Let's say we have one contestant who played 1 game and won (1 out of 1), another who played 50 and won 49 of them (49 of 50), and so on. I need to produce a ranking of those contestants even though the count of games is different. The higher number of played games should give higher certainty. The lower should result in lower certainty and therefore when normalizing the scores, how can I mathematically do this such that people with lower certainty get their scored adjusted lower accordingly? In other words, how can I incorporate the certainty of the score to normalize all scores?