A while back, FSUncensored argued that "size matters" when it comes to a defense's front 7. He wrote:
Size in the front 7 is a crucial component of a successful defense
Below the jump, I will see if the numbers support this claim.
Is Bigger Really Better?
Must a defense have a bulky front 7 in order to succeed? What ever happened to "speed kills" and "lean mean fighting machine?" I mean, isn't there more than one way to do defense? Not according to FSUncensored. Consider this quotation from FSUn:
"In Part One (see link in lead paragraph), we discussed the reasons for FSU's poor run defense. Simply put, FSU's defense lacks the minimum bulk in its front 7 defenders to stop the run without dedicating another defender (from the secondary) to the cause. Committing an extra defender to the run overtaxes our secondary, leaving us vulnerable to play action fakes."
FSUncensored's argument in full can be found here and here.
I take FSUn to essentially be saying this: if the front 7 lacks the minimum bulk, then either (a) we cannot stop the run or (b) our pass defense suffers. That is, if the front 7 of a defense lacks the minimum bulk, then that defense will not be successful. (I'll call this claim the Bulk Thesis) I'll take "successful" to mean either (I) holding quality opponents to fewer than 5 YPP; or (II) an Adjusted Defensive Efficiency score of -0.40 or better. I'm using two different metrics for the sake of thoroughness. A successful defense need only satisfy one of the two.
"Minimum Bulk"
So what is "minimum bulk?" This is difficult to figure. But rather than setting an arbitrary number, I'll take it to be the average total weight of the top 10 (by Adjusted Defensive Efficiency or "ADE") defenses in the country in 2008, minus the standard deviation, which comes to a grand total of 1778 total lbs. I'm not sure how else to come up with a "minimum bulk" number. I should also mention that I'm not very skilled with statistics -- I'm learning as I go. So I'm likely to commit some horrible fallacy. Caveat lector.
TEAM | ADE rank | F7 Weight | |
1 | Florida | 1 | 1845 |
2 | BC | 2 | 1902 |
3 | Wake | 3 | 1875 |
4 | Clemson | 4 | 1785 |
5 | USC | 5 | 1860 |
6 | UConn | 6 | 1738 |
7 | UNC | 7 | 1865 |
8 | TCU | 8 | 1876 |
9 | Texas | 9 | 1807 |
10 | VAtech | 10 | 1770 |
AVG | 1832.3 |
Standard deviation = 54
Minimum Bulk = 1778 lbs.
Now we can immediately see that two teams in the top 10 will fall below our minimum bulk number of 1778 lbs.: UConn and VATech. So right away we seem to have some evidence contrary to FSUn's Bulk Thesis. However, this is statistically expected. Now we can give the Bulk Thesis more detail:
Bulk Thesis: if the total weight of the front 7 of a defense is less than 1778 lbs., then either (I) that defense will not hold quality opponents to fewer than 5 YPA; or (II) that defense will not have an Adjusted Defensive Efficiency score of -0.40 or lower.
FSUn may balk at the amount of specificity I've given his claim, but otherwise there doesn't seem to be any way to test it. And if his claim can't be tested then it doesn't amount to much, since it is, after all, a claim about measurable things.
So here's what we need in order to show that the Bulk Thesis is false -- we need a significant number of examples of defenses who meet the following criteria:
(i) the total weight of their front 7 is less than 1778 lbs.
AND either
(ii) they hold quality opponents to fewer tha 5 YPA
OR
(iii) they have an ADE score of -0.40 or lower.
Now look at the following chart from 2008
TOTAL DEFENSE AGAINST WINNING TEAMS ONLY (ranked by Yards per Play):
Name | G | Rush Yards | Pass Yards | Plays | Total Yards | Yards/Play | Yards/G | |
---|---|---|---|---|---|---|---|---|
1 | USC | 7 | 621 | 1022 | 442 | 1643 | 3.7 | 234.7 |
2 | Boston College | 10 | 940 | 1782 | 638 | 2722 | 4.3 | 272.2 |
3 | TCU | 6 | 282 | 1336 | 371 | 1618 | 4.4 | 269.7 |
3 | Florida | 10 | 1076 | 1741 | 636 | 2817 | 4.4 | 281.7 |
3 | Tennessee | 6 | 695 | 1033 | 390 | 1728 | 4.4 | 288.0 |
6 | Connecticut | 7 | 702 | 1193 | 423 | 1895 | 4.5 | 270.7 |
6 | Wake Forest | 9 | 1198 | 1425 | 578 | 2623 | 4.5 | 291.4 |
6 | Iowa | 7 | 715 | 1354 | 459 | 2069 | 4.5 | 295.6 |
9 | UCF | 6 | 916 | 1272 | 463 | 2188 | 4.7 | 364.7 |
10 | Ohio State | 8 | 904 | 1636 | 530 | 2540 | 4.8 | 317.5 |
10 | Clemson | 8 | 1328 | 1346 | 561 | 2674 | 4.8 | 334.3 |
10 | California | 8 | 1218 | 1589 | 587 | 2807 | 4.8 | 350.9 |
10 | Boise State | 6 | 965 | 1248 | 462 | 2213 | 4.8 | 368.8 |
14 | Utah | 6 | 665 | 1272 | 397 | 1937 | 4.9 | 322.8 |
14 | Notre Dame | 6 | 1022 | 929 | 398 | 1951 | 4.9 | 325.2 |
14 | North Carolina | 9 | 1343 | 1925 | 663 | 3268 | 4.9 | 363.1 |
If we toss out Utah and Boise State (weak schedule), and Notre Dame (they ran a 3-4 in 2008) all 12 of the remaining teams gave up less than 5 YPA against quality opponents, and thus they all satisfy criteria (ii). Now which of these teams also satisfy criteria (i)? Only 2: UConn (1738) and UCF (1762). And let's be honest -- UCF didn't play a single ranked team, so they're out. Hmmm, only UConn remains. FSUn's Bulk Thesis is looking pretty good.
Now what if we look at ADE scores?
TEAM | ADE |
Florida | -0.567 |
Boston College | -0.547 |
Wake Forest | -0.511 |
Clemson | -0.491 |
USC | -0.483 |
Connecticut | -0.46 |
North Carolina | -0.443 |
TCU | -0.437 |
Texas | -0.421 |
Virginia Tech | -0.406 |
Mississippi | -0.405 |
Only these 11 teams had a score of -0.40 or lower. We've already eliminated most of them, but the new candidates are USC, UNC, and VATech. Of these, which satisfy criteria (i)? Only VATech (1770 total lbs.). So after all that work, we are left with only two examples that defy the Bulk Thesis: Virginia Tech and UConn (our original two suspects!). Very disappointing. Perhaps I've been too stringent.
Revising the Criteria
Suppose I were to change the minimum bulk to 1800 lbs. (still much less than the average of the top 10), and I were to raise the minimum ADE to, say -0.30. Would I get different results? You bet. The highlighted teams below satisfy our revised criteria.
TEAM | ADE |
BULK |
YPP against winning teams |
Florida | -0.567 | 1845 | 4.4 |
Boston College | -0.547 | 1902 | 4.3 |
Wake Forest | -0.511 | 1875 | 4.5 |
Clemson | -0.491 | 1785 | 4.8 |
USC | -0.483 | 1860 | 3.7 |
Connecticut | -0.46 | 1738 | 4.5 |
North Carolina | -0.443 | 1865 | 4.9 |
TCU | -0.437 | 1876 | 4.4 |
Texas | -0.421 | 1807 | 5.6 |
Virginia Tech | -0.406 | 1770 | 5.1 |
Mississippi | -0.405 | 1800 | 5.1 |
Iowa | -0.393 | 1833 | 4.5 |
Ohio State | -0.322 | 1811 | 4.8 |
Florida State | -0.32 | 1735 | 5.1 |
Cincinnati | -0.315 | 1697 | 5 |
Pittsburgh | -0.314 | 1760 | 5 |
Oklahoma | -0.312 | 1748 | 5.7 |
California | -0.335 | 1881 | 4.8 |
ECU | -0.326 | 1829 | 5.1 |
South Carolina | -0.327 | 1882 | 5.2 |
Tennessee | -0.319 | 1781 | 4.4 |
Now I get Clemson, UConn, VATech, FSU, Cincinnati, Pitt, Oklahoma and Tennessee -- all as examples contrary to the Bulk Thesis. This seems like a significant number (8) of quality examples. Out of the 21 teams with the best ADE scores (and respectable schedules) nationally, more than a third of them defy the Bulk Thesis. But if the Bulk Thesis were correct, wouldn't we expect the vast majority of the best defenses to conform to it? I'll put it another way.
If bulk on the front 7 (relative to your competition) is crucial (necessary) for successful defense, then it follows that nearly all successful defenses should have bulky front 7s. In other words, FSUn is suggesting that successful defense is an excellent predictor of front 7-bulkiness. But if successful defenses have bulky front 7s only 62% of the time, this isn't a reliable enough number for prediction -- it's just slightly better than a coin-flip. That is, we shouldn't reliably expect that successful defenses will have huge front 7s. So with the revised criteria, the Bulk Thesis appears false or on shaky ground at best. (Note: If you put the ADE number back down at -0.40 but leave the bulk number at 1800, then the percentage improves to 72%.)
Conclusion
So you can take my post one of two ways:
- You can go with the original proposed criteria (minimum bulk of 1778, YPP < 5.0, ADE at least -0.40) and conclude that FSUncensored's Bulk Thesis is absolutely confirmed by the numbers. Two counter-examples just isn't enough to disprove his claim.
- You can go with my revised criteria (min. bulk of 1800, YPP<5.0, ADE at least -0.30), and conclude that the Bulk Thesis is false -- there are a significant number of defenses that defy the principle. These defenses have front 7s that are lighter than other elite lines, and yet they performed at the highest levels in 2008.
Now if you take option (1), then you have to agree that the Noles desperately need to beef up their front 7. If we start the year with a total weight under 1778, then we may be in for trouble. The numbers suggest we won't be able to hold our YPP under 5 (excluding cupcakes), which is crucial.
If you take option (2), then you might think that there is another way to do things. We can have a leaner, meaner defensive front, similar to VATech, Clemson and Tennessee. But still, all three of these front 7s were in the 1770 range last year, so we might as well agree with FSUn!
At the end of the day, even though I suspected that the Bulk Thesis was false, the numbers have once again validated FSUncensored's excellent football instincts.