MCT Trading Bot

MCT (stands for ML-Crypto-Trading) is a high-performance trading engine that features a Dynamic Expression Engine allowing users to define position sizing and risk management logic using familiar, spreadsheet-like functions.

Tip

Financial Logic Without the Code. If you can write an Excel-style formula, you can define trading logic in MCT. The engine provides access to real-time market data, balances, and technical indicators (RSI, EMA, SMA) directly in your configuration.

Caveat Utilitor! For educational and research purposes only.

---

Community

Join our Discord community to discuss strategies, report issues, and collaborate with other traders:

---

Core Features

• 🧠 AI Agent & LangGraph Workflows: Interactive multi-agent assistant with specialized subgraphs for configuration tuning, on-demand backtesting, trading performance analytics, and a proactive Trading Oracle.
• 📐 Dynamic Expression Engine: Define position sizing and custom strategy logic using spreadsheet formulas (rsi, sma, ema, balance, equity, pnl, etc.).
• 🤝 BFT Consensus Engine: Flexible multi-strategy quorum voting supporting 1-of-1 single strategy, 1-of-2 independent triggers, and 2-of-2 strict consensus.
• 📊 High-Fidelity Backtesting: Multi-asset independent clock simulation with realistic execution frictions (slippage, latency, fees), drift detection, and tick replay.
• 🌐 REST API & Real-Time WebSockets: FastAPI server providing order analytics, monthly heatmaps, latency metrics, configuration proposals, and live log streaming (/api/v1/logs/ws).
• 📈 Runtime Observability & Metrics: Built-in collectors for CPU/memory usage, WebSocket health, order lifecycle timings, and exchange telemetry.
• 📜 Configuration Version Control (VCS): Git-like history tracking, audit trail, and rollback for runtime configuration changes.

---

Quick Start

The easiest way to run the bot is using Docker or standard Python.

Prerequisites

• Docker & Docker Compose (or Python 3.11+)
• PostgreSQL: A running instance (local or remote) to store trading history.

1. Configuration

Create your asset configuration and environment variables.

trading-config.yaml

assets:
  - name: "Bitcoin (Crypto.com)"
    base_ticker_symbol: "BTC"
    quote_ticker_symbol: "USD"
    exchange: "CRYPTO_DOT_COM"
    min_quantity: 0.00005
    quote_decimals: 2
    quantity_decimals: 5
    candles_timeframe: "MIN1"
    schedule: 1
    strategies:
      - name: "HammerAccumulationStrategy"
        type: "STATIC"
        class_name: "HammerAccumulationStrategy"
        action: "BUY"
      - name: "RsiOversoldBuy"
        type: "DYNAMIC"
        action: "BUY"
        expression: "rsi(14) < 20"
      - name: "RsiOverboughtSell"
        type: "DYNAMIC"
        action: "SELL"
    consensus:
      buy: 1.3
      sell: 0.5
dynamic_quantity: "max(min_qty, (balance * 0.02) / close)"

Consensus Configurations

MCT uses a Byzantine Fault Tolerant (BFT) voting system where trade actions execute when consensus quorum is satisfied:

$$\text{True Count} \ge \text{Factor} \times (\text{Total} - \text{True Count})$$

You can configure consensus per asset depending on your trading strategy:

• Single Strategy (1-of-1 Consensus): Configure 1 strategy with consensus.buy: 1.0 and consensus.sell: 1.0. The trade triggers whenever that single strategy votes True.

  strategies:
    - name: "RsiOversoldBuy"
      type: "DYNAMIC"
      action: "BUY"
      expression: "rsi(14) < 30"
  consensus:
    buy: 1.0
    sell: 1.0

• Multi Strategy (1-of-2 Consensus — Trade on Each Strategy): Configure 2 strategies with consensus.buy: 0.5 (or any factor $\le 1.0$). If either strategy signals a trade (1 of 2), quorum is met ($1 \ge 0.5 \times 1$).

  strategies:
    - name: "RsiOversoldBuy"
      type: "DYNAMIC"
      action: "BUY"
      expression: "rsi(14) < 30"
    - name: "BreakoutBuy"
      type: "DYNAMIC"
      action: "BUY"
      expression: "close > sma(50) and volume > 100"
  consensus:
    buy: 0.5
    sell: 0.5

• Multi Strategy (2-of-2 Consensus — Trade on Quorum / Agreement): Configure 2 strategies with consensus.buy: 1.3 (or any factor $> 1.0$). A trade triggers only when both strategies agree ($2 \ge 1.3 \times 0$, but $1 < 1.3 \times 1$).

  strategies:
    - name: "RsiOversoldBuy"
      type: "DYNAMIC"
      action: "BUY"
      expression: "rsi(14) < 30"
    - name: "HammerAccumulationStrategy"
      type: "STATIC"
      class_name: "HammerAccumulationStrategy"
      action: "BUY"
  consensus:
    buy: 1.3
    sell: 0.5

.env

APP_ENV=production
CRYPTO_DOT_COM__API_KEY=your_api_key
CRYPTO_DOT_COM__SECRET_KEY=your_secret_key
DATABASE_CONNECTION_HOST=localhost:5432
POSTGRES_USER=postgres
POSTGRES_PASSWORD=your_password
POSTGRES_DATABASE=trading_bot

2. Run Modes

📊 Backtesting (Historical Data & Replay)

Test your strategies against historical CSV or recorded audit data with realistic execution frictions (slippage, latency, fees):

docker run -v $PWD:/workspace toniton/ml-crypto-trading \
  --assets-conf=/workspace/trading-config.yaml \
  --backtest-mode=true \
  --backtest-source=/workspace/history/ \
  --backtest-latency-ms=500.0 \
  --backtest-slippage-ticks=2 \
  --backtest-fee-rate=0.001

🧪 Simulated Trading (Paper Trading)

Run the bot with live market streams while intercepting and executing orders in memory:

docker run --env-file .env -v $PWD:/workspace toniton/ml-crypto-trading \
  --assets-conf=/workspace/trading-config.yaml \
  --simulated=true

🚀 Live Trading (Real Exchange)

Execute live orders on the configured exchange:

docker run --env-file .env -v $PWD:/workspace toniton/ml-crypto-trading \
  --assets-conf=/workspace/trading-config.yaml

🌐 REST API & AI Agent Server

Start the integrated FastAPI and AI Agent server for interactive chat, log streaming, and metric APIs:

docker run --env-file .env -p 8000:8000 -v $PWD:/workspace toniton/ml-crypto-trading \
  --assets-conf=/workspace/trading-config.yaml \
  --server=true \
  --api-port=8000

---

Documentation

For comprehensive information, please explore the documentation in the docs/ directory:

• Introduction & Features — System motivation and comprehensive feature breakdown.
• Core Concepts — Consensus model, Dynamic expressions, AI Agent architecture, and Backtesting.
• Configuration & System Properties — CLI flags, environment variables, and YAML schemas.
• Architecture & Diagrams — Manager layer, Agent graphs, API server, and system flow.
• Logging Architecture — Application, trading, audit replay, and WebSocket log streaming.
• Community & Contributing — Contributing guidelines and Discord community.

---

License

This source code is available on GitHub under the GNU Lesser General Public License v3.0.

Copyright

Copyright © 2026 Toni Akinjiola

All rights reserved except as expressly provided under the applicable license.