Blending the Worlds of Sports & Mathematics

I have been asked what exactly it is to bring together or blend math and sports.  First, I have been involved in sports ever since I was a small child, baseball, football, basketball, I loved them all and played most sports up until I was about 40.  So when I was given the chance to work for a company that dealt with sports, it seemed like a dream job.  In fact, most of my friends back in Los Angeles couldn’t believe I got paid to do a job where part of my job was watching games on TV every day.  And in fact, I have had a TV in my office for as long as I have worked at National Sports (since 1986) so that I could watch ESPN, ESPN News and of course football, baseball and basketball games.

I have always been fascinated with the prospect of predicting the outcome of games based on past results.  And in fact, predicting the outcome based on some algorithm or formulas.  Over the last 20 years I have experimented with numerous algorithms, most of which had varying degrees of success.  Sometimes they did well, other times seemed totally crazy.  There isn’t any shortage of programs on the Internet that claim to predict the outcome of football games based on some sort of prediction model or mathematical algorithms. However, baseball is a different animal.  Most companies don’t like dealing with baseball and you won’t find many out there producing software to predict baseball outcomes.  This is mainly because baseball is a daily sport, but more so because of pitchers.  Baseball lines are largely based on the pitching matchups for the contest and that makes programming the software even more difficult.  Now you have to keep track of every pitcher in baseball, including relievers and be able to allow the customer to update their database daily with those stats.  This makes for a labor intensive program that prohibits the cost.  Typically, people will pay up to about $99 for a program they think will produce results with the games.  Anything over $99 and it becomes more difficult to make sales.  Add to that fact that you need to give them free access to download the stats and lines every day for the program.

Back in 1996 our company decided to produce a statistical program for college and pro football called “Smart Line.”  This program allowed the user to choose their own stats for the teams and then basically handicap their own games based on the stats produced by these stats.  We failed miserably at this venture.  Why did we fail?  There were a few reasons. 1) the Internet wasn’t what it is today and back then people needed what was called a “shell” account to access the internet and then our servers so they could download the weekly stats.  This did not work well at all since it wasn’t reliable and the few customers we did have needed complicated instructions to download stats.  2) We charged a seasonal price to download those stats.  They should have been given this free as part of purchasing the software.  3) The software was priced anywhere from $299 to $599, depending on what they could get the customer to purchase it for. (remember my note on the $99 ceiling).  4) The marketing was horrible and never made it to the retail stores.  It was only sold via our company by salesmen.   5) The program did not give a prediction on who should win the game.  This one was our on biggest customer complaint (among many), they wanted the program to tell them who would win the game.  The software itself was a fine product and gave lots of wonderful statistics, but everything else about it failed and so did the product.

Today, the best of these programs work if they sell for around $59, have free, one-click downloads of stats and give selections on the games.

Which brings us to the question at hand, can one predict the outcome of games by using mathematics.  The answer is yes and no.

First you have to understand how sports betting works.  In football and basketball the sportsbook puts up a line on the game (baseball and hockey includes what are called money lines and I won’t go into that here).  Now some people have the misconception that this line is a prediction of how much one team should beat the other team by.   For example, the Denver Broncos are 6-point favorites this Sunday against the Kansas City Chiefs.  The 6-point line is a number that a group of people determined through various methods will generate equal betting on both Denver and Kansas City. This is important, the Line is a number that the books believe will generate equal betting on both the favorite and underdog. Ideally, the sportsbook doesn’t care who wins as long as they have equal betting on both sides.  That’s why the bettor is charged a vigorish or “juice” on the game.  This is usually about 10%.  If you want to win $100 on the Broncos, you will have to bet $110, with the $10 being the “vig.”  It’s that $10 that the sportsbook is looking to make as their profit margin on the game.  If they take  100-thousand dollars on that one game, they have made 10-thousand profit if they get equal betting.   If the betting on the game is one sided, the sports will then move the line towards the team being bet on so that they will generate interest on the other side.  If betting is coming in on Denver in our example, the line may go to Denver minus 7-points in an effort to get people betting on Kansas City.  The last thing a book wants to do is start gambling on the right side winning and thus risking taking a loss.  Though it does happen and quite often that a book will be on the short end of games, weekends and the rare season.

Because of this “vig” the player needs to hit just over 52% of their games if they hope to break even.   If a sports bettor could consistently hit 55% on a season he/she would make a living betting sports.  However, statistics show that the longer anyone bets, the closer to 50% overall their picks will be.  So again, you can see that approximately 2% difference is the profit the books are looking for.   So how can we increase our odds at sports betting and raise that 50% winning to say 54, 55 or even higher percentages?  One way is through a combination of math and analysis. There is NO WAY you can rely on the mathematical outcome of a program or algorithm to  predict a game without any human analysis. There are just too many variables the computer cannot take into account to predict the outcome of a game based on past performances, including weather, coaching, injuries and situations to name a few.  A good example is late in a pro football season when players might be rested by teams that have locked-up spots. There really is no way to program those personnel changes into a system. This is where you have to use your own analysis on the games.  Same thing with key injuries (the QB is out), or weather (blowing snowstorms) and even coaching plays a part.  And lets not forget one big variable that cannot be accounted for, referees.  However, that all being said, you can increase your odds (in my opinion) by using the mathematical models as a good starting point for picking games.   I have seen my own systems get hot and produce many winners over the years.

In fact, I won the 2009 Station Casinos Mini-Contest using my mathematical formulas as the start point for choosing teams.  The mini contest was in the last four weeks of the pro football season and I won with a record of 15-4-1 (picking your top five games each week), a half game ahead of the 2nd place finisher.  But I will say this, the key to winning was in the final week with Philadelphia Eagles playing at Dallas Cowboys.  As it turned out, everyone was on the Eagles +3 points, and in fact my own program stated I should look at the Eagles.  Initially I wrote down the Eagles as a play that week, but after further analysis on my part I decided not to play the game.  What happened?  The Cowboys blew out the Eagles and with that loss most of the top 20 in the mini contest took a loss.  My deciding not to play the Eagles and in fact moving over to a winning side on the Oakland Raiders is what won me the contest.  Had I blindly taken the Eagles as the program advised, I would have finished 14-5-1 and thus lost by a 1/2 point. So as you can see, you still need some analysis to determine if their are any other factors to be considered in the game before taking the computer’s side blindly.

But I am a strong advocate of blending sports and mathematics.   But, don’t forget that there are still human factors that need to be considered.

I will go into more about this subject as we start the football season in September.  Also, I will go into some excellent formulas used for college basketball that have proven to work over the years and are available for free.

I will also be doing a post on football contests here in Vegas for 2010.  I am not sure if I will be getting into any this year (hopefully Stations will have the same one they had last year).

Each week during the pro and college football season I will post my top five computer plays each week.  Again, these will be my computer plays and not based on any of my own analysis.   So be sure to check back quite often here at MichaelFunaro.com this football seasoon.