TradeBot — Learning Market Dynamics from Synthetic Correlations

Simulate common trading bots on real crypto data and prove that a neural networks is able to capture the pattern!

This project proves that a neural network can learn meaningful market patterns that stem from commonly used pre-programmed bots on the real market. It also permits its implementation in a real-time trading bot, and includes different tools for getting, analyzing and pre-processing data. It is meant to make this whole process more easy!

First install with:

pip install tradebot

Example code:

For a more detailed tutorial that goes through all functions of this library, i uploaded a Kaggle notebook, showing the capture of a real-time trading bot:

🧠 Project Overview

1. Load historical BTCUSDT data
2. Compute technical indicators and prepare model inputs
3. Diagnose data distributions and correlations
4. Generate synthetic market data with controlled structure
5. Train neural network on real vs synthetic data
6. Simulate a trading bot using trained model