[Preface]
As a loyal constituent of Red Sox Nation, I have grown accustomed to analyzing player production through Bill James's Sabermetrics. Examining contemporary and historical performance through the lens of on base percentage, OPS, and runs generated, a clearer illustration of potential future returns is made more tangible.
Like baseball, I think football, especially college football, can be understood and analyzed through methods dissimilar to traditional statistical outputs. These analytical approaches focus on the relationship between a teams production per attempt (generally) and the associated differential accruing to an opponent.
[Why Statistical Contrarianism?]
To be quite honest, there is nothing inherently wrong with looking at traditional football statistics to determine value and production. Evaluating quarterbacks through passing efficiency or points responsible for will give an adequate illustration of how one passer compares to another passer. Total offense statistics can provide a reasonable depiction of how a team's offensive system compares to conference counterparts.
However, just because something isn't broke doesn't mean it can't be fixed.
Traditional statistical formulae are flawed. For example, look at the formula for the aforementioned passing efficiency -
Passing Efficiency
[ (8.4 x {Yards}) + (330 x {Touchdowns}) - (200 x {Interceptions}) + (100 x {Completions})] / [Total Attempts ]
This formula, while encompassing all passing related outcomes, is fundamentally flawed. It fails to take into consideration fumbles resulting from a quarterback sack. It does not consider game situation which may affect number of yards, completions, interceptions, and attempts. And finally, the values applied to each passing category are virtually arbitrary numbers that once represented the average passer.
To quell any skepticism as to the validity of thinking of statistics outside of the box, check out Football Outsiders, Big Ten Wonk, Sabernomics, and Ken Pomeroy. (Note: Pomeroy is a daily must-read for the college basketball season.)
[Alternative Football Statistical Formulae]
Football statistics and Sabermetrics are not bosom buddies. Unlike baseball, there isn't a large number of games to draw a large statistical pool to complete a thorough analysis. In the arena of college football, the statistical pool is limited even more because of the the few number of regular season games per year (12) and a maximum statistical output of four years.
Therefore, it should be expected that variances and statistical anomalies will be more prevalent in a college football sabermetric analysis that that of a baseball or even the NFL illustration.
In addition to the few number of games to draw data, football has other issues that must be considered in making an analysis. Runningbacks and wide receivers will touch the ball fewer times than a quarterback and this may affect the significance of some assembled data, especially in games requiring a quarterback to pass more or a running back to bleed the clock.
And one must not forget the variable football has that baseball doesn't - the weather. Weather can be adjusted for in the formula, but its presence does confound some analysis.
[What this all Means]
Well, nothing.
Really, this has very little bearing on anything in particular. The blog will often look at different types of statistical approaches over this summer to try and figure out why Syracuse has been so absymal the last few years and what needs to occur during the Robinson regime to restore order to this madness.
[Formulae to be Used]
With Syracuse moving to the West Coast Offense (WCO), it's time Orange nation gets used to looking at some of these burgeoning pass-associated statistics. The receiver rating statistic has no real bearing on a team's possibility of success, but rather is used to only compare individual receivers production. The statistics stolen from Bud Goode are great indicators for team success. They are general team categories and do not look specifically at individual player performance.
Yards/Pass Attempt Differential
- Blatantly stolen from Bud Goode's site
Points/Pass Attempt Differential
- Blatantly stolen from Bud Goode's site
Receiving Rating
- This one is quite good. It quantifies the production of a receiver when he touches the ball on a game-to-game basis. Blatantly stolen from Ross Smith.
Offensive/Defensive Efficiency
- Examines the number of plays need to score or yeild points. This approach to offensive/defensive efficiency focuses on ability to score/yeild points rather than yards because a team wins and loses by the number of points scored, not the number of yards allowed/gained per play. Yardage also skews efficiency since it cannot consider field position.
This method is also great because it gives a strong indication of opportunities to score considering the number of drives available in a game.
In terms of yardage efficiency and points scored, teams should generally generate 12 yards of offense for every point scored. Professional Football Research has all this information on this glorious statistic.
If you know of any other alternative statistical models, please let me know at matthew.glaude@quinnipiac.edu. I'm looking for an alternative to passer efficiency/passer rating that can effectively encompass game situation and production.
If worse comes to worse, I'll just sit down and write something out myself patterned after basketball's point per weighted shot and passer rating. And then I'll blow my brains out.
As a loyal constituent of Red Sox Nation, I have grown accustomed to analyzing player production through Bill James's Sabermetrics. Examining contemporary and historical performance through the lens of on base percentage, OPS, and runs generated, a clearer illustration of potential future returns is made more tangible.
Like baseball, I think football, especially college football, can be understood and analyzed through methods dissimilar to traditional statistical outputs. These analytical approaches focus on the relationship between a teams production per attempt (generally) and the associated differential accruing to an opponent.
[Why Statistical Contrarianism?]
To be quite honest, there is nothing inherently wrong with looking at traditional football statistics to determine value and production. Evaluating quarterbacks through passing efficiency or points responsible for will give an adequate illustration of how one passer compares to another passer. Total offense statistics can provide a reasonable depiction of how a team's offensive system compares to conference counterparts.
However, just because something isn't broke doesn't mean it can't be fixed.
Traditional statistical formulae are flawed. For example, look at the formula for the aforementioned passing efficiency -
Passing Efficiency
[ (8.4 x {Yards}) + (330 x {Touchdowns}) - (200 x {Interceptions}) + (100 x {Completions})] / [Total Attempts ]
This formula, while encompassing all passing related outcomes, is fundamentally flawed. It fails to take into consideration fumbles resulting from a quarterback sack. It does not consider game situation which may affect number of yards, completions, interceptions, and attempts. And finally, the values applied to each passing category are virtually arbitrary numbers that once represented the average passer.
To quell any skepticism as to the validity of thinking of statistics outside of the box, check out Football Outsiders, Big Ten Wonk, Sabernomics, and Ken Pomeroy. (Note: Pomeroy is a daily must-read for the college basketball season.)
[Alternative Football Statistical Formulae]
Football statistics and Sabermetrics are not bosom buddies. Unlike baseball, there isn't a large number of games to draw a large statistical pool to complete a thorough analysis. In the arena of college football, the statistical pool is limited even more because of the the few number of regular season games per year (12) and a maximum statistical output of four years.
Therefore, it should be expected that variances and statistical anomalies will be more prevalent in a college football sabermetric analysis that that of a baseball or even the NFL illustration.
In addition to the few number of games to draw data, football has other issues that must be considered in making an analysis. Runningbacks and wide receivers will touch the ball fewer times than a quarterback and this may affect the significance of some assembled data, especially in games requiring a quarterback to pass more or a running back to bleed the clock.
And one must not forget the variable football has that baseball doesn't - the weather. Weather can be adjusted for in the formula, but its presence does confound some analysis.
[What this all Means]
Well, nothing.
Really, this has very little bearing on anything in particular. The blog will often look at different types of statistical approaches over this summer to try and figure out why Syracuse has been so absymal the last few years and what needs to occur during the Robinson regime to restore order to this madness.
[Formulae to be Used]
With Syracuse moving to the West Coast Offense (WCO), it's time Orange nation gets used to looking at some of these burgeoning pass-associated statistics. The receiver rating statistic has no real bearing on a team's possibility of success, but rather is used to only compare individual receivers production. The statistics stolen from Bud Goode are great indicators for team success. They are general team categories and do not look specifically at individual player performance.
Yards/Pass Attempt Differential
- Blatantly stolen from Bud Goode's site
Points/Pass Attempt Differential
- Blatantly stolen from Bud Goode's site
Receiving Rating
- This one is quite good. It quantifies the production of a receiver when he touches the ball on a game-to-game basis. Blatantly stolen from Ross Smith.
Offensive/Defensive Efficiency
- Examines the number of plays need to score or yeild points. This approach to offensive/defensive efficiency focuses on ability to score/yeild points rather than yards because a team wins and loses by the number of points scored, not the number of yards allowed/gained per play. Yardage also skews efficiency since it cannot consider field position.
This method is also great because it gives a strong indication of opportunities to score considering the number of drives available in a game.
In terms of yardage efficiency and points scored, teams should generally generate 12 yards of offense for every point scored. Professional Football Research has all this information on this glorious statistic.
If you know of any other alternative statistical models, please let me know at matthew.glaude@quinnipiac.edu. I'm looking for an alternative to passer efficiency/passer rating that can effectively encompass game situation and production.
If worse comes to worse, I'll just sit down and write something out myself patterned after basketball's point per weighted shot and passer rating. And then I'll blow my brains out.
0 Responses to “Syracusemetrics! Well, Sort Of....”
Post a Comment