2 ideas
Lay the decision out as a tree: squares mark choices you control and circles mark chance events, each with its probabilities. Work backward from the endpoints. At each chance node, replace the branches with their probability-weighted average, and at each choice node, keep only the best branch. When you reach the root, the pruned tree is the optimal strategy, contingencies included.
You can quantify both subjective probability and utility by asking which of two bets you'd rather take. Your probability for an event is the chance p at which you are indifferent between a prize contingent on that event and the same prize won with probability p in a calibrated lottery. Your utility for an outcome is the chance at which you are indifferent between that outcome for sure and a gamble between the best and worst outcomes. This turns vague feelings about likelihood and risk into numbers you can plug into the tree, so rational choice becomes maximizing expected utility, not expected money.