real-time viewing
tail -f user_data/logs/freqtrade.log

# Filter specific information
tail -f user_data/logs/freqtrade.log | grep "BUY\|SELL"

# View recent errors
grep "ERROR" user_data/logs/freqtrade.log | tail -20
Enter fullscreen mode Exit fullscreen mode

3. View Current Positions 

# Method 1: Use freqtrade command (need API enabled)
freqtrade show_trades -c config.json

# Method 2: View database
sqlite3 user_data/tradesv3.dryrun.sqlite "SELECT * FROM trades WHERE is_open=1;"
Enter fullscreen mode Exit fullscreen mode

Output Example:

Current trades:
โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”“
โ”ƒ ID  โ”ƒ Pair      โ”ƒ Open Rate  โ”ƒ Current  โ”ƒ Profit% โ”ƒ Duration   โ”ƒ
โ”กโ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ฉ
โ”‚ 1   โ”‚ BTC/USDT  โ”‚ 43500.00   โ”‚ 43680.00 โ”‚ +0.41%  โ”‚ 1h 25m     โ”‚
โ”‚ 2   โ”‚ ETH/USDT  โ”‚ 2280.00    โ”‚ 2295.00  โ”‚ +0.66%  โ”‚ 45m        โ”‚
โ””โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜
Enter fullscreen mode Exit fullscreen mode

4. Performance Statistics 

# View statistics
freqtrade show-stats -c config.json
Enter fullscreen mode Exit fullscreen mode

Output Example:

Performance:
โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ณโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”“
โ”ƒ Pair        โ”ƒ Trades  โ”ƒ Win Rate   โ”ƒ Profit %   โ”ƒ Duration โ”ƒ
โ”กโ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ•‡โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”โ”ฉ
โ”‚ BTC/USDT    โ”‚ 15      โ”‚ 80.0%      โ”‚ +2.35%     โ”‚ 3h 22m   โ”‚
โ”‚ ETH/USDT    โ”‚ 12      โ”‚ 75.0%      โ”‚ +1.85%     โ”‚ 2h 45m   โ”‚
โ”‚ BNB/USDT    โ”‚ 8       โ”‚ 87.5%      โ”‚ +1.52%     โ”‚ 4h 10m   โ”‚
โ”œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ผโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ค
โ”‚ TOTAL       โ”‚ 35      โ”‚ 80.0%      โ”‚ +5.72%     โ”‚ 3h 25m   โ”‚
โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”ดโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜
Enter fullscreen mode Exit fullscreen mode

16.4 Signal Analysis and Recording

1. Record Buy/Sell Signals

Create signal recording table (Excel/Google Sheets):

DateTime Trading Pair Signal Type Price Indicator Status Market Environment Notes
2025-09-30 10:05 BTC/USDT BUY 43500 RSI=28, EMA Golden Cross Ranging up -
2025-09-30 12:30 BTC/USDT SELL 43850 RSI=68 Short-term resistance +0.80%
2025-09-30 14:20 ETH/USDT BUY 2280 Bollinger Lower Band Oversold bounce -

2. Daily Summary Template 

# Dry-run Log - 2025-09-30

## Today's Overview
- Running time: 16 hours
- Trading signals: 7 buys, 5 sells
- Current positions: 2 trades
- Today's P&L: +1.35%

## Signal Details

### Buy Signals
1. **10:05 BTC/USDT @ 43500**
   - Conditions: EMA20 crosses above EMA50 + RSI<30
   - Result: In position (+0.80%)

2. **14:20 ETH/USDT @ 2280**
   - Conditions: Price touches Bollinger lower band + Volume expansion
   - Result: In position (+0.65%)

### Sell Signals
1. **12:30 BTC/USDT @ 43680**
   - Trigger: ROI 2% target
   - Profit: +0.41%
   - Holding time: 2h 25m

## Issues and Observations
- โš ๏ธ Found one false breakout signal (BNB/USDT 15:30)
- โœ… Stop loss setting reasonable, not triggered
- ๐Ÿ’ก Many signals in ranging market, need to add filter conditions

## Tomorrow's Plan
- Continue observing ranging market performance
- Consider adjusting RSI thresholds
Enter fullscreen mode Exit fullscreen mode

3. Signal Quality Assessment

Assessment Dimensions:

# Buy signal quality
def evaluate_entry_signal(trade):
    score = 0

    # 1. Profit rate (40 points)
    if trade.profit > 2.0:
        score += 40
    elif trade.profit > 1.0:
        score += 30
    elif trade.profit > 0.5:
        score += 20
    elif trade.profit > 0:
        score += 10

    # 2. Holding time (30 points)
    if 2 <= trade.duration_hours <= 8:
        score += 30  # Ideal holding time
    elif trade.duration_hours < 2:
        score += 15  # Too short
    else:
        score += 20  # Too long

    # 3. Exit method (30 points)
    if trade.exit_reason == 'roi':
        score += 30  # ROI exit is best
    elif trade.exit_reason == 'exit_signal':
        score += 25
    elif trade.exit_reason == 'trailing_stop':
        score += 20
    else:
        score += 10  # Stop loss

    return score

# Rating
if score >= 80: print("Excellent signal โญโญโญโญโญ")
elif score >= 60: print("Good signal โญโญโญโญ")
elif score >= 40: print("Average signal โญโญโญ")
else: print("Poor signal โญโญ")
Enter fullscreen mode Exit fullscreen mode

16.5 Common Problem Handling

Problem 1: Frequent Buying/Selling (Overtrading)

Phenomenon:

Multiple signals per hour
Holding time < 30 minutes
Fee ratio > 20%
Enter fullscreen mode Exit fullscreen mode

Causes:

  • Strategy conditions too loose
  • Market volatility causing frequent triggers
  • Time frame too short

Solutions:

# Add signal cooldown period
from datetime import timedelta

class Strategy001(IStrategy):
    last_entry_time = {}
    cooldown_minutes = 60  # 60 minute cooldown

    def populate_entry_trend(self, dataframe, metadata):
        pair = metadata['pair']
        current_time = dataframe['date'].iloc[-1]

        # Check cooldown period
        if pair in self.last_entry_time:
            time_since_last = current_time - self.last_entry_time[pair]
            if time_since_last < timedelta(minutes=self.cooldown_minutes):
                return dataframe  # Still in cooldown, no signal

        # Normal signal logic
        dataframe.loc[
            (your_conditions),
            'enter_long'] = 1

        # Update last entry time
        if dataframe['enter_long'].iloc[-1] == 1:
            self.last_entry_time[pair] = current_time

        return dataframe
Enter fullscreen mode Exit fullscreen mode

Problem 2: No Signals for Long Time

Phenomenon:

No buy signals for 6 hours
Enter fullscreen mode Exit fullscreen mode

Causes:

  • Strategy conditions too strict
  • Market doesn't match strategy environment
  • Data acquisition issues

Check Steps:

# 1. Check data updates
freqtrade list-data -c config.json

# 2. View strategy indicators (enable debug mode)
freqtrade trade -c config.json --strategy Strategy001 -vvv

# 3. Loosen condition strictness (temporary test)
# Modify strategy, relax conditions
Enter fullscreen mode Exit fullscreen mode

Problem 3: Signals Inconsistent with Backtesting

Phenomenon:

Backtesting: 80% win rate, +20% return
Dry-run: 50% win rate, -5% return
Enter fullscreen mode Exit fullscreen mode

Possible Causes:

  1. Overfitting: Strategy only suitable for historical data
  2. Data Differences: Backtesting data different from real-time data
  3. Execution Delays: Real-time signals have delays
  4. Market Environment Changes: Market conditions different from backtesting period

Response Methods:

# 1. Out-of-sample validation
freqtrade backtesting -c config.json --strategy Strategy001 --timerange 20251001-20251231

# 2. Extend real-time validation period
# At least dry-run for 2 weeks before evaluation

# 3. Check if strategy is overfitted
# See if parameters are too precise (like RSI=27.3)
Enter fullscreen mode Exit fullscreen mode

Problem 4: Bot Crashes or Stops

Error Log Example:

ERROR - Exchange binance does not support fetching OHLCV data
ERROR - Network error: Connection timeout
ERROR - Database locked
Enter fullscreen mode Exit fullscreen mode

Solutions:

# Network issues
# Check network connection
ping api.binance.com

# Use proxy (if needed)
export HTTP_PROXY=http://127.0.0.1:7890
export HTTPS_PROXY=http://127.0.0.1:7890

# Database locked
# Stop all freqtrade processes
pkill -f freqtrade

# Delete database lock files
rm user_data/tradesv3.dryrun.sqlite-shm
rm user_data/tradesv3.dryrun.sqlite-wal

# Restart
freqtrade trade -c config.json --strategy Strategy001
Enter fullscreen mode Exit fullscreen mode

๐Ÿ’ก Practical Tasks

Task 1: Start First Dry-run 

# 1. Check configuration
cat config.json | grep "dry_run"

# 2. Start Dry-run
freqtrade trade -c config.json --strategy Strategy001

# 3. Let it run for at least 1 hour

# 4. View statistics
freqtrade show-stats -c config.json
Enter fullscreen mode Exit fullscreen mode

Task 2: 24-Hour Monitoring

Run Dry-run for at least 24 hours, record:

## 24-Hour Dry-run Report

### Operation Overview
- Start time: _______
- End time: _______
- Strategy name: _______
- Time frame: _______

### Trading Statistics
- Total signals: _______
- Buy signals: _______
- Sell signals: _______
- Current positions: _______
- Average holding time: _______

### P&L Statistics
- Total return: _______%
- Profitable trades: _______ trades
- Losing trades: _______ trades
- Win rate: _______%
- Max drawdown: _______%

### Issues Recorded
1. _______
2. _______
3. _______

### Conclusion
โ˜ Performance meets backtesting expectations, can continue
โ˜ Poor performance, need to adjust strategy
โ˜ Found serious issues, need to redesign
Enter fullscreen mode Exit fullscreen mode

Task 3: Compare Backtesting vs Dry-run 

# Backtesting (same time period)
freqtrade backtesting -c config.json --strategy Strategy001 --timerange 20250928-20250930

# Record backtesting results
# Compare with actual Dry-run performance
Enter fullscreen mode Exit fullscreen mode

Comparison table:

Metric Backtesting Dry-run Difference
Trade Count ? ? ?%
Win Rate ?% ?% ?%
Total Return ?% ?% ?%
Max Drawdown ?% ?% ?%
Average Holding ? ? ?

Analysis:

If Dry-run performance is close to backtesting (within ยฑ20%):
  โœ… Strategy reliable, can consider live trading

If Dry-run performance is much lower than backtesting:
  โš ๏ธ May have overfitting or other issues
  Need to extend observation period or adjust strategy
Enter fullscreen mode Exit fullscreen mode

๐Ÿ“Œ Key Points Summary

  1. Dry-run is essential before live trading: Run at least 1-2 weeks
  2. Real-time performance โ‰  backtesting performance: Allow 20% difference
  3. Record and analyze every signal: Build signal quality assessment system
  4. Adjust problems in time: Don't blindly enter live trading
  5. Focus on strategy adaptability: Performance in different market conditions
  6. Be patient: Dry-run is a learning and debugging process

โžก๏ธ Next Lesson Preview

Lesson 17: Telegram Notification Configuration

In the next lesson, we will learn:

  • Create Telegram Bot
  • Configure notification settings
  • Receive trading signal pushes
  • Control Bot remotely through Telegram

Preparation:

  • โœ… Complete at least 24 hours of Dry-run
  • โœ… Register Telegram account
  • โœ… Prepare device to receive notifications

๐ŸŽฏ Learning Assessment Criteria:

  • โœ… Can independently start Dry-run mode
  • โœ… Can view and record real-time signals
  • โœ… Can analyze signal quality
  • โœ… Can handle common problems

After completing these tasks, you've taken the first step toward live trading! ๐Ÿš€

Top comments (0)