Ancient Greeks Preferred Small Penises
Now that I have your attention allow me to direct it towards today's data dive. At the request of a true friend and colleague, I have organized a dataset consisting of 9 critical NFL team statistics and will be using them to predict the outcomes of this weeks upcoming games.
I am not going to bore you with the details of how I collected or organized the data. Instead, I will jump right into the evaluation process. The statistics that I used for this analysis are: 3rd Down Conversion %, Rushing Yards/Game, Completion %, Passing Yards/Game, Passing Int./Game, Fumbles/Game, Opponents Yards/Game, Take-aways/Game and Opponents 3rd Down Conversion %.
I used the Logistical Regression function available in excels Data Analysis Toolpak to conduct my analysis. I used the statistics listed above as my independent variables and I used team records as the dependent variable. Here are my results:
R-Squared measures the strength of the relationship between your model and the dependent variable on a 0 – 100% scale. Our R-Squared came out to .968, which at a high-level is a very positive outcome. It means that our data fits our regression model quite well. We can assume this means that the statistics we chose are tightly correlated to a winning outcome.
As your eye's wander south on the results of the regression analysis you will notice a column titled Coefficients. Each of our critical statistics has an associated coefficient value, that indicates its impact on the dependent variable (the outcome). We see that 3rd Down Conversions have a coefficient of .55, meaning that for each 1% increase in 3rd Down Conversion the probability of winning increases by .55%, assuming all other independent variables are held constant.
I used these results to compare matchups this week of games that have tight spreads in the hope of uncovering value that is unnoticed by the rest of my gambling degenerates. Of the 7 games that I analyzed, I found 2 potential opportunities that I will be going all in on this week. See the table below to ride with me.