Algorithmic Trading Systems — Independent AI/ML and labeled-signal research (2024 – Present)
Built an AI multi-agent trading system that uses an LLM and integrates with TradingView via MCP to drive trading decisions. Performed machine learning research using XGBoost with meta-labeling and Triple Barrier labeling to develop scalping signals from market data. Implemented Pine Script v5 indicators and Python trading/bot components to support labeled-signal generation and strategy evaluation. • Engineered LLM-powered multi-agent workflow (Technical, Fundamental, Financial, Mediator) for signal reasoning. • Applied XGBoost with meta-labeling and Triple Barrier labeling on a 3-year XAU/USD dataset. • Developed Pine Script v5 tools for Supply/Demand + Hull Moving Average and EMA breakout/retest/reject strategies. • Implemented Binance Futures execution bot and an MT5 forex trading agent with monitoring dashboard.