Everyday Statistics, with Eddie Davila

- [Instructor] Five minutes left in the game. Your team is ahead. Can this last? Can your team actually hold on and win the championship? At this point in the match, what are the odds your team will win this game? Nowadays, when watching a sporting event, either the telecast or a website may track the live in game win probability, the probability that one team will win the game at that moment in the game. Two equal teams would start the game at 50 50 odds. Then, team A scores first. Their odds might jump to 65%. Right before half time, team A scores again. Now their odds increase to 80%. These win probability tools are developed by statisticians. The goal is to create a tool that is statistically accurate. If the tool says that home teams winning by 10 points at half time win 85% of the time, the 20 games in which that scenario presented itself we would expect that 17 times the home team would win the game, and three times the visiting team would complete a big comeback. But, how are these win probabilities calculated? And, can you trust them? Well, each sport is different, therefore each win probability calculation requires a different algorithm. For example, an NBA team's win probability might take into account the present score, the time remaining, which players are on the court, and how many fouls each player has, and also which team has possession right now. An NFL win probability might also consider things like score, players, possession, and time remaining, but, here, they may consider the down, field position, weather, and the playing surface. A World Cup match might not only have to consider the odds of winning and losing but also the odds the game will end in a tie. Baseball is often a game of specific individual match ups. One batter with certain abilities against a pitcher with specific pitching strengths. The number of balls, strikes, outs, runners on base, and the inning are all considered in calculating the present win probability. And, in every sport, player fatigue and injuries impact probabilities. Also, home teams often have a slight advantage. But, should that increase their odds of winning by 1% or 4%? And, should the odds reflect these particular players in this game in this particular situation? Or, do we just look at how many times in history NBA teams losing by 15 points with four minutes remaining have come back and won the game? And, if we're looking at history, should we consider every single NBA game played or only those played under the current system of rules and current NBA strategies? This is why two different win probability tools may give you different probabilities. Each win probability tool likely utilizes different calculations that use different statistics. So, next time, a win probability chart dooms your team to a 5% win probability, don't lose all hope. Remember, every win probability tool is different. Maybe this model is wrong. Perhaps your team's odds are better than you think. And, if you think all these win probability tools are sub par, perhaps use your statistical knowledge to create your own superior win probability tool.