A sophisticated logistic regression model that calculates win probabilities based on multiple performance metrics
P(Home Win) = 1 / (1 + e-z)
Where z = β₀ + Σ(βᵢ × Xᵢ)
Probabilities above 70% indicate strong favorites, while 55-60% suggests a slight edge. Values between 45-55% represent essentially even matchups.
P(Home Win) = 1 / (1 + e-(β₀ + ΣβᵢXᵢ))
This logistic function converts the weighted sum of input parameters into a probability between 0 and 1.
Each parameter (Xᵢ) represents the difference between home and away team metrics, except for binary factors like home field advantage.
Given differences: TP = +0.15, SPS = +0.8, HFA = 1, INJ = -0.1
z = 0.5 + (2.0 × 0.15) + (1.2 × 0.8) + (0.24 × 1) + (-0.5 × -0.1)
z = 0.5 + 0.3 + 0.96 + 0.24 + 0.05 = 2.05
P(Home Win) = 1 / (1 + e-2.05) ≈ 88.6%