It is fascinating that the gambler's ruin problem which is so ubiquitous in modern probability theory (cf. the Levin-Peres text on Markov chain and Mixing Times) actually dates back to a letter from Pascal to Fermat written in the 1650s (cf. this manuscript by Oystein Ore ).
I wanted to see some mathematical research where variants of the classical gambler's ruin were studied (e.g. by accomodating multiple gamblers and/or letting the bet sizes for the rounds to be deterministic or non-deterministic variables, changing the rules for winning/losing a round, etc.).