Balatro Challenge 1: from a hand of 8 without any pairs discard n (max 5) and draw n from the remaining 44 cards. Which n maximizes your chance of obtaining at least a pair?
I coded it in Julia. 0.077s/million simulation per n.
Beat it with your lang!
An idea came to me about GBM algos like XGBoost: What if I add a few randomly generated features to a tabular dataframe and call XGBoost on it. I stop the growth of a node if a random feature is selected and use that as my stop-growth criterion. Is this a known approach?
