Predictive Metrics In Football Handicapping
Ground Rules
Sports handicapping has long been the province of shady practioners who have promised "superior results" and delivered anything but. This cast of characters has adopted a variety of clever methods to separate fools from their money and the parade never stops. You'll find none of that here. The start of college football season is approximately 50 days from today. I don't know if this experiment will succeed or fail. What I do know that I will operate with honesty, integrity and opacity. Once I upload picks and commentary onto site, the text will go unaltered. They will stand as a record that I will defend as either success or failure this fall. This experiment has to be done in real-time in order to offer any value. As a result you'll know virtually the same time as I do if it is working.
Adventures In Football Handicapping: NFL Versus NCAA
For those of you that recall the pre-ESPN era of sports coverage will likely remember the iconic image of the ski jumper crashing at the open of ABC's Wide World of Sports while the narrator said "...and the agony of defeat." Well, over the years I have had an enduring image in my head of sports gamblers sitting around TVs in sportsbooks, sports bars and anywhere else who were "crushed" while watching "The Play" in the 1982 Stanford-Cal game. Imagine thousands of sports bettors simultaneously shrieking at the Stanford players to tackle the Cal kick return team on that Saturday in November 1982. THAT image personifies the "agony of defeat" for me.
Football handicapping, especially college football, might be one of the most difficult endeavors in sports handicapping. There is a limited sample set of performances to calibrate your models by. Each team will get 12-14 offensive possessions per game. A great team might score on 45%-55% of those possessions. The NFL plays 8 home games and 8 road games. Major college football differs in that it is very commonplace for teams to have 9 home games and as few as four road games. In the NFL there is greater parity with regard to the level of play. Major colleges often pad their schedule with several "David versus Goliath" match-ups that make it difficult to accurately forecast what might happen when two more evenly matched teams square off against one another. There is greater turnover in the lineup from year to year in NCAA football than in the NFL. The list of headwinds in college football is practically endless.
In my college basketball efforts I purposefully avoid analyzing match-ups that have Vegas lines greater than 17 points. Once the outcome of the game is clearly decided both coaches and players make decisions during garbage time in a vastly different fashion. As a result the bettor is left too susceptible to back-door covers that can be incredibly maddening. In college football I plan to avoid analysis on games where the line is greater than 28 points. That figure may even prove to be too high but it will be my threshold as the season begins.
Ahhh, but where there is uncertainty, there is opportunity. Recalling my old days playing high school football was my first exposure to sports analytics. We were drilled to know that if we achieved most of our game objectives that we'd be successful and win. Following the games the coaches would break down the game film and assign players grades on our performance. The gridiron gospel according to my former coaches went like this. You make more big plays than your opponent (15+ yds), stay on schedule with regard to down and distance, shorten the field with special teams, finish drives that get inside the opponent's 40 yd line, and win the turnover battle. The team that best accomplished those objective almost always won the football game.
Predictive Metrics In Football Modelling
As a result I've set forth five measureable football metrics that I believe can produce a sufficient edge against a casino sportsbook. I came across the following stats on FootballInsiders.com from the 2013 season that you might find enlightening:
1) If you win the big play battle, you win 86 percent of the time.
2) If you win the stay on schedule battle, you win 83 percent of the time.
3) If you win the drive-finishing battle, you win 75 percent of the time.
4) If you win the field position battle, you win 72 percent of the time.
5) If you win the turnover battle, you win 73 percent of the time.
According to the data study these stats don't change much from year to year. As a result I give a "hat tip" to my old high school football coach for teaching me more about the game than I gave him credit for at the time. If you win 4 out the five key metrics you win over 95% of the time. As for that other 5% of the time...well, Stanford-Cal in 1982 keeps coming back to me.
As I have said in previous posts, I won't be discussing specifics of my sports metrics in any blog post, just the overall concept. The results of selective testing of my hypothesis looks promising in the lab, but needs to be tested in the real world in real-time before I can say it is a success or not.
More to follow soon.
Jack
