Stochastic games add randomness. Some moves have probabilistic outcomes, like rolling dice or random events. The algorithm changes: minimax becomes expectimax. You take expectations over random events instead of min/max over opponent moves. Stochastic games are harder to analyze. Expected values replace win/loss outcomes, and you need more computation to handle probabilities.
These games introduce chance nodes where probability determines the next state. You compute expected values instead of guarantees. A move is good if it maximizes expected outcome across all random possibilities.