Our model simulates prediction market probabilities as a jump diffusion process blending gradual drifts from market consensus, sudden jumps from breaking news, and time decay toward resolution, then layers in AI-driven priors, liquidity adjustments, and regime-specific risks to pinpoint fair values, edges, and optimal position sizing, turning noisy event contracts into probabilistic alpha.