Football Prediction

As a clinical and sports psychologist with a passion for all that is NFL Football, I have written elsewhere about how I discovered a fascinating new finding in the science of sports. I discovered that the mental performance that I have been working on improving with my clients for years was actually much more valuable and influential than even I realized.

It shocked me when I discovered that nobody in history (and that is a pretty bold statement) had taken the time to quantify how smart a team had played, or how well they had performed mentally. Sure, the talk radio hosts, television gurus, coaches, and writers had all spoken about the need to be mentally tough and to make smart decisions, but it was a thrill to me when I learned that despite this vitally important part of any sport, nobody had decided, as I did in the early 2000s, to quantify this all-important factor. I captured an NFL tiger by its tail!

I made my discovery after creating this new statistic that in one simple number described how well a football team had performed, but what was unique is that a big part of that performance included in the statistic was how well they had done mentally in managing pressure and reducing careless mistakes and penalties, as just a couple examples. My journey toward making NFL football picks was, thus, a quite unexpected one that began in simply trying to help NFL and college coaches understand better how their teams were performing.

I was paid good money to create reports on team performance after reviewing the coach tapes of offense, defense, and special teams. These are the ones used by teams without the hype of media and commercialism that you mostly see out there. I was paid to review every play, create the performance statistic for each team, and then communicate a report back to the coach, often warning him about what was going on with his team in a unique way that highlighted the mental performance as well as the overall execution physically.

The reason my reports worked is that final scores and more traditional statistics are often grossly misleading. Most teams have a quality control department that somewhat overlaps with what I was doing, but my product was unique enough to add significant value. Sometimes coaches were advised to be on guard against a letdown, as when team performance was shoddy despite a good victory on the scoreboard. At other times, coaches were encouraged to be positive and uplifting, as when the team lost despite outperforming their opponent on my statistic in a big way.

The entire premise of my approach was and still is that there is a vast difference between how well an athlete or team performs, and the final score or outcome. In many ways, my approach was one of the most accurate depictions of actual performance possible as it included all 150-160 plays in the game in the scoring. It was both comprehensive and also a great measure of team consistency. It was a measure of mental performance too, and I believe that is what made is stand apart. It is also what gives my NFL football picks today a special advantage.

As I have stated elsewhere, picking a football team against a spread, or saying which team will cover the spread given a published line, is one of the hardest things to do in sports. Many do not realize how hard this is. To give you an example of how truly challenging it is, I looked at the top 36 football handicappers alive today and studied their long-term records that often go back 15 years. Beating the spread proved for most of these handicappers to be extremely difficult and many of them were below 50% success. It soon became clear to me that those who could achieve 52 or 53% success in picking against a spread were actually like rock stars in the industry. Those who had no experience in this, or faked being knowledgeable due to just having a radio or television show, were getting close to 50% success, which is rock bottom. If most were to theoretically pick a million games, they would not get more or less than 50%. It is the pits when randomness rules.

It is just as hard to get 45% success as it is to get 55% success. In a binary decision of picking one team or the other against a known line (like Dallas -7, which means that Dallas is favored by 7 points and if you pick Dallas you need Dallas to win by 8 or more to win the bet) if you are the most terrible handicapper in history your long-term record would be exactly 50%, showing that you are no better than a random coin flip in your picks. A dog, monkey or fish could be taught to pick one team or the other and would ultimately get 50% success if they were able to pick 100,000 games. It would actually be more like 49.98% or 50.02%, but you get the point.

The handicappers who are really good are able to make these binary decisions just a tad better than everyone else on the planet. Not to mention names, but I studied one handicapper who has been very prominent in publishing magazines of NFL prognostication every year for many years, and he is regarded as one of the best by far in the business. Guess what his record is over about 1000 picks? It is 52.8%! And that is great. It shows that whatever he is doing allows him to get almost 3% better than chance. More power to him, but in my view that is not enough!

The general population should know that if a person making football picks for money were to achieve on average 52.38% over a long period of time, they would break even financially. That is the magic percentage that is needed to just be able to get your money back after paying for the “vig” or the cost of the bet. The vig can vary from house to house, and game to game, but 10% is a usual standard. Of course, there are some picks that can be had for a 5% vig, and there are some over 10%, but 10% seems to be the most common vig, so that 52.38% level to break even is well documented and based on the 10% level.

As such, this famous handicapper that I alluded to with his 52.8% success over 1000 picks is barely making any money at all! There are a very select few handicappers who have been able to achieve 55 or 56% success, but from what I have studied, these are usually based on a smaller sample of games, in the 50 to 150 game range. Nobody I have seen who has picked over 1500 games can get above 53%. It is almost an impossible feat … but I think I can do it!

A good basic working knowledge of statistics is necessary in order to really understand this all. The greater the sample size in anything studied, the closer the results of that sample approximate the true population of whatever you are looking at. This applies to anything from political polls to the color of automobiles on the road. If I study only 10 games and get 10% success or 90% success, it is totally meaningless. If I look at 100 games, it starts to look more like reality. But even a 60% or 40% success over 100 games might be due to chance more than anything else. By 300 games, and definitely by 1000 games picked, the sample success rate starts to really approximate the population reality.

In my research to convert my coveted MPI statistic into an NFL football picks tool, I looked at almost 10,000 NFL football games over 40 years of past games from the 1970s to the current day. As such, my analysis had a huge sample size, and one of my statisticians told me one day, “you have essentially studied an entire population of NFL games” and he was exactly right. I essentially studied every game in the modern era of NFL football. I did not go back prior to the 1970s, however, as I felt that the game was so much different back then that I would perhaps have more data but I would be studying a different sport. If you watch tapes, all they did was run, run, run with much less creativity than the game played today.

I am comfortable knowing that I devoted the time to study almost 10,000 past games with my programs and analyses. I did some fancy tweaks, with the help of my professional statisticians, to avoid some of the common errors that data analyzers often make when looking at historical or archival data. For example, it is very important, in statistical language, that the estimates made are what are called “unbiased estimates” and going into all those details are beyond the scope of this article. Just know that I took every possible precaution to ensure that I was not making some of the backtracking mistakes, errors of bias, or other pitfalls that amateurs routinely make.

After looking at all these past games with my MPI statistic, I was then able to juggle an impressive number of variables such as the published line for that game, past games that season, the final score, who I would have picked based on where the game was played, who was favored and by how much, who the teams were, and so many other mental factors. In the end, I got a very impressive set of standards using always one question and one question only, how well would my pick have done against that known spread on that day?

This was at minimum a 5-year journey, working every spare hour I had after working with clients that day, and it involved countless analyses, javascript programs, endless spread sheets, and more than I can possibly say. It was a labor of passion to try to build as close to possible the perfect prediction machine against the spread, and I am excited to say that I think I have done so! With the data known before games, and by running two particular programs, I now make a select few picks on any given NFL week in the 17- week season, and I am confident that I have defeated that nasty 50/50 success rate that even a 2-year old child would get. More than that, I am confident that I can get closer to 55% long-term!

What is known so far? In my first full year of making predictions each week in advance at I have a record of 54 wins, 41 losses, and 3 pushes (a “push” is the word used for a tie in handicapping parlance). That is exactly 56.84% success, which rounds nicely to 57%. Before I get too excited, however, I am reminded by my countless statistics professors as a graduate student in sports psychology and psychology, that a sample size of only 95 games is way too small to get overly thrilled. I just need to keep plodding along, making my picks, and growing that sample size. By the end of this coming NFL season, I will have about 200 games in the bucket. I would have to have a terrible year of around 43% to be at 50% success after 200 games. If I even have a modest success this coming season of 51% my overall average will be around 53 or 54% which is better than the best in the industry, so you can see why I am so excited by my start.

What does it mean if I am eventually shown to be really good at making NFL football picks? For one, and this is the reason I started this all in the first place, it will be a strong endorsement for the value of mental performance in football, and really all sports. Since what I am doing in making predictions is relying heavily on a mental component, the success I am having will ultimately be shown to be due to the fact that I am focusing on some aspect of sports that has been largely ignored! The first take-home message, then, will be that sports psychology as a profession was right from the very beginning about how important mental skills are, and mental training will grow in popularity if coaches want to win. This would prove that empirically. Argue with science at your own peril.

I guess a second benefit of having success picking NFL games against a spread will be that people will want to have access to my picks. They can have them now for free, by just going to the website, but they might not believe in them yet. I believe in them. I did the research over 5 years and nearly 10,000 games, and I was not messing around. This is serious. Yet at some point, if my picks are proven to work over several hundreds of games, I might creatively find some way to profit from this venture. After all, nothing is free forever. I am not sure what form my entreprenuership of this might take, but you can be sure that a big part of my mission will be to continue to extoll the benefits and value of focusing on mental performance in order to help people by getting that message out. If I can also profit by franchising out the program, setting up a football picks service, or building a national radio or tv forum for my picks, that would be fun. It all will depend on one thing, and one thing only, how good my picks are, and how important the mental game is to success.

I encourage each and every one of you to start tuning into the fun at where all my NFL football picks will continue to be posted throughout the 2019 season and beyond!

Winning Cappers

recent results

Winning Cappers