What's New

1. Advanced Statistical Analysis

We have introduced a dedicated analysis engine that processes market data to extract key statistical signals before the AI even begins its evaluation. This ensures the model is looking at the hard math behind the price action.

• Trend Confidence ($R^2$): The system now calculates the "quality" of a trend using Linear Regression. This allows the AI to distinguish between a clean, tradable trend and choppy, low-probability noise.
• Normalized Slope: Measuring the exact percentage change per interval to determine true trend velocity.
• Z-Score Anomalies: The AI now monitors how many standard deviations the current price is from the mean. This provides strict, mathematical boundaries for identifying overbought or oversold conditions relative to recent volatility.
• Structural Pivots: Automatic detection of fractal highs and lows to identify key support and resistance structures.

2. Sharper Decision Context

These new metrics are now a core part of the AI's decision-making process.

Instead of asking the AI to interpret a raw list of prices, we now provide it with a "Statistical Health Check" for every asset. The model receives explicit confirmation (e.g., "Strong Uptrend Confirmed, Statistical Support at $X") alongside the raw data. This hybrid approach combines the intuition of Full Language Models with the precision of algorithmic quant indicators.

3. Signal Integrity Upgrade

To ensure maximum reliability, we have enforced a strict "Closed-Candle" policy for these new calculations. All statistical features are derived exclusively from finalized market intervals. This eliminates "repainting"—where indicators change mid-candle—ensuring that the AI's real-time decisions are stable, consistent, and fully aligned with historical performance logic.