P(rain) = (members with rain + ½) / (members + 1)
σ = √(σ_poll² + drift² · days/30)
P(A wins) = Φ( (A − B) / σ )
posterior odds = prior odds × LR, odds = p / (1 − p)
p* = σ( a · Σ wᵢ·logit(pᵢ) / Σ wᵢ ), logit(p) = ln(p / (1−p))
p ~ Beta(k+1, n−k+1)
P(tomorrow) = (k+1)/(n+2)
P(≥1 in m days) = 1 − B(k+1, n−k+1+m) / B(k+1, n−k+1)
λ ~ Gamma(c+½, t)
P(≥1 in w) = 1 − (t/(t+w))^(c+½)
P(E | sign) = P(sign|E)·P(E) / [P(sign|E)·P(E) + P(sign|¬E)·P(¬E)]
P(all) = Π pᵢ P(at least one) = 1 − Π (1 − pᵢ)
Brier = mean (p − outcome)² (0 = perfect, 0.25 = always 50%)
Log score = mean −ln P(what happened) (0.693 = always 50%)