The math is simple. No machine learning. No hidden coefficients. The same calculations a research desk built into Excel forty years ago, modernized, run nightly, surfaced on a chart.
// MOVING AVERAGESFive windows. Simple, unweighted, trailing.
MA(3), MA(5), MA(7), MA(20), MA(30). Simple unweighted means over the last N games inclusive of the most recent. An MA is undefined until N games exist. MA(5) is the recent picture. MA(20) is the baseline. The crossover between the two is the engine of every shift signal.
// SLOPESOLS regression on the MA series itself.
slope_5 is the ordinary least squares slope of MA(5) computed over its last five points. slope_20 is the OLS slope of MA(20) over its last ten points. Units are runs per game per game. The classifier uses slope thresholds to distinguish noise from real shift. A small positive slope_5 is not enough on its own to call a team Heating Up.
// STANDARD DEVIATIONσ(20), Bessel-corrected.
Sample standard deviation over the last 20 game run totals with divisor n minus one. σ(20) governs the WagerTrend envelope width. Wider σ means a noisier team and a wider envelope.
// WAGERTREND ENVELOPEMA(20) plus or minus two sigma.
Upper band is MA(20) plus 2σ(20). Lower band is MA(20) minus 2σ(20), clipped to zero on display because run totals cannot go negative. The unclipped value is exposed in the envelope status card so you know whether a team is theoretically below floor. Crossings of the envelope band are not signals by themselves; they are inputs into the pattern library.
The envelope is the WagerTrend indicator. The math is two-sigma; the name is proprietary.
// CHANNEL LINESSupport and resistance from local extrema.
Local maxima and minima in the last 20 games (interior points at least as high or low as both neighbors) are identified. OLS regressions are fit separately on maxima (resistance) and minima (support) against game-index position. The lines are projected five game-days forward. Channel lines with R² below 0.15 do not render. R² between 0.15 and 0.50 render with a caveat. R² of 0.50 or higher signals a strong trend.
// HOT AND COLD ZONESAbove p75 and below p25 of the team's season.
The hot zone background tint marks runs above the team's season p75 in the active view. The cold zone marks runs below the season p25. Percentiles use linear interpolation. The zones are descriptive, not predictive. They show where the team has lived in extremes, not where it is going next.
// MOMENTUM OSCILLATORMA(5) minus MA(20), in runs per game.
The small panel beneath the main chart. Positive readings mean recent above baseline. Negative below. The spread magnitude (absolute value) is the variable that splits the four SHIFTING sub-states (marginal, sweet, strong, extreme) you will meet in Module 2.
// CROSSOVER DETECTIONThree filters before a crossover counts.
A crossover is analytically meaningful only when three conditions hold at once. First, the sign of MA(5) minus MA(20) changed within the last three games. Second, the post-crossover spread magnitude exceeds 0.5 runs per game. Third, the absolute slope_5 exceeds 0.15 runs per game per game. If any of the three filters fails, the crossover is noise. The Method does not chase a crossover that fails any one of these filters.
If any filter fails, the crossover is noise.