2025 Backtested Results
9-agent LangGraph pipeline backtested across 2,847 signals spanning Forex, Stocks, Crypto, Commodities, and Indices. All results are from historical simulations.
Each signal is produced by a 9-agent ensemble: 8 specialist analysts (Fundamental, Technical, Sentiment, Macro, Order Flow, Regime, Correlation, Quant) run in parallel, followed by a bull/bear debate protocol. A TraderAgent then synthesizes the final probability signal. The ensemble is deterministic Python end to end — the figures on this page are reproducible from the same inputs, and were produced without an LLM in the loop. Claude Opus 4.6 and Sonnet 4.6 are available as a reasoning layer on Pro and Enterprise, on top of the same arithmetic; they do not set direction, entry, stop or size.
A signal is marked WIN if price reaches the research target within the analytical window. A signal is marked LOSS if price hits the invalidation level first. Signals that expire without hitting either level are excluded from win rate calculations.
Backtest uses half-Kelly criterion sizing with a 2% max risk per trade. No leverage. No compounding of gains (fixed $10K allocation). Slippage modeled at 0.05% per trade. Commission: $0 (simulating commission-free brokers).
These results are from backtested simulations using historical data. They do not represent actual trading results. Past performance does not guarantee future returns. AI Trading Copilot provides analysis and signals only — it does not execute trades or hold customer funds. This is not financial advice. Trading involves substantial risk of loss. You should only trade with capital you can afford to lose. Contact quantneuraledge@gmail.com for questions.