I have been getting a lot of questions about how exactly my program works so I've decided try to help clear up how exactly it works and how I use it for making my picks. As I've noted in the past, my program is based completely on NFL stats. The main areas I look are passing stats, rushing stats, return stats, and scoring stats. I take these for each team, both offensively and defensively.
I don't just limit myself to stats you find on ESPN though, I take these stats and make some new ones in areas that I think are important. For example, we know Arizona has scored 28.9 points per game, thrown for 7.57 yards per attempt
(This is a bit different than the number you'll see on ESPN since they don't count a sack as a pass attempt, but subtract sack yards from passing yards. If you ask me, that doesn't make much sense), rushed for 3.45 yards per attempt, and has returns of 16.31 yards per attempt. These stats may look important on their own, but what exactly do they mean?
How do these numbers compare to what their opponents allow on average? How efficient are they in turning these yards into points? These are two of the main categories that my program calculates that you won't find on NFL websites, currently Arizona throws for 1.33 yards per attempt better than expected, rushes for -0.44 yards per attempt worse than expected, and scored 1 point for every 13.59 yards of offense (or 1 point for every 21 passing yards and 1 point for every 9.48 rush yards if you want to get more specific).
I understand some of that may be confusing, I apologize if it is, it is very difficult to explain. That is just a brief idea of what my program calculates, it goes over many more things and goes more in depth than that even. Hopefully you have a little bit better of an idea of what it does though. So, when the formulas are all done, it will arrive at certain stats, such as expected pass attempts and expected yards per pass attempt for the upcoming game based on the team's stats and its opponent's stats. The final step it takes it to multiply these two numbers to get predicted passing yards, then, using the scoring efficiency stats, determines how many points the team will get off of these yards. Once passing, rushing, returns, and miscelaneous points (INT Returns, PATS, 2 Point Conversions etc...) are factored in, the final result is a predicted score for the game. I look at the predicted margin of victory or loss, and add the spread, and see what the resulting number is. The higher the number, the greater the value, thus the better the results should be. Currently,
if a team has a value of higher than 5 in my program, and is not a favorite of more than 9 points (unfortunately I didn't notice this until this week), the team is a stunning 12-2 ATS. Of course
past performance does not guarantee future results, but I am going to be paying quite a bit of attention to this until it begins to fail me.
Now, let me just take a moment to say, I understand that you cannot use NFL stats alone to predict outcomes of games. I know there are tons of other variables that affect each game, such as key player injuries, a team's recent turnaround, or other factors such as this. That is where I come in. While my program tells me who to pick based on the numbers, I look at the games and figure out where my program is most likely to be wrong. Also, this week I am ready to give a test run of the program that will factor in recent performance of teams to better factor in injuries that have recently occurred. Take the Chiefs for example, for the first 2/3 of the season they had no passing attack whatsoever. However, now that Thigpen is in they have performed much better than early in the season. So, instead of taking season long stats that still weigh in those weeks without Thigpen, I will be looking at the past 4 weeks when Thigpen has been quarterback to determine their passing stats instead. Anyways, bottom line is I will never recommend a game just because my program said it was a good play, I will always closely examine the matchup and see if anything could throw the program off and look for things the program can't predict, such as weather.
I really do hope this better explains what I do. I'm aware its probably kind of wordy in places but I did struggle with trying to figure out how to explain all of this. I just ask that although we just came off of a bad week, you continue to follow along and trust I will get things turned around. I can assure you I am putting the necessary time in to do this, and I look forward to many good weeks to come.
Stumble It!