ThorpeBot Overview

ThorpeBot is a cryptocurrency trading bot that leverages the power of multiple Large Language Models (LLMs) to make trading decisions by consensus. It combines insights from three different AI models:

• Gerald (OpenAI) - Powered by GPT-4, provides sophisticated market analysis
• Claude (Anthropic) - Delivers careful reasoning and risk assessment
• Will (Fireworks) - Offers diverse perspectives based on open-source models

By requiring consensus among these models, ThorpeBot aims to reduce bias and increase confidence in trading decisions.

System Architecture

ThorpeBot consists of several interconnected components:

LLM Consensus Mechanism

ThorpeBot's core innovation is its LLM consensus mechanism:

1. Each model (Gerald, Claude, Will) independently analyzes the same market data
2. Models recommend BUY, SELL, or HOLD with confidence scores
3. A decision requires at least 2 out of 3 models to agree
4. If consensus is reached, the trading action is executed

Backtest Mode

The current Vercel deployment runs in demo mode, which simulates trading activity:

• Uses historical or simulated price data
• Demonstrates the consensus mechanism with realistic model responses
• Shows portfolio performance over time
• All transactions are simulated (no real trades)

To start a backtest, click the "Run Backtest" button in the Model Chat page.

API Reference

ThorpeBot provides a RESTful API for data access:

Deployment Options

ThorpeBot can be deployed in different ways:

git clone https://github.com/yourusername/tradebot.git
cd tradebot
docker-compose up -d