In the process of determining the rankings I put out, I calculate a handful of interesting stats for each team. Starting in ~2015, I started figuring out a way to use those stats to attempt to project game outcomes a week at a time. I don't really have a specific goal for this other than my own entertainment, but if I felt so inclined I could probably use this data to make some money gambling.
I will be putting out a handful of projections each week, usually just for relevant and/or interesting matchups. I'll gladly give out more or specific ones if somebody asks for them.
What the projections are:
Using the data that goes into the rankings I run a couple different regression approaches to project upcoming games. Three different regressions have all shown to have nearly equal aggregate performance. In some games or weeks one will do better, sometimes another does better.
What I'm publishing:
What you'll find in the projection posts is the handful of matchups I've chosen to publish that week along with two numbers. The first number is the projected spread based on the mean of the three regression methods. The second number is the total range of projected spreads across the three methods.
EXAMPLE: For some matchup if you see: 10.0 and 2.0, this means that the three methods project a margin of 10 points on average, but one of the methods probably projected 11 and one probably projected 9, thus the range of 2.0.
How to interpret the table:
The final thing to know is how to read the table. The projected spread should be read as the score of the Away Team minus the Home Team. So a positive number means the road team will win, while a negative number means the home team will win by the projected margin. In this way, the negative numbers here are comparable to the same numbers as assigned to home teams by oddsmakers in Vegas or elsewhere.
EXAMPLE: If Utah is at home against say USC and Vegas gives Utah a -7, this is the same as my tables listing: USC, Utah, -7, range.
Or if Utah is home against USC and Vegas give USC the -7, this would be: USC, Utah, 7, range.
Finally, I do not encourage gambling, I'm not telling you if or how you should gamble, and I accept no liability or blame for any decisions you make with this data. If you feel inclined to gamble and want to use my numbers in some way for that, suit yourself, but you should probably use my data in conjunction with many other sources and your own intuition. Although my projections have done relatively well against oddsmakers en masse in past seasons, there is absolutely zero reason to believe that my projections will be better than Vegas on any game, week, or season.
Comments
Post a Comment