
AlphaTrader is an autonomous AI trading agent built for the AI Trading Agents Hackathon (March 30 - April 12, 2026). The system combines four complementary trading strategies to maximize risk-adjusted returns while minimizing drawdown. ARCHITECTURE: Core Strategies: - RSI + Bollinger Bands (50% weight): Identifies overbought/oversold conditions through technical analysis - Momentum Strategy (30% weight): Captures trending moves based on 5-day price acceleration - Mean Reversion (20% weight): Exploits price deviations from 20-day moving averages Ensemble Voting: Final signals generated via weighted voting system. If aggregate confidence exceeds threshold, agent executes trades autonomously. Integration & Execution: - Kraken CLI for live market data and trade execution - Claude API for intelligent decision-making and strategy adaptation - SQLite database for audit trail and performance tracking - Streamlit dashboard for real-time portfolio monitoring Risk Management: - Position sizing: Max 20% of capital per trade - Stop loss: -5% per position - Take profit: +10% per position - Daily loss limit: -10% of portfolio Expected Performance: - Target: +15-20% ROI over 2 weeks - Win rate: 55-60% (ensemble advantage) - Sharpe ratio: >1.5 The agent trades autonomously 24/7, logging every decision (strategy, signal confidence, result) for transparency and hackathon evaluation. Technology Stack: Python 3.11, Kraken CLI, SQLite, Streamlit, Claude API, Pandas.
12 Apr 2026