Signal Scoring

Every detected pattern receives a composite signal score in [0, 1] that combines pattern accuracy, momentum confirmation, and structural context. Scores drive alert thresholds and dashboard ranking.

Score Components

ComponentWeightDescription
Harmonic confidence40%Pattern ratio accuracy from detection (see individual pattern docs)
RSI divergence30%RSI-14 value at the D-point candle
S/R confluence30%D price proximity to the nearest prior support/resistance level
score = 0.4 × confidence + 0.3 × rsiScore + 0.3 × confluenceScore

RSI Scoring

RSI-14 is calculated at the candle corresponding to D.

DirectionFormulaMaximum
Bullish(30 − rsi) / 301.0 when RSI = 0
Bearish(rsi − 70) / 301.0 when RSI = 100

Values below zero are clamped to 0 (RSI not yet in extreme territory).


S/R Confluence

prevSwingPrice is determined by scanning up to 120 candles before X, looking for the nearest contiguous cluster of candles whose wicks break through the X price level.

DirectionCluster extreme used
BullishLowest low of the cluster (support wick)
BearishHighest high of the cluster (resistance wick)

If D is within 0.2% of prevSwingPrice, the full 30% confluence weight is awarded; beyond that the score tapers to 0.


Score Thresholds

ScoreMeaning
≥ 0.7Triggers admin email alert
< 0.7Stored and visible in dashboard, no alert

Scores are also used to rank signals in the analytics dashboard and watchlist.