NFL Predictions

One of the most challenging things to do in all of sports is to consistently and successfully make NFL predictions against a known line or spread. This is a far different challenge than simply picking the team that will win the game, but even that is hard to do and the best in the business usually can never consistently get 70% of those picks correct. Contrast that with the 50% success that most everyone on the planet gets when they try to pick the team that will successfully cover the spread!

What I find amusing is all the football websites that make predictions on all the NFL games in the upcoming season 6 months before the games are even played! That is totally ludicrous, insane, and impossible … pick your own adjective to describe that! It is hard enough to even pick games during the actual week of games when the teams are known, the injury reports are out, and there is a track record for data. But 6 months before the season begins? Yes, it is pure entertainment, and very poor entertainment at that!

The line makers in Las Vegas and offshore are not idiots. And while they make a line early in the week prior to the Sunday games, that line can shift and change quite drastically as information is accumulated prior to the games. Anyone who makes NFL predictions needs to know why these line makers are so good at what they do. Their financial survival depends on it. They are really not interested in who wins the game, but they are very interested in getting as close to possible an equal amount of money on each team before kickoff, thus ensuring that whoever wins, they will win financially! This line is a living and changing status that is expertly calculated in order to return the highest amount of profit to the casino or sports book. Think about it, each person who picks a game is paying approximately 10% at the window, so the house gets their 20% and pays back 10% plus the original amount back to the winner. If roughly the same amount of money is on each team, the house wins and wins big.

The risk to the house is if there is a gross inequity in teams picked and the wrong team covers. For example, if 95% of the money were to end up on Dallas and only 5% on the Giants, and Dallas covers, then the house has to pay back most of the money and they end of taking a huge bath. But if it is close to 50/50, the house always wins. The line is adjusted to encourage more people to pick the team that people are not betting on. So, in our Dallas/Giants example, let’s say that Dallas is favored by 4 initially and most of the wagers come in on Dallas. The public, in this case, believes that Dallas will easily cover the 4 by winning by at least 5 points. The house smartly changes the line to Dallas -5.5, for example, and now these Dallas bettors are not so sure, but many more Giants fans jump in knowing that they only lose is if their team loses by 6 or more points, so now the money goes from 95% Dallas, 5% Giants, to 55% Dallas, 45% Giants, and this process continues right up until kickoff where is might end up 52/48 or 51/49. The house wants equal money on each side.

While the line is indeed a function somewhat of the seasoned expertise of the Vegas linemakers, it is also adjusted due to public opinion expressed as dollars placed on each team. I did a study once to see how accurate the linemakers really were in predicting the outcome, and it was quite interesting. My data showed that the sports books get as close as 12 to 13 points from the final score on average. For example, let’s say that Dallas is favored to win the game by 7 (expressed as Dallas -7) against the Giants. In this instance, the most accurate linemaker in the world would say Dallas -7, and Dallas would win 24-17, but that rarely happens. Much more common would be for Dallas to win by 19 or to lose by 5 (off by 12 points in either direction). Final scores that would be closer to an average accuracy of the Vegas linemakers might, therefore, be more like Dallas 34, NY Giants 14, or NY Giants 24, Dallas 17. Those two outcomes (Dallas winning by 20 or losing by 7) would fit right about in the range of how accurate Las Vegas usually is over many games. While it does not seem impressive on its surface, against the spread it is very good! The experts get within 12 or 13 points on average, while fans, media and inexperienced observers are usually close to 16 to 18 points off the final score.

Now that we’ve established that Vegas is extremely good at what they do, but also vulnerable since they are really quite far off from the final score (but very good at getting equal money on each team so that they profit) there are indeed ways to exploit the natural inaccuracies that exist week to week in an NFL season. As a licensed clinical and sports psychologist who has worked with NFL teams and players, studied football for years, and wrote a book about a totally new scoring system that included mental performance and was able to capture football performance better than any single statistic yet invented (I call it the Mental Performance Index or “MPI” for short), I have a lot to share on this issue.

I am privy to something that Las Vegas never even considered, or if they have they did not spend 20 years refining a new statistic that correlates with winning the Super Bowl about 85% statistically (on a Pearson Product Moment Correlation) and discussed in my book “The Mental Performance Index” supported by many icons in the NFL including Don Shula, Tom Flores, Steve Sabol and Lesley Visser. The next best statistic, turnover differential, is around 45 to 50% at best. By combining my knowledge of mental performance, statistics, and football, I was able to come up with something that is truly exciting as we enter the 2019 NFL season.

One thing you should consider about psychologists is that we are very interested in and knowledgeable about statistics and the accuracy of measurements. It is the hallmark of good science. My background in both sports psychology and clinical psychology scientifically means that I sat through at least 8 very strong courses on statistics and research methods, and I also taught statistics at both the undergraduate and graduate school levels. I did not stop there. In the formulation of my algorithm, I hired several statisticians along the way to directly help with the project and guide my overall mission. The end result is that I did it the right way. Starting in 2018 I began sharing my findings with the world.

The book, revised in 2013, describes my MPI well, and since then I have worked tirelessly to convert this statistic into something that could be used to forecast or predict which team has the best chance of covering the elusive spread. Remember, this is made very difficult by both oddsmakers who are professionals at this, and public money, which has its own intelligence attached to it but is also grossly misled by ignorant sports talk show hosts who pretend to know which team will cover. From this view, I love the so-called experts on tv! They make my job easier with their misinformation.

Using everything we know about a team before kickoff, I built a model that considered these known factors as well as a variety of other statistics that ensued from looking at 40 years of past NFL games. While I would never give away the secret recipe, I will offer a few clues to whet the appetite. I rely on my patented MPI data, but I also rely on more traditional factors such as how each team did against common opponents that season, by what score, and in what setting, and also the stability or variability of each team playing that week in terms of performance. Teams that are more consistent on my MPI are more reliable overall and much easier to predict.

I will use an analogy to explain part of what I do, but this is just a small part. About 15 other factors go into it, but this is a way to share the excitement. Let’s say that you go sport hunting with paint guns, and you would like to capture a rabbit that is running erratically all over the place with quick jigs and jags making it hard to hit with your paint gun (I’m not a big advocate of real hunting so I am used a paint gun analogy!). That rabbit would be very difficult to shoot, and very hard to know where he would be at any moment! On the other hand, let’s say you go to a farm with a number of cows that move very slowly and steadily. Of course, those targets would be much easier to hit with the paint gun. Rabbits are the NFL teams that perform erratically on my MPI and in terms of overall performance. If the two teams playing in the upcoming game are like rabbits, it makes it even harder to predict that game. On the other hand, if both teams have been consistent in mental and physical performance, like cows on a farm or pasture, that game is much easier to predict!

In making my picks each week, I love it when I find cows playing cows, and rarely make picks on games when rabbits play rabbits! This is the value of knowing what we call variance or standard deviation in statistics, and it is but one of a variety of important aspects of performance that I analyze in making my picks each week. Some weeks there are as many as 5 to 8 games that are easier to predict and within my acceptable range for a pick, while on other weeks there may be no good games or only 1 that week that meets my criteria. If I were to just blindly pick all 16 games each week, my success rate would be much lower than what I get when I am able to cherry pick and only make picks on teams that fall within a certain range of success. Forget about just picking the favorite to cover against the spread. My research of many thousands of games shows that you will actually do better picking the underdog! But it is just slightly above 50% success, so forget about doing anything valuable there.

The work that I put into formulating this method of prediction was exhausting, and it really took 5 years as a passionate side project. I find the notion of predicting the future really audacious and fun. I mean, who would not want to be the Nostradamus of football if they loved football? I was already being called the “Freud of Football” by the Washington Post, so I pursued this tiger over 5 years with all my energy and knowledge. I was seeking a variety of predictable factors, and the combinations of them that would lead to the optimal success against the spread. I am convinced that I hit a home run!

Most of the world does not know about this yet, and that is somewhat by design because I have only been making public predictions in advance for one season, beginning in 2018. My comprehensive research indicated that my NFL predictions should be able to get anywhere from 55% to 57% success against the spread consistently over an infinite number of games. Even it only ends up at 54% or 53.5% long-term, I would still have built the best prediction machine in NFL football history!

If you want to test this, just find any handicapper who has had his results of picking a side against a line documented accurately over 1000 to 2000 games and you will inevitably discover, as I did, that success long-term is rarely more than 51%! As I have written about in other articles, one of the best icons in football picks against a spread gets just about 52.8% over about 1000 games. Another handicapper, the genius behind Jimmy “The Greek” Snyder in the 1970s, was shown to have a long-term success rate around 49.8% over 1000 picks when his record was studied in detail. When I saw that, my own confidence in picks soared!

The problem is that science is never glamorous and scientific success is often not realized until long after discovery. So most do not yet know about the MPI because to prove this will take careful, slow and accurate analysis. I discovered that the mental game in football was being largely ignored. I did something about it and invented a statistic. I studied that statistic and found it was the key to winning the Super Bowl. I wrote a book about my findings and was endorsed by NFL leaders and stars. I next converted this statistic through many years of research into a prediction machine. Yet, like any good science, I do not declare victory until this has stood the test of time. Another way to say this is that I need more games, or a higher N in statistics parlance, to show the value of my system.

Results so far? In 2018, I established a record against the spread of 54 wins, 41 losses, and 3 pushes. This equates to a 57% success rate. While this seems very impressive (and I know it is valid) I cannot come out and say this is a winner until I have a couple more seasons at least in the tank. I encourage everyone to join me on this mission, benefit from my picks if you want, and see what I am doing over at https://www.FootballShink.com. I will continue to make my picks in advance of games against a known line, and publish all the details for legitimacy before the game, and then publish the results after the game and the ongoing record.

In sum, this is an exciting work in progress. It will ultimately show the world how truly key the mental game is to success in football, and all sports too. It will also show people what I believe will be the best prediction algorithm every created for NFL football.