gavindiaz/LSTM-PPO-Reinforcement-Learning-Trading-Bot-for-MetaTrader-5-XAUUSD-
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
Vittus Mikiassen
2026-06-29 16:23:10 +02:00
committed by GitHub
parent dd39f474bb
commit b2ff18180e
1 changed files with 291 additions and 100 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+291 -100
View File
@@ -1,12 +1,8 @@
# MT5 XAUUSD LSTM PPO Trading Bot
# xLSTM PPO Reinforcement Learning Trading Bot for MetaTrader 5 (XAUUSD)
A MetaTrader 5 reinforcement learning trading bot for XAUUSD (Gold).
An AI-powered trading bot that combines **Extended LSTM (xLSTM)** and **Proximal Policy Optimization (PPO)** to learn profitable trading strategies on XAUUSD.
The bot combines an LSTM neural network for sequence learning with Proximal Policy Optimization (PPO) to learn trading decisions directly from historical market data.
Unlike traditional bots that rely on fixed rules, the PPO agent learns when to Buy, Sell or Hold from thousands of market examples.
The program should be in a folder or the desktop where the 5m csv file with OHLC data downloaded from dukascopy, 3 years. is, in training it will then make a LSTM-PPO-saves folder where the training is saved (used for making decisions, also in testing).
Unlike traditional indicator-based Expert Advisors, this project learns directly from historical market data using reinforcement learning while incorporating Smart Money Concepts (SMC), technical indicators, market structure, and adaptive risk management.
---
@@ -14,112 +10,138 @@ The program should be in a folder or the desktop where the 5m csv file with OHLC
### Reinforcement Learning
- LSTM policy network
- PPO training
- Continuous online training
- Live inference on MT5
- Automatic checkpoint saving/loading
- xLSTM policy network
- Proximal Policy Optimization (PPO)
- Sequence-based learning
- Reinforcement learning trading environment
- Reward shaping
- GPU/CPU training support
### Technical Indicators
---
## Smart Money Concepts (SMC)
The bot automatically detects:
- Order Blocks (OB)
- Mitigated Order Blocks
- Breaker Blocks
- Market Blocks (MB)
- Rejection Blocks (RB)
- Fair Value Gaps (FVG)
- Inverse Fair Value Gaps (IFVG)
- FVG Retracements
- Equal Highs (EQH)
- Equal Lows (EQL)
- Market Structure Shifts (MSS)
- Swing Highs
- Swing Lows
- Previous Day High (PDH)
- Previous Day Low (PDL)
- Asia High
- Asia Low
- Trend Lines
- Liquidity Sweeps
---
## Technical Indicators
Included features include:
- EMA 7
- EMA 21
- EMA Difference (Momentum)
- EMA Difference
- EMA Slopes
- ADX
- +DI
- -DI
- Stochastic
- Stochastic Oscillator
- VWAP
- VWAP Bands
- VWAP Slope
- VWAP Upper Band
- VWAP Lower Band
- Volume Moving Average
### Smart Money Concepts
- Bullish Order Blocks
- Bearish Order Blocks
- Bullish Fair Value Gaps
- Bearish Fair Value Gaps
- Bullish Rejection Blocks
- Bearish Rejection Blocks
- Equal Highs
- Equal Lows
- Market Breaks
- Session Detection
- Indecision Candles
### PPO State Features
---
Current state contains:
## Adaptive Trade Management
- OHLC
- EMA trend
The bot automatically manages:
- Adaptive Stop Loss
- Adaptive Take Profit
- Partial Profit Taking
- Break-even Movement
- Dynamic Position Sizing
- Risk-based Lot Calculation
---
## Machine Learning Features
The model learns from more than 50 engineered features including:
- Smart Money Concepts
- Trend
- Momentum
- ADX trend strength
- DI Direction
- Stochastic
- VWAP
- VWAP Bands
- VWAP Position
- VWAP Slope
- Volume MA
- Order Blocks
- Fair Value Gaps
- Rejection Blocks
- Equal Highs/Lows
- Buy Score
- Sell Score
- Liquidity
- Session information
- Price distances
- Volume
- Volatility
- Historical sequences
---
## Current Strategy
## Trading Environment
Current reward structure:
The custom Gymnasium environment supports:
- Take Profit: 20 pips
- Stop Loss: 40 pips
- Risk/Reward: 1 : 0.5
- Buy
- Sell
- Hold
The current focus is high-probability momentum trades rather than large swing trades.
Rewards consider:
- Profit
- Drawdown
- Risk
- Trade quality
- Spread
- Commission
---
## Training
## Project Structure
The PPO agent trains continuously over historical MT5 data.
```
.
├── bot.py # Main application
├── download/ # Historical CSV data
├── models/ # Saved models
├── logs/ # Training logs
├── scaler.pkl # Feature scaler
└── README.md
```
Example metrics during training:
---
- Win rate: 70–85%
- Profit Factor: 1.5–3+
- Weekly performance: typically 10–30R during training (varies by market conditions)
## Installation
These figures are training statistics only and are not guarantees of future performance.
Clone the repository
### Quarterly stats
```bash
git clone https://github.com/pressure679/LSTM-PPO-Reinforcement-Learning-Trading-Bot-for-MetaTrader-5-XAUUSD-
cd LSTM-PPO-Reinforcement-Learning-Trading-Bot-for-MetaTrader-5-XAUUSD-
```
## Weekly PPO Training Performance by Quarter
Install dependencies
Training period: **~June 2024 – June 1, 2026** (102 rolling weekly training windows)
| Period | Approx. Dates | Avg Weekly R | Avg PF | Avg Max DD | Avg Recovery Factor |
| ------ | ---------------------- | -----------: | -----: | ---------: | ------------------: |
| Q1 | Jun 2024 – Sep 2024 | 6.20R | 1.10 | 8.80R | 2.04 |
| Q2 | Sep 2024 – Dec 2024 | 7.39R | 1.12 | 9.80R | 2.25 |
| Q3 | Dec 2024 – Mar 2025 | 16.89R | 1.25 | 8.14R | 5.42 |
| Q4 | Mar 2025 – Jun 2025 | 43.73R | 1.29 | 10.81R | 9.20 |
| Q5 | Jun 2025 – Sep 2025 | 36.54R | 1.38 | 8.94R | 8.72 |
| Q6 | Sep 2025 – Dec 2025 | 75.07R | 1.38 | 8.62R | 19.42 |
| Q7 | Dec 2025 – Mar 2026 | 164.37R | 1.82 | 7.23R | 54.69 |
| Q8* | Mar 2026 – Jun 1, 2026 | 126.55R | 1.49 | 8.76R | 34.54 |
*Q8 contains the final 11 weeks of training.
### Observations
* Profit Factor increased from approximately **1.10** during the earliest training period to **1.4–1.8** in the later periods.
* Average weekly drawdown remained relatively stable between **7R and 10R** despite substantially higher returns.
* Recovery Factor improved significantly over time, indicating that profitability increased faster than drawdown.
* The strongest performance occurred during the final two quarters while maintaining comparable risk characteristics.
```bash
pip install -r requirements.txt
```
---
@@ -127,45 +149,214 @@ Training period: **~June 2024 – June 1, 2026** (102 rolling weekly training wi
- Python 3.11+
- MetaTrader 5
- MetaTrader5
- pandas
- numpy
- torch
Install dependencies:
Python packages:
```bash
pip install MetaTrader5 pandas numpy torch
```
torch
stable-baselines3
sb3-contrib
gymnasium
MetaTrader5
numpy
pandas
scikit-learn
ta
xlstm
```
---
## Running
## Historical Data
Train:
Place historical data inside the `download/` folder.
Example:
```
download/
└── xauusd-m5-bid-2023-01-01-2026-06-01.csv
```
The bot automatically calculates all technical indicators and Smart Money Concept features before training.
---
## Training
Train the reinforcement learning model:
```bash
python bot.py --train
```
Live trading:
Training automatically:
- Loads historical data
- Calculates technical indicators
- Builds SMC features
- Normalizes inputs
- Creates the trading environment
- Trains the PPO agent
- Saves the trained model
---
## Live Trading
Run the live trading bot:
```bash
python bot.py --test --symbol "XAUUSD"
python bot.py --test
```
Train and trade simultaneously:
The bot will:
```bash
python mt5-xau-lstm-ppo-bot.py --train --test
- Connect to MetaTrader 5
- Read live candles
- Generate feature vectors
- Predict actions
- Execute trades
- Manage open positions automatically
---
## Feature List
Current feature set includes:
```
k
k_smooth
adx
+di
-di
EMA721_DIFF
EMA7_Slope
EMA21_Slope
indecision
bullish_ob
bearish_ob
bullish_fvg
bearish_fvg
bullish_ifvg
bearish_ifvg
eqh
eql
bullish_mb
bearish_mb
bullish_rb
bearish_rb
bullish_bb
bearish_bb
bullish_mss
bearish_mss
bullish_ob_mitigation
bearish_ob_mitigation
bullish_fvg_retracement
bearish_fvg_retracement
bullish
bearish
above_vwap
below_vwap
pdh
pdl
asia_high
asia_low
pdh_distance
pdl_distance
asia_high_distance
asia_low_distance
vwap
vwap_upper
vwap_lower
bullish_ifvg_retracement
bearish_ifvg_retracement
swing_high
swing_low
support_trend_line
resistance_trend_line
seq_len
session
```
To train download a m5 xauusd csv file form kaggle or dukascopy and place the folder in "download" relative to the directory the mt5-xau-lstm-ppo-bot.py is in.
---
## Risk Management
Supports:
- Fixed percentage risk
- Dynamic lot sizing
- Adaptive stop losses
- Adaptive take profits
- Partial exits
- Break-even protection
- Maximum open positions
---
## Training Statistics
The bot reports:
- Win Rate
- Profit Factor
- Sharpe Ratio
- Sortino Ratio
- Weekly Return
- Maximum Drawdown
- Mean Win
- Mean Loss
- Average R
- Trade Count
---
## Roadmap
- [ ] Multi-symbol training
- [ ] Multi-timeframe observations
- [ ] Prioritized Experience Replay
- [ ] Hyperparameter optimization
- [ ] Walk-forward validation
- [ ] ONNX export
- [ ] Transformer/xLSTM hybrid policy
- [ ] Trade visualization dashboard
- [ ] Pattern similarity search
- [ ] Explainable AI trade analysis
---
## Disclaimer
This project is for educational and research purposes only.
This project is intended for educational and research purposes only.
Trading leveraged products involves substantial risk. Always test thoroughly on historical data and demo accounts before risking real capital.
Trading financial markets involves substantial risk. Past performance does not guarantee future results.
Always test on a demo account before trading with real funds.
---
## License
This project is licensed under the MIT License.