How PATTERNicus works PATTERNicus is a powerful engine for finding, analyzing and predicting. It uses a dual-metric quantitative similarity engine to scan historical price data. For each analysis, it extracts the most recent price pattern (your chosen window, e.g. 3 months) and normalises it, then slides that template across decades of historical candles, scoring every possible position using two equally-weighted measures: Pearson correlation (which captures the overall shape similarity between two normalised price curves) and directional agreement (which counts what percentage of individual up/down moves match between the two segments). The two scores are averaged into a single 0–100% similarity score, the top non-overlapping matches are ranked, and the proportion that led to a bullish or bearish outcome over your chosen forward window becomes the probability estimate. The more matches found, and the higher their average similarity, the more statistically meaningful the signal — which is what the Signal Quality rating reflects.