Week 13:
Betting Football Totals –
A Theory of Relativity
We live in a remarkable age. This statement is true in many respects but perhaps our age is as remarkable for technological advances as anything. The development and enhancement of the personal computer has made life easier an/or more efficient for untold millions upon millions of people in all walks of life. For those who crunch numbers – the PC has become a way of life. It can be argued that the greatest beneficiaries of the PC have been those who invest in the stock market and those who bet on sports.
The ability to wade through thousands of items of data instantaneously has allowed the creativity of handicappers to flourish. For those of us who engage in this fascinating intellectual exercise the thrill that comes from the discovery of a new tool to add to our handicapping arsenal is unrivaled, though often matched when validated upon the winning of a wager upon which this new discovery was based.
In the past I have shared some thought on what I have labeled “Factor X,” a number that attempts to measure, quantify and explain a team’s efficiency or inefficiency in converting yards to points on offense and preventing the same on defense. It was from this discovery that I sought to find a similar way to handicap Totals, or Overs and Unders in football.
Towards that end I have come up with my Football Theory of Relativity, which is explained momentarily. First and foremost we must understand that tradeoffs that come with simplicity versus complexity. When I first began handicapping sports professionally over two decades ago I wanted to delve into virtually every area that could be measured statistically to get a total feel for what is happening on the field of play and how those events translate into success or failure against the pointspread.
Although the data collection and analysis techniques of those days were primitive and tedious compared to what exists today, the fundamentals underlying my research have not changed. Unlike many who will back track data and find patterns that have existed over past seasons (some call this ‘data mining’) I have always started from the perspective of hypothesizing that certain circumstances should lead to certain results. For example, I would think that it would follow that a team that performed extremely poorly in a given game would tend to rebound in their next game and give a better performance. Basically I would come up with some objective parameters that I would test and observe the results.
One major area of research I have conducted in recent seasons has been to find ways that simplify the analysis without sacrificing the underlying thought that there must be a reasonable and logical connection between the hypothesis being set forth and the results to be expected. In recent years I have concentrated on the relative relationships between teams, their statistics and their performance.
Many handicappers have believed for some time that it may easier to forecast Totals than to predict side winners in football, especially professional football. When you are forecasting Totals you are not as much concerned about which team is stronger or better but rather how the game is likely to unfold or be played.
In support of this contention I point to the fact that nearly 70 percent of the time the total points scored in an NFL game falls outside the normal range of Over/Under lines set by the oddsmakers.
Within the past month I ran a series of three features in our weekly College & Pro Football Newsletter that looked at Totals in the NFL. Without getting into the detailed data it was shown that over the past 20 plus seasons the Over/Under line was between 35.5 and 46.0 almost 80 percent of the time. Yet 36.1 percent of all games featured total points below 35.5 and 34.4 percent of all games featured total points greater than 46.0. Thus, 70.5 percent of all games fell either below the low end (35.5) of the 80 percent range or above the high end (46.0) of that range.
The conclusion is that there exists a great opportunity to succeed in betting totals because of that huge variance from the normal range of Over/Under lines.
Over the years I have used many mathematical models to project where a Total for a specific game would fall but I’ve also tried to simplify the process by looking at the relative rankings of the teams and apply those rankings in a structured manner.
It makes sense that a team that struggled on offense or plays exceptionally well on defense would be a candidate for a low scoring game. Similarly, teams with potent offenses or vulnerable defenses would make the best candidates for high scoring contests. Often, teams will exhibit one characteristic but not the other, or might be merely average in each. Or their opponent might match up with opposite characteristics. I wanted a method that was sound, easy to use and easy to explain.
Hence my Theory of Relativity for betting Totals.
Essentially I am looking for matchups that involve two teams with weak offenses and strong defenses or two teams with strong offenses and weak defenses. Makes sense, of course, but how do we measure strength and weakness in a consistent and meaningful manner?
The simple answer is that I use three sets of measures – yardage, points and yards per play – both offensive and defensive – and I look for games in which both teams show significant variations from the league average. I use team rankings to make those assessments.
An example from this week’s schedule will best illustrate how I perform the rather simple calculations. Following the remainder of this article I shall present the results for all 16 games on the Week 13 NFL schedule as well as show you how I use this method in College football.
I begin by ranking the 32 teams in the three aforementioned categories – both on offense and defense. Let’s use this week’s Green Bay at Chicago game as an example.
Green Bay ranks # 20 in total offense (yards per game), # 17 in points scored and # 23 in yards per play gained on offense. On defense the Packers are # 9 in total defense, # 17 in points allowed and # 8 in defensive yards per play. Chicago’s rankings on offense are # 28, 24 and 29 while the Bears rank -- not surprisingly -- # 1, 1 and 1 in the three defensive categories.
The next step is to subtract the defensive rankings for each team in each category from that same team’s offensive rankings (NOT those of its opponent).
Thus Green Bay’s results would be + 11 for yardage (17 minus 9), 0 for points (17 minus 17) and + 15 for yards per play (23 minus 8).
For Chicago their yardage rating is + 27 (28 minus 1); their points rating is + 23 (24 minus 1); and their yards per play rating is + 28 (29 minus 1).
The more positive the rating the stronger the tendency for the team to be an UNDER team while the more negative the rating the stronger the potential for an OVER.
I look for matchups where we have two teams that rate either “positive” or “negative” in the three categories. A “positive” matchups presents a potential play on the UNDER. A “negative” matchup presents the possibility for an OVER play. But I must insist on certain parameters before giving even more serious consideration to making the play.
The maximum rating a team can have in any of the three categories is + 31 (if they rank # 32 on offense in that category and # 1 on defense). The lowest possible rating is at the other extreme, - 31 (by ranking # 1 on offense and # 32 on defense). The strongest plays will occur when EACH teams’ ratings are at least either + 16 or - 16, representing a difference relative to the rest of the league of at least one half of all teams, A lesser play is in order when the ratings are either at least + 8 or - 8 for each team, representing at least one quarter of the NFL’s 32 teams.
In our example above the Green Bay/Chicago game comes close to qualifying as a STRONG UNDER play in the yards per play category based on the ratings of + 15 (Green Bay) and + 28 (Chicago). The game qualifies as a weaker play on the UNDER using the total yardage ratings of + 11 (Green Bay) and + 27 (Chicago). No Play is indicated using the points criterion as Green Bay’s rating of 0 does not agree with Chicago’s strong UNDER rating of +23.
In developing a structured system for using this method I might assign a value of + 1 when a weak UNDER play is indicated (such as in our example using the yardage and yards per play ratings) and a value of + 2 for a strong UNDER play indication (as would have been the case in the yards per play ratings if Green Bay had been + 16 or higher). A play would be indicated if a matchup had a value of at least + 3 although a case can be made if there is at least a + 2 value. Using the three categories above (yardage, points, yards per play) the maximum possible value would be + 6 ( if each category received a + 2 value). For OVER plays the values would be - 1, - 2, etc.
Looking at the schedule for week 13 there are seven matchups which produce a value of at least + or - 1. In fact, 4 games have a value of +/- 1, 2 games have a value of +/- 2 and 1 game has a value of - 3. These games are as follows –
Buffalo at Miami – Value of + 1 – Play indicated on the UNDER based on the “Points” category in which Buffalo has a rating of + 13 (30 minus 17) and Miami has a rating of + 9 (23 minus 14).
Tennessee at Indianapolis – Value of - 1 – Play indicated on the OVER based on the “Yards Per Play” category in which Tennessee has a rating of - 10 (16 minus 26) and Indianapolis has a rating of - 10 (4 minus 14).
Jacksonville at Cleveland – Value of + 1 – Play indicated on the UNDER based on the “Points” category in which Jacksonville has a rating of + 13 (16 minus 3) and Cleveland has a rating of 20 (28 minus 8).
Green Bay at Chicago – Value of + 2 – Play indicated on the UNDER based on the “Yardage” category (value of + 1) in which Green Bay has a rating of + 11 (20 minus 9) and Chicago has a rating of + 27 (28 minus 1) and on the “Yards Per Play” category (value of + 1) in which Green Bay has a rating of + 15 (23 minus 8) and Chicago has a rating of + 28 (29 minus 1)
Denver at Kansas City – Value of - 1 – Play indicated on the OVER based on the “Yardage” category in which Denver has a rating of - 10 (7 minus 17) and Kansas City has a rating of - 18 (5 minus 23).
Oakland at San Diego – Value of - 2 – Play indicated on the OVER based on the “Points” category (value of - 1) in which Oakland has a rating of - 11 (14 minus 25) and San Diego has a rating of - 13 (2 minus 15) and on the “Yards Per Play” category (value of - 1) in which Oakland has a rating of - 8 (13 minus 21) and San Diego has a rating of - 11 (5 minus 16).
Seattle at Philadelphia – Value of - 3 – Play indicated on the OVER based on the “Yardage” category (value of - 2) in which Seattle has a rating of - 25 (1 minus 26) and Philadelphia has a rating of - 16 (9 minus 25) and on the “Yards Per Play” category (value of - 1) in which Seattle has a rating of - 15 (3 minus 18) and Philadelphia has a rating of - 13 (10 minus 23).
If this seems terribly complicated please read it through several times and you will actually see that this is a rather simple yet structured manner for identifying playable totals. If you don’t keep team rankings on your own they can be found on many websites, including that run by the NFL.
The method also has some validity in college football where there are 119 Division I-A teams. For the colleges I would recommend using a difference of at least 30 for a weak indicator (as compared to 8 in the NFL) and a difference of at least 60 for a strong indicator (as opposed to 16 in the NFL).
It should be noticed that it is best for a minimum of 4 games to have been played before the rankings are considered meaningful with at least 6 games or more likely to produce the most reliable results.
Good luck this weekend and in the future using our Theory of Relativity for handicapping football Totals.
Andy Iskoe has been a professional handicapper for more than 20 years and has been based in Las Vegas since 1991. He has fared well in many major handicapping competitions and contests including cashing multiple times in the prestigious Hilton Super Contest in which he finished third in 2002.
For information on Andy Iskoe’s Newsletters and Premium Selections offerings please visit www.thelogicalapproach.com. He may be reached via email at
logicalapproach@aol.com
What do you think? Make your opinion known on the Message Boards
