Trust Layer
Transparency Center
TradingGPT Pro is decision-support software for market research, strategy testing, alerts, and controlled execution workflows. It is not a broker, exchange, investment adviser, or guarantee engine.
AI Scope
- Research Mode produces analysis, explanations, summaries, and Trade Plan Audit context.
- Execution Mode requires user-connected broker access, write permission, and risk acknowledgement.
- Automation Mode is paper-first; live mode requires manual activation, approval, limits, and stop-loss controls.
- AI confidence is qualitative unless a verified historical cohort exists.
Known Limits
- Market data can be delayed, missing, stale, or unavailable depending on provider and instrument.
- News interpretation can miss context or change after new facts arrive.
- Backtests can be overfit and may exclude slippage, spread, commission, borrow cost, liquidity, or regime changes.
- AI can hallucinate. Critical trade decisions must be checked against live data and broker order tickets.
Backtest Truth Engine
- Grades sample size, drawdown, risk-adjusted return, exposure realism, data source, and stress proxy.
- Flags missing trading-cost assumptions and insufficient trade count.
- Does not prove future performance or replace walk-forward and out-of-sample testing.
Execution Guardrails
- Live trade automations require stop loss, approval, timeout cancel, execution caps, and broker write access.
- Paper-to-live gate checks paper execution history before live promotion.
- Portfolio Risk Firewall blocks malformed or oversized live trade actions and warns on concentration.
- Paused, disabled, or approval-blocked automations cannot silently continue executing.
Data and Performance Review
Audited trade ideas can be captured in the AI Performance Ledger. User ledger rows are private by default. Public rows are shown only when explicitly marked public.