Crypto Self-Learning Bot (Template)

A starter template for a self-learning crypto trading bot.
Includes Python backend scripts for data ingestion, feature engineering, a simple ML model (Random Forest), a self-learning loop, backtesting utilities, sample reports, and a minimal web placeholder.

This is a template- you'll need to add your own API keys and tweak strategies before using with real money.

Features

• Fetch OHLCV via CCXT (script stub)
• Compute common technical indicators (RSI, SMA, EMA)
• Train a simple supervised ML model (RandomForest) to predict short-term movement
• Simple self-learning loop that simulates trades and updates the model
• Backtesting metrics and sample JSON report
• GitHub Actions workflow for scheduled runs

Quickstart (local)

1. Create a Python virtualenv:

python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

2. Edit .env (create from .env.example) with your keys and config.

3. Run a backtest (sample):

python -m src.bot --symbol BTC/USDT --timeframe 1h --mode backtest

4. Run live (VERY RISKY - only after testing):

python -m src.bot --symbol BTC/USDT --timeframe 1h --mode paper

Structure

crypto-self-learning-bot/
├─ src/
│  ├─ data.py
│  ├─ features.py
│  ├─ model.py
│  ├─ bot.py
├─ config/
│  └─ bot.config.json
├─ reports/
│  └─ sample-report.json
├─ web/
│  └─ README.md
├─ .github/workflows/daily-bot.yml
├─ .env.example
├─ requirements.txt
└─ README.md

Notes & Warnings

• This template is educational. Do not use with real funds until thoroughly tested.
• Trading cryptocurrencies involves substantial risk.

Crypto Self-Learning Bot

A full-stack crypto strategy lab with a Python backtest engine and a production-style Next.js dashboard.

This project gives you a complete loop:

• load market data,
• engineer technical features,
• train a model,
• run a trading simulation,
• persist metrics/history,
• and visualize everything in a polished web app.

What is included

Backend (Python)

• Data layer for sample CSV and live exchange OHLCV via ccxt.
• Feature engineering pipeline (SMA, EMA, RSI, MACD, returns).
• Supervised model wrapper (RandomForestClassifier).
• Backtesting engine with long-only execution logic.
• Rich report generation with:
  • final equity,
  • return vs buy/hold,
  • max drawdown,
  • Sharpe,
  • Calmar,
  • trade count, win rate, exposure.
• Persistent artifacts:
  • reports/sample-report.json (latest run),
  • reports/backtest-history.jsonl (run archive),
  • reports/latest-equity-curve.csv (chart-ready time series),
  • reports/model.joblib (optional saved model).

Frontend (Next.js)

• Executive dashboard with branded layout and visuals.
• KPI cards, equity curve chart, market context chart, and recent run table.
• Server-side data loading from backend artifacts.
• API routes for integration:
  • GET /api/report
  • GET /api/config
  • GET /api/history
  • GET /api/equity

Repository structure

• src/data.py data loading (CSV + exchange)
• src/features.py indicators and feature columns
• src/model.py ML training/inference wrapper
• src/bot.py CLI entrypoint and report pipeline
• config/bot.config.json default strategy/runtime settings
• reports/ generated run artifacts
• web/ Next.js dashboard

Quickstart

1) Backend setup

python3 -m venv .venv
./.venv/bin/pip install -r requirements.txt

Run a backtest:

./.venv/bin/python -m src.bot --mode backtest --save_model

The command updates:

• reports/sample-report.json
• reports/backtest-history.jsonl
• reports/latest-equity-curve.csv
• reports/model.joblib (when --save_model is used)

2) Web setup

cd web
npm install
npm run dev

Open http://localhost:3000.

Production build:

npm run build
npm run start

CLI options (backend)

Main command:

./.venv/bin/python -m src.bot --mode backtest [options]

Important options:

• --symbol trading pair (default BTC/USDT)
• --timeframe candle timeframe (default 1h)
• --initial_balance starting capital (default 10000)
• --fee_rate transaction fee per trade (default 0.001)
• --predict_horizon target horizon in candles
• --data_mode sample or exchange
• --data_path custom CSV path when using sample mode
• --exchange_name exchange id for ccxt (default binance)
• --limit OHLCV row limit in exchange mode
• --report_dir artifact output directory (default reports)
• --save_model persist model to disk
• --model_path model output path (default reports/model.joblib)

Running with live exchange candles

./.venv/bin/python -m src.bot \
  --mode backtest \
  --data_mode exchange \
  --exchange_name binance \
  --symbol BTC/USDT \
  --timeframe 1h \
  --limit 1200 \
  --save_model

Safety and disclaimer

This software is for research and education. It is not investment advice. Crypto trading is high risk. Do not deploy with real capital without robust validation, controls, and monitoring.

License

See LICENSE.

(Full app build: backend, dashboard, and professional docs)

(Resolve merge conflicts and fix bot syntax error)