Some people set out to handicap the NFL using tools such as angles, spread records, and head to head history. Others set off to bet armed with "inside information", close scrutiny of injury reports, or motivational analysis. Many turn to traditional avenues of prediction and base their wagers on looking at points, yards, first downs and other conventional NFL stats. Meanwhile the so-called sharps spend countless hours following line movements and shopping around for perceived value. All share the same goal of beating the point spread whether by looking at information others don’t have, or by more skillfully interpreting common data.

However, if you step back for a moment and think about it, there would seem to be another way to go after the challenge of betting successfully on the NFL. Our goal could be to try and create better numbers -- our own statistics and ratings that would better reflect the true level of teams’ performances and which would prove to be more accurate and more predictive in terms of projecting scores for upcoming games.

An analogy can be found in what has happened in horse racing over the years. For a long time people were content to look at a horse’s final time and use that as a guide to ability. Then along came a number of smart people (including of course Andy Beyer) who realized that the final time was not in and of itself very accurate. Tracks varied in their general characteristics, such that a 1 minute, 10 second time for a six furlong race at Santa Anita would not equate with the same time at Churchill Downs. Moreover, the condition of a given track changes on a daily and even between race basis, depending on variables like water, sand, harrowing and the like. So it was that certain horseplayers set out to create their own speed figures that would take into account a whole range of factors, and not just accept times at face value. Many of them had great success since their ratings were more accurate than what was commonly available. These days the sophistication applied to calculating speed ratings is tremendous, with some organizations incorporating wind velocity, post position bias, weight carried, pace effects and other hard to quantify data to make numbers that will keep them one step ahead of the general horse betting populace.

There are many reasons to believe something similar can be done for the NFL, since the official statistics can be very misleading as to what actually transpires on the field. Consider the following examples of how typical information can distort the truth -

• Inclusion of "non-plays" in averages -- this is commonplace especially with the QB "kneel-down" at the end of the game to run time off the clock. Every time the QB sits down, his team gets charged with a rushing play for -1 or -2 yards. You can find games where a team runs the kneel-down four times (once or twice at the end of the first half, two or three times to run off the seconds in the fourth quarter). Suppose a team rushed 20 times for 100 yards with its real plays, but has 4 for -4 on the kneel downs. Instead of thinking "gee, they averaged a solid 5.0 yards a carry" (20 carries for 100 yards), you think "man, they only got a pretty average 4.0 yards per carry" (24 carries for 96 yards). The same kind of errors creep into official stats with all the QB spiked passes to stop the clock in the two minute offense...yup, it counts as an incompletion to the NFL statkeepers.

• Not taking penalties into consideration -- penalties are a significant factor in NFL football, and yet they are treated separately from the main stats. A good team may be able to draw its opponents into committing defensive pass interference a couple of times in a game. These plays often go for big yards, and yet they don't show up in a team's passing statistics. At the same time, a bad offensive line may constantly have to resort to holding to keep the QB from getting pummeled and yet the negative yards from the holding don't get factored in either.

• Perhaps most importantly conventional statistics don't weight a team's numbers by the situation. If it's 3rd down and 20 to go and I run a draw play that picks up 8 yards, that's in no way a sign of a good rushing play, more a fact of the defense happily giving up the yards to force the punt. On paper though it shows up as a carry for 8 yards and a good gain. On the other hand, if it's fourth and goal at the one-yard line and I power my way into the end zone for a touchdown, I get credited with a carry for a measly one yard. Multiply this effect several times, and the team picking up cheap yardage on the third and long draw play can end up looking like a better rushing team than the powerhouse fullback club.

In light of these and other "mistakes" that enter the official statistics, we might seek to create statistics which really matter and reflect the "truth behind the numbers" as to how the thirty-one respective NFL teams stack up. A first concept for a new breed of NFL "gambling statistics" might be --

Nothing is more fundamental in football than the notion of the drive..."you got the ball and then what happened?"

By charting how a team's drives wind up, we can establish what sort of success they have in scoring Touchdowns (TD), getting into Field Goal range and then having a kicker who can knock it through the uprights for three (FG), struggling to make the necessary yardage and having to Punt the ball (PT), and even the dreaded turnover siblings, Fumble and Interception (T/O). With this information for each team, both on offense and on defense, you can start to accurately assess strengths and weaknesses. You weed out much of the detail but get quickly to the summary of how efficiently a team used its chances to score and prevent scores. Take a look at this example of the "Drive Chart Stats" for last year’s NFC Championship game:

A quick glance shows that while Minnesota was much more potent on offense, scoring touchdowns 9% more often than New York (29% of possessions vs. 20%), they gave up touchdowns 9% more often on defense, as well as yielding field goals much more frequently. In addition, the Giants’ defense was much better at forcing turnovers (16% vs. 11%). With this data in hand and considering the Giants had the additional benefit of home field advantage, it wouldn’t be hard to view the New Yorkers as the better bet given they were 2 1/2 to 3 point underdogs in most sportsbooks.

Drive Charts are the first weapon in our arsenal! However, once we begin compiling the Drive Chart figures, the next obvious jump is to start tracking teams on a play by play basis and thus is born our -

There is no more refined level of detail possible in the game of football than to look at each individual play of a game and plot results. To do this is an incredibly time consuming process whereby you and your cohorts would need to watch EVERY SINGLE PLAY of EVERY SINGLE NFL GAME (well, regular season on, we could skip the pre-season and pro bowl). After you record all the pertinent details into your databases, you could then write and run sophisticated programs that can extract and order the information for what you want ("data mining" is the popular term for this). However all this effort could well be worthwhile as when looking at play by play records you can start to really appreciate the positives and negatives about teams and also individual players. To make it efficient for people to understand at a glance, you’d want to put together a couple of key indicators:

• Average Yards Gained per play everyone who follows football is used to seeing this type of thing, but we can really make it count by not including "non-events" like QB kneel downs, while adding in yardage gained and lost on penalties.

• "Play Success Rate" (PSR) this is a novel statistic to pay attention to, since what it reflects is how often a team achieves "success" on a given play. What constitutes "success" you may ask? Well, you might want to use the formulas which were put forward in an outstanding book (sadly out of print as far as I know) written by Bob Carroll, Pete Palmer and John Thorn titled "The Hidden Game of Football". The idea is to generally view a successful first down play as one that picks up 40% or more of the yardage needed to make a new first down (so on 1st and 10 you need to get 4 or more yards to be successful, on 1st and 20 you'd need to get 8+ yards). On second down you need to get 60% of the remaining yardage necessary to pick up the first (so 6+ yards if it's 2nd and 10, but only 3+ yards if it's 2nd and 5 to go). Finally with third down you need to get enough yardage to get another first down rolling (e.g. on 3rd and 2 you need 2+ yards, on 3rd and 14 you need 14+ yards). These rules would not be hard and fast, as for example a team picking up yards to move into field goal range could be credited with a "successful" play even if they don't meet the requisite percentage of the "to go" yardage. Overall, these PSR numbers would truly reflect a team's consistency in execution and getting the job done. A team with a nifty PSR on first down (say 50%) is going to put a lot of pressure on the other side's defense, since they'll be facing "second and short" with regularity. On the other hand, if a team doesn't come through early in the down sequence, the offense will constantly be facing blitzes as the other side discounts the possibility of a running play.

• Big While we respect the teams that have high PSR ratings for their ability to play hard every down, you also want to note which teams have the "big play" capability. To do this we could credit each running play that picks up 10+ yards and each passing play that gets 20+ (and defensive interference can count for a big play in this regard) as a "Big Gain" and tabulate the results for each team. You'll soon see which teams are "going deep" and which ones try to win the war in the trenches!

• Effectiveness Rating (EFR) similar to the horse racing speed figures, the EFR would be a complex formula that attempts to incorporate many of the above factors into a single number

Put it all together and you have incredible insights into how good a team may be! Take a look at the Play-By-Play ratings for last year’s Minnesota-New York playoff game:

While it’s always easy to call a game after the fact, the play-by-play ratings going into this game projected the Giants would run and pass the ball much more effectively than the Vikings. The Play_By_Play Projected Score was: New York Giants 30, Minnesota 19.

Access to Drive Chart Stats and Play-By-Play Ratings would already represent a near quantum leap in the quality of statistical data available for the typical NFL bettor, but it’s possible to foresee even more sophisticated numbers. Imagine if you could get your hands on official game film where you can see every play from a variety of camera angles. Currently a vast number of the player positions on the field have little or no meaningful statistics representing how well they play and what conditions they are facing. With game film it would truly be possible to open up the hidden aspects of the game by providing block-by-block ratings for offensive linemen, or to understand which defensive linemen are constantly being double-teamed. You could see which running backs are good at breaking tackles and which ones get their yards strictly on the holes the line opens up for them. You can be sure that NFL teams are tracking their tackle percentages, missed blocks and like statistics, and it’s not that unreasonable to think that in the not too distant future, sports gamblers - at least the sharp ones - will be using these kind of numbers too!