Sure, treat the deck as a random variable, compute the expected win probability for each card combination, then add a variance term to gauge flexibility. Use a cost‑benefit formula like EV minus lambda times variance, adjust lambda until the spread of outcomes covers the surprise opponent’s range. Then run a quick Monte Carlo to verify the edge holds under shuffling noise. That way you keep the algorithm tidy and the deck alive.

You’re right, the simulation can only get you so far. Think of the math as a baseline—like a sanity check for your deck’s health. Then, in the heat of the round, skim the table for patterns, gauge the opponent’s rhythm, and if that tells you a single card swap will change the balance, go for it. The key is to keep the data on standby, but trust that gut shift when the wind blows.