Tuesday, September 25, 2007

The Penalty Mystery – Perhaps Explained

I know this comes amidst “AUBURN MUST DIE!” week, but it has been bothering me, and that itch must be scratched.

For those of you scratching yourselves at this point, the “Penalty Mystery” refers to the evidence assembled (by the SMQ) that more penalized teams actually won more often in 2006. We commented on the phenomena here, here and here.

The issue comes to immediate relevance to Florida fans as the Gators, as they were in 2006, are one of the NCAA’s most penalized teams. At this point in the season, by total penalties committed, Florida is second at 42 to only Texas Tech’s 44.

First of all I wanted to establish that more penalized teams do indeed have more success. Using the NCAA’s statistics we find that heavily penalized teams are more successful than their less penalized brethren – but just barely. To establish this, we used yards per game penalized as the relevant data, which led to the following –

Teams with Winning Records49.44 yards per game average penalties
Teams with Losing/Even Records48.44 yards per game average penalties

Not a big difference, that 1 yard per game. However, it is surprising that more penalized teams are NOT SIGNIFICANTLY LESS SUCCESSFUL than less penalized team.

At the very least, penalties don’t really matter to success.

The question is why?

What we sought out to find is a series of data relations that showed a positive, or most hopefully a “perfect”, correlation between higher penalties and success at some recognized positive statistical attribute. I broke down the team’s statistics by quintiles (The portion of a frequency distribution containing one fifth of the total sample)

I first went to time of possession, which I had earlier theorized resulted in more penalties (as offenses being more likely to commit fouls).

Time of possession doesn’t seem to be the answer. While I found that teams that possessed the ball longer (over half the game, or 30 minutes) did commit more penalties, I didn’t quite find the correlation I was looking for.

What I found was the following –

Over 30 minutes possession – 49.25 yards per game penalized
Under 30 minutes possession – 48.69 yards per game penalized

Or, by quintile -

Top Quintile (Most TOP) – Average 32:06 TOP, 49.49 YPG penalized
2nd Quintile - Average 30:47 TOP, 48.73 YPG penalized
3rd Quintile - Average 29:51 TOP, 48.06 YPG penalized
4th Quintile - Average 29:06 TOP, 50.04 YPG penalized
Bottom Quintile (Least TOP) - Average 28:02 TOP, 48.41 YPG penalized

What causes us trouble here is the 4th quintile, in which these teams are more penalized than the top teams. Some correlation sure, but not what I wanted.

Could it be offense? Are successful offensive teams more likely penalties (as high offense being an indicator of success)?

No again. Our data revealed –

Top Quintile (Most YPG) – Average 427.77 YPG, 52.82 YPG penalized
2nd Quintile - Average 371.27 YPG, 47.07 YPG penalized
3rd Quintile - Average 341.21 YPG, 45.54 YPG penalized
4th Quintile - Average 315.66 YPG, 51.21 YPG penalized
Bottom Quintile (Least YPG) - Average 268.85, 48.01 YPG penalized

I had been hopeful for this stat, as some of the big time offenses in 2006 were the most penalized (Hawaii, West Virginia and even Florida). But this falls apart at the bottom quintiles, which are more penalized than the 2nd and 3rd. However, there is something here in the fact that the most successful offensive (top quintile) teams are the most penalized.

The answer was there if I looked deeper.

It was in the breaking down of the offensive stats between passing and rushing that I think I’ve found it. Initially, in looking at the passing stats, I saw the following heavily penalized teams near the top –

Hawaii – 1st in passing (441 ypg), 7th most penalized
New Mexico St – 2nd in passing (399 ypg), 3rd most penalized
Texas Tech – 3rd in passing (370 ypg), 8th most penalized
BYU – 4th in passing (324 ypg), 17th most penalized

Then, in looking at rushing, I saw the opposite –

Navy – 1st in rushing (327 ypg), 6th least penalized
Air Force – 3rd in rushing (229 ypg), 7th least penalized
Clemson – 5th in rushing (218 ypg), 10th least penalized

We did our quintiles and found the following. Passing first –

Top Quintile (Most YPG) – Average 284.54 YPG, 51.54 YPG penalized
2nd Quintile - Average 226.3 YPG, 50.04 YPG penalized
3rd Quintile - Average 198.68 YPG, 48.96 YPG penalized
4th Quintile - Average 177.7 YPG, 48.46 YPG penalized
Bottom Quintile (Least YPG) - Average 140.4, 45.59 YPG penalized

A “perfect” correlation, with the highest passing teams being the most penalized, and on down. Also note, that for our data here, a difference in yards per game penalized between the highest quintile and the lowest of nearly 6 yards per game is significant.

Next, we look at the rushing stats –

Top Quintile (Most YPG) – Average 200.49 YPG, 44.84 YPG penalized
2nd Quintile - Average 159.74 YPG, 48.85 YPG penalized
3rd Quintile - Average 133.36 YPG, 49.27 YPG penalized
4th Quintile - Average 116.99 YPG, 50.23 YPG penalized
Bottom Quintile (Least YPG) - Average 83.71, 51.65 YPG penalized

Not “perfect”, but damn close. Certainly there is a strong correlation.

Heavy passing oriented offenses commit more penalties, at least in 2006.

Why is this so? If I would have to venture a guess (and without going through the game-by-game stats of the highest passing offenses it is a guess), my guess would be the holding call. When an offensive line is buying time for a quarterback in a passing scheme, it seems more likely that the opportunity for a hold is enhanced. And holds, being from the infraction point, can be yardage killers in the YPG stats.

The final assumption we must make is that offenses that are heavily passing oriented are more successful. This is hard to prove data-wise, as some of the least successful rushing offenses are truly troubled teams. The top 20 passing offenses in YPG in 2006 were (with records) –

1 Hawaii 11-3
2 New Mexico St. 4-8
3 Texas Tech 8-5
4 Brigham Young 11-2
5 UTEP 5-7
6 Purdue 8-6
7 Louisville 12-1
8 Houston 10-4
9 Kentucky 8-5
10 Missouri 8-5
11 Baylor 4-8
12 Tennessee 9-4
13 Notre Dame 10-3
14 Southern California 11-2
15 Washington St. 6-6
16 Ball St. 5-7
17 California 10-3
18 LSU 11-2
19 South Carolina 8-5
20 Pittsburgh 6-6

Not a bad list, with 13 of 20 having winning records.

Of the bottom 20 passing offenses in YPG, only Navy, Ohio, Arkansas , Wake Forest and West Virginia had winning records, or 5 of 20.

Is this the final word on the issue? Perhaps not, and I would like to look at the 2007 data before (and perhaps older years) before making a final call. But it does appear that passing offenses commit more penalties.

So perhaps penalties are the price for those high-flying aerial attacks. Because, as your parents told you, everything has a price.

Now, back to Revenge Week!


Anonymous said...

Passing teams play in games that have substantially more plays and thus have more oppurtunity for penalties on both offense and defense. I think that is your real connection.

Henry Louis Gomez said...

They used to say that three things can happen when you throw the ball and two of them are bad. That needs to be revised. 4 things can happen and 3 of them are bad.

Anonymous said...

Actually, gone gator, I just ran the numbers, and passing the ball doesn't really result in more plays per game. The top quintile in % of plays being pass attempts had only 1.48 more plays per game than what was average in 2006. The bottom quintile averaged only 0.44 plays per game fewer than average.

And, as it turns out, there is no relationship whatsoever between offensive plays per game and penalties. There is no pattern whatsoever. I've put all my results here.