Fill it only 86.7%, in fact. So at the beginning of the NHL postseason I sat down and attempted to reason out the somewhat unpredictable outcome of the ensuing playoffs. I am an armchair statistics buff so I used what I had, combined with some good old rational thought, and attempted to devise a method to predict the outcome of the Stanley Cup Finals. Here is how my thought process worked.
In the playoffs, nothing matters but wins. Overtime losses don’t matter. Points don’t matter. So I wanted to find out what statistics correlate significantly with number of wins achieved during the regular season. I used the statistics from every NHL team to expand the sample size. From these calculations I found that 3 statistics significantly positively correlated with winning (goals per game, 5 on 5 goals for and against ratio, power play %) and 1 significantly negatively correlated with winning (goals against per game). However, all of them had at least some correlation, although not above 0.5. So I developed two separate scores based on a summation of the 4 stats, or a summation of all. What I did was take the correlation coefficients for each statistic and multiplied it by the actual statistic achieved by each team. I then added the products and arrived at a sum that I called the All Score, or the 4 Score. I then ranked the teams by descending all score or 4 score. I predicted the outcome of each series by comparing the scores of each competing team. By doing this, both of these scores correctly predicted the outcome of 13 of the 15 Stanley Cup playoff series, an 86.7% success rate. However, the truly amazing outcome was the prediction of the all-score. From the get go, before the playoffs even started, I used the all-score to predict the outcome of the entire tournament. It predicted an ANA v OTT final with ANA being victorious. Holy shit. The only series I predicted incorrectly were the first round DAL v VAN series, and the second round BUF v NYR series. So, we’ll will see if I am as successful next year!