Freshly Made-Up Statistics of Offensive Efficiency
by CPT Hoolie on Aug 28, 2009 6:44 PM CDT
Email
PrintWhile trying to predict the effect of the losses of Javon Ringer and Brian Hoyer to MSU, I read all kinds of stats like "Javon Ringer only averages 4.2 yards per carry" as though it's a bad thing. Bad? Good? From the face of it, who knows? [Not very good, it turns out]. So I started toying with the idea of adjusted stats. One of my favorite stat creations of the sabermetricians is the concept of the "+", the "plus", where stats are adjusted based on league factors (and, in the case of baseball, park and team factors). So I set out to create some new stats that used the concept of the "plus". At the same time I wanted to fix what I see are two shortcomings in college football stats.
First of all, "sacks" and their resulting lost yardage are counted as rushing attempts, when they really should be included as passing attempts.
Secondly, there isn't really a way to penalize a team for an interception save an arbitrary weighting of interceptions per pass attempt as a component of the Passer Rating stat. So I decided to subtract the interception return yardage from the normal passing yardage as a penalty to the quarterback for letting his throw get picked off in the first place. For team stats, this also affects the team for allowing the defensive player to return the ball.
So the two new stats I just made up just now are:
NYPPA [Net Yardage per Passing Attempt] where NYPPA = (pass yards - |sack yards| - |interception return yards|) / (attempts + number of sacks). I also calculated Adjusted NYPPA, NYPPA+, where NYPPA+ = (team or player NYPPA / FBS overall NYPPA) * 100. NYPPA+ of 100 equals exactly average.
NYPRA [Net Yardage per Rushing Attempt] where NYPRA = (rushing yards + sack yards) / (attempts - number of sacks). so NYPRA+ = (team or player NYPRA / FBS overall NYPRA) * 100
I did the initial calculations using team passing stats from cfbstats.com. When I sorted by NYPPA+ I noticed that generally the "best" teams (Florida, Oklahoma, Texas) had the highest NYPPA+. To investigate further, I then ran a correlation between my three new made-up stats and win percentage (win percentage stats coming from stassen.com).
Coefficient of Correlation between NYPPA+ and win percentage is 0.65. Between NYPRA+ and win percentage is only 0.36. Thus, the better at passing a team is, the more likely they have a higher win percentage.
So. What does this mean for MSU, losing Brian Hoyer? Hoyer's NYPPA+ in 2008 was a below-average 98.32. Hoyer regressed slightly during his last two years, going from NYPPA+ of 102.28 in 2007 to 98.32 in 2008, both representing mediocre performances. And Ringer? Ringer posted a NYPRA+ of 86.94 in 2008. Some of this comes from the playcalling, but the truth remains that Javon Ringer, while a workhorse back, was not by any means one of the top running backs in the country.
I for one am optimistic that Michigan State's offense will not be any worse off in 2009, despite the loss of Ringer and Hoyer. Fact is, Hoyer wasn't very good in 2009, but he wasn't terrible either. Michigan State was winning games, so there was no impetus for Dantonio to make a switch at quarterback. Kirk Cousins [small sample size alert!] had an NYPPA+ of 112.77 in 2008. It looks like the passing offense should be improved overall in 2009.
This is a FanPost, written by a member of the TOC community. It does not represent the official positions of The Only Colors, Inc.--largely because we have no official positions.
6 comments
|
0 recs |
Do you like this story?
Comments
Ringer's status as a good back
Was based on his stamina and ability to break huge runs late in games. I’d also argue that based on his NFL preseason performances, we can see he’s a talented back. What the NYPRA+ shows is what a back does, behind their O-line.
Light a man a fire, he'll stay warm for a day.
Light a man afire, he'll be warm for the rest of his life.
That's the problem
with football stats: it’s even harder to separate individual performance from team performance than it is with basketball stats.
Anyway, a couple methodological notes:
1) You may be getting a higher correlation between win% and pass efficiency in part because you’re accounting for turnovers (which, of course, tend to have a large impact on winning/losing) in the passing game but you’re not accounting for them in your run efficiency stat. This is understandable, given that I don’t think you can get data on just fumbles lost on running plays.
2) On the topic of interceptions, I think you have to account for the loss of possession. I’ve seen an estimate of 45 yards in terms of the negative value of an interception. That makes intuitive sense, given that the offense could have punted the ball and at least pushed the other team downfield in giving up possession.
Keep up the number crunching. Just 7 days until we have some fresh data to analyze.
Cheer for The Only Colors: Green and White!
by KJ@theonlycolors on Aug 29, 2009 7:58 AM CDT up reply actions
This stat isolates the effect of turnovers in terms of yards directly attributable to the turnover. The interception return yardage is the only thing directly attributable to the offense. (And no, I couldn’t find fumble return stats without going through each game log, but it would have made the NYPRA stat even lower.) Once the return is complete, the subsequent yardage yielded to the opponent is the responsibility of the defense.
The only time a turnover leads directly to points for the other team is in the case of a pick-six or a fumble return for TD. Pick-sixes happen on 10% of all interceptions but only 0.31% of all pass attempts. Is that small enough to be statistically insignificant? I would need to calculate the rate at which teams score in terms of scores per possession.
Intuitively, I look at turnovers in football with the same disdain that the guys at Fire Joe Morgan used to look at scrifice bunts: your at-bats (in the case of football, plays or even possessions) are too precious to throw away for a small return. I would need to calculate the rate at which teams score in terms of scores per possession to see how offensive turnovers correlate with win percentage. Hey, you just gave me an idea for another article!
Perhaps
Could you just make an arbitrary yardage value for fumbles? Seems like a reasonable yardage value could be used to evaluate running.
Light a man a fire, he'll stay warm for a day.
Light a man afire, he'll be warm for the rest of his life.
Maybe,
but I was trying to avoid arbitrary additions or subtractions. I am thinking that fumbles might fall under another separate stat, maybe something along the lines of range factor in baseball?
by CPT Hoolie on Aug 29, 2009 12:48 PM CDT via mobile reply actions
Comments For This Post Are Closed
Secondary Sidebar
User Tools
Related FanPosts on The Only Colors
FanPosts
Community blog posts and discussion.
