https://github.com/un4b0mber/ai-trading-bot

AI-powered trading bot that predicts BUY / SELL / HOLD signals and executes real-time trades via exchange APIs
https://github.com/un4b0mber/ai-trading-bot

3d ai bot cnn-model convolutional-neural-networks cryptocurrency deep-learning gate-recurrent-unit ml python sequence-models

Last synced: 4 months ago
JSON representation

AI-powered trading bot that predicts BUY / SELL / HOLD signals and executes real-time trades via exchange APIs

• Host: GitHub
• URL: https://github.com/un4b0mber/ai-trading-bot
• Owner: un4b0mber
• License: apache-2.0
• Created: 2025-07-21T15:06:22.000Z (4 months ago)
• Default Branch: main
• Last Pushed: 2025-07-22T09:10:22.000Z (4 months ago)
• Last Synced: 2025-07-22T09:39:47.446Z (4 months ago)
• Topics: 3d, ai, bot, cnn-model, convolutional-neural-networks, cryptocurrency, deep-learning, gate-recurrent-unit, ml, python, sequence-models
• Language: Python
• Homepage:
• Size: 6.84 KB
• Stars: 1
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# 🤖 AI Crypto Trading Bot – Deep Learning with 3D CNN + GRU

A deep learning–powered crypto trading bot using a hybrid 3D CNN + GRU model to predict market signals (`BUY`, `SELL`, `HOLD`) and execute real-time trades via the Binance API.

---

## 🧠 Key Features

- 🔍 **Hybrid Deep Learning Model**  

  Combines 3D Convolutional Neural Networks and GRU layers to analyze spatio-temporal patterns in market data

- 🧾 **Real-Time Trading**  

  Executes live trades using the Binance API, based on the model’s predictions

- 🛠️ **Customizable Parameters**  

  Easily tune training parameters like `epochs`, `batch size`, `learning rate`, etc.

- 💾 **Model State Management**  

  Saves and loads the best-performing model via PyTorch's `state_dict`

- 💹 **Backtesting Module**  

  Run strategy simulations on historical data to evaluate performance

- 🔗 **Binance Integration**  

  Fetches real-time market data (default: BTC/USDT) with easy support for other pairs

- 📦 **Modular Codebase**  

  Designed for clarity and experimentation — each stage is separated and reusable

---

## 🧠 Why 3D CNN + GRU?

- **3D Convolutional Neural Networks (3D CNNs)** are excellent at capturing *spatio-temporal features* — that is, patterns across both indicators (features) and time.

- **GRU (Gated Recurrent Units)** are powerful for *sequence modeling*, allowing the model to remember trends and time dependencies.

- The combination enables the system to recognize complex multi-dimensional market patterns and short-term trends — essential for high-frequency crypto trading.

This hybrid model aims to improve predictive performance over traditional 2D CNNs or LSTMs used alone.

---

## ⚙️ How It Works

The pipeline consists of five core Python scripts, each responsible for a key step in the trading workflow:

1. **Data Preparation**  

   Collects raw crypto data from Binance, computes technical indicators, assigns labels (`BUY`, `SELL`, `HOLD`), normalizes inputs, and applies oversampling to balance the dataset.

2. **Model Training**  

   Defines and trains the hybrid deep learning model using a combination of 3D CNN and GRU layers. Includes the training loop, loss tracking, and model saving.

3. **Backtesting**  

   Tests the trained model on historical data to estimate profitability and performance over a selected time period. Outputs include trade logs and profit metrics.

4. **Dynamic Optimizer**  

   Connects data prep, training, and backtesting in one loop. Automatically adjusts parameters (e.g., learning rate, batch size) every N iterations to search for better results.

5. **Live Trading Bot**  

   Uses the trained model to make predictions in real-time and places trades via the Binance API based on the current market conditions.

---

## Install Dependencies

**Install all required libraries with:**

`pip install requests websocket-client pandas numpy ta torch scikit-learn python-binance python-dotenv`