How XAUMarket AI Operates
Explore the multi-layer pipeline combining real-time TwelveData market feeds, dual independent AI models, and institutional risk management.
Real-Time TwelveData Feed & Technical Indicators
When an analysis is triggered, the system fetches live XAU/USD OHLCV candlestick data across 4 distinct timeframes (M1, M5, M15, and H1) directly from TwelveData. Technical indicators are calculated mathematically without look-ahead bias:
Parallel Dual-AI Isolation
Market data is routed into two isolated AI models running concurrently. Neither model sees the other's assessment, preventing cognitive bias:
• Analyst 1 (Qwen 2.5 Vision): Evaluates rendered M1 to H1 chart images to identify visual price action patterns, support/resistance breakouts, and pinbar rejections.
• Analyst 2 (DeepSeek Chat): Evaluates raw numerical indicator data, momentum shifts, and market structure alignment.
DeepSeek Synthesis & Divergence Audit
The DeepSeek Judge model receives outputs from both analysts along with live price data and multi-timeframe RSI values to perform a final audit:
• Calculates Divergence Score (0-100) to measure agreement.
• Verifies entry range overlap instead of artificially widening disjoint entry zones.
• Checks M1/M5 RSI to block buying into overbought territory or selling into oversold bounds.
Institutional Risk Engine & SL Clamping
Before any signal is presented to the user, the backend RiskEngine recalculates Risk/Reward mathematically. Stop Loss is tightly clamped based on leverage (100x - 2000x) and M5 ATR multiples to protect against liquidation.