How Really Good are the Experts in Predicting NFL Football Games? What Does it Mean to be an Expert at Making NFL Picks? When I began developing a totally new way to look at football performance, I demanded that my approach included the smartest sports psychology principles available. I was helping an NFL team and some of their players individually, and one day I had an incredible “aha moment” when I realized something vital. I was asking my players and coaches to emphasize process over performance, but I had no way of measuring how well they complied with my mental coaching other than vague self-report of coach perspectives.
One of the oldest and wisest principles of sports psychology is to get the performers focusing on what they are specifically doing, the tasks they are engaged in and how well they do those tasks rather than obsessing or worrying about outcome or results. There are two problems with worrying too much about results. For one, it is totally in the future, and as such, the athlete or team living in the future will not be focused on the present. That present centered focus is so key to success, so in thinking too much about results, successes or failures, the athlete guarantees defeat by losing focus. The second problem with future oriented thinking about outcome is that it create fears, anxieties, and unnecessary stress. In many of my articles and lectures you will hear me saying “if you think one time about results during a competition you have already lost.” The truth is that you have already lost the battle of optimal focus and you have also increased the odds of unhealthy worry.
In the early years of developing my Mental Performance Index, I was aggressively seeking every which way to measure what I was telling my athletes in all sports to do. I was finding objective and observable behaviors that indicated that a team was in the moment, not playing with excess anxiety or stress, staying completely focused on the present, and avoiding carelessness often seen in turnovers, penalties, clock abuse or needless performance failures that everyone could agree indicted horrible focus. I gathered data that reflected either optimal or poor mental performance and another phrase I used for this was “smart play,” suggesting that it is smart to stay focused and just execute in the moment, and not smart to think too far ahead, get anxious or make needless mistakes.
In many ways, my MPI was a measure of team consistency in football. I had my hunches that this approach of measuring the smart or not smart moment was going to reveal something very important about overall performance and the statistics confirmed this. I hired several statisticians and statistics teams after collecting all the MPI data on every team on every play in Super Bowl History. The statisticians then went to work analyzing how vital my MPI was to winning the game on Super Bowl Sunday, and it turned out that it was by far the most important statistic to winning.
My excitement about the MPI was derived from the fact that I had discovered (or actually created) a new way of looking at overall team performance in football, and this look was encapsulated in a statistic that could be easily generated and depict how well a team had performed. I had no idea that I would hit the ball out of the park. My statisticians were impressed that the correlations between the MPI and winning were so high (in the .85 range) compared to the next best more traditional statistic (turnover differential) which did not quite reach .50! In other words, my clinical instincts in 2001 to measure the moment and include an analysis of mental performance were supported and rewarded by actual empirical data showing that this was the key to winning.
I was surprised that nobody had tried to measure mental performance in football, and I felt a passion to get this new finding out in a big way, so I wrote a book about it in 2011 and revised this book in 2013. The title was “The Mental Performance Index: Ranking the Best Teams in Super Bowl History.” It was almost as if the mental skills had gained a new exciting life and were to forever be viewed as not only important to success, but quite likely to be the most important key to success. Had I been delusional, I do not think I would have received cover endorsements from the likes of Don Shula, the winningest football coach in history, Steve Sabol of NFL Films, and many others. I do not think I would have had my foreward writer Tom Flores (yes, the Tom Flores with 4 Super Bowl rings) get behind the project and rave about it. And I doubt that Lesley Visser, the first and only female inducted into the pro football hall of fame, would have been excited enough to write the epilogue on NFL coaching legend Bill Walsh. The bottom line is that this book on the MPI is a terrific addition to our understanding of football and really all sports.
There was only one small problem. After I revised the book in 2013, I realized that the sports world, and football coaches in particular, were either not buying my revolutionary new understanding, or more probably had just not read the book. American Football is one of the most traditional and conservative sports, seldom willing to go into new areas, and my publisher was not Random House! The word was not getting out as fast as I would have liked, so I changed my strategy. Since I had such valuable data in this new statistic, and many including friends, pro athletes, and coaches encouraged me to take it to a new level of prediction, I decided that I would convert it to a predictor and figure out if I could make accurate NFL expert picks based on the information I was receiving about teams. If I could show that I could predict games against the spread better than anyone else out there with this tool, people in all sports would eventually want to know how I was doing it and the mental performance I had shown to be so critically important would get its day in the sun.
This was not an easy task to convert the MPI rating system to something that would be good at making NFL expert picks against a known line. It took about 5 years working a few hours a day on the project between 2013 and 2018. The reason it took so long is that I needed to examine hundreds of possible combinations of variables over 40 years of past NFL games, and do it right. I needed to find a way to create an adapted version of the MPI that could be applied to past games. I was up to the challenge. Nothing that is valuable ever comes easy in life, yet I had a quiet confidence that with the MPI it was going to be hard to fail. I think I was spot on!
Using a combination of solid sports psychology principles, careful and proper statistical analyses, football knowledge, and a massive database, I studied what it takes to beat the Las Vegas or offshore spread in NFL games. All along, my only key question was “how does doing this lead to success in getting it right against the spread?” I have written elsewhere about what it means to beat the spread, so I will not address this in the current article, but suffice it to say that the entire planet of those making football picks average about 50% success, which is no better than a coin flip because the linemakers, and public money, are smart and they have essentially handicapped the game with a line.
What is funny is that I had original visions of hitting 70% or even 65% success against the spread, but those early notions proved foolish, and I am now far wiser. 50% represents chance, or the worst performance possible. A newborn baby or a cat with his paw making NFL picks against a spread will get closer and closer to 50% success over time based on pure chance. It really is a coin flip. The spread more or less evens the teams to get equal money for the house on each side. To win money at the betting window when a person lays $110 to make a $100 bet, the person over time would need to average 52.38% over many picks just to break even! That 2.38% over chance is what it costs to make the bet, and that fee is called the vigorish or vig for short.
What was the standard among experts? Knowing about NFL expert picks it seemed would be a good place to start, so I decided to study how the best handicappers in the world were doing. I tuned into one of the best websites in the world for this which has a panel of 37 NFL experts whose picks are published and archived, some went all the way back to 2003. I studied all the handicappers and their success rates in picking sides against a spread over several different years, and never once found an error or an inflation in success rates. The handicappers on that site are well known nationally, publish magazines and articles in pre-season guides, and have often been doing it a long time.
I completed a study of all 37 handicappers on this site, and their NFL expert picks since 2003, and I would like to share some fascinating results when it comes to making picks against a known official line. The take home message is that the overall average is below what it would take to make money!
Of all the 37 NFL handicappers combined, there were a total of 15,228 wins, 13,976 losses, 873 pushes. This equates to 52.14% against the spread (-323 on average in terms of money).
In terms of just the top 10 NFL handicappers based on their total wins, the results were no better with a total of 7,562 wins, 6,914 losses, 412 pushes. This equates to 52.24% against the spread (-316 on average)
For the 27 NFL handicappers who have had data of making picks on at least 300 games, we again saw similar results of 14,341 wins, 13,143 losses, 804 pushes or 52.18% against the spread
What does all this say? Is says that the absolute best handicappers in the world are averaging 52% success against the spread in NFL football. It does not matter if you look at all 37 handicappers, the top 10 handicappers, or the 27 with over 300 games, they all get very close to the same results of 52% and it is below what it would take to even break even at 52.38% if the bets were all -110. If the bets, however, are -105 (in other words, you pay $110 or $105 dollars to win $100) the percentage needed to break even is a little more than 51%, but since the format for many is -110, I am using that for the sake of example.
What is impressive about my little study above is the sheer numbers of observations, and it is interesting that no matter how you slice it, the success rates tend to hover very close to 52% at best.
Now contrast that with my research using the MPI over 40 years of past NFL games. I was able to look at many thousands of games over 40 years and get a success rate between 56 and 57% overall against the spread. While that was only archival research, and not a look forward into the future, with the help of several key statisticians I was able to make sure that I did not violate statistical principles and that I only got what is called an “unbiased estimate.” Another way to say that is that while I was predicting games in the past using the MPI algorithm, it was as if I was predicting games in the future.
I was still far from satisfied. My data showed that predictions using my new approach were yielding great success, but the only way to show this to anyone convincingly and to convince myself 100% too, was to make totally new predictions on games that had not yet been played. That is precisely what I did in the inaugural year of this at FootballShrink.com in 2018.
Let’s see how I am doing after one year and some Super Bowl picks made in advance: 54 Wins, 41 Losses, 3 Pushes, and this all equates to 57% against the spread. I realize that I have only made 98 future predictions so far, and that this is still a very small number statistically. But given that I have looked at almost 10,000 games over 40 years in developing this prediction monster, can you blame me for being excited that as a person now making NFL expert picks, I very well might have developed the best tool of prediction ever? I am modest until I see a couple more hundred games, but my enthusiasm is already hard to quell.
If I am correct, and this new algorithm is able to even get 53 or 54% over a long period of time, I will have cracked the code in predicting games against a known line. When this happens, I am also convinced that rather then me lining up to try to show people my book, and all the exciting findings we made in studying mental performance, people will be lining up to try to use this algorithm, see my picks, and learn more about what has been a truly fascinating pursuit.
Stay tuned as the 2019 NFL predictions are coming soon on a weekly basis in year two of this test at FootballShrink.com.