Much thanks to Bill Connelly from Rock M Nation and Football Outsiders College for answering a few questions regarding Texas Tech and the Big 12. Bill does advanced stats for Football Outsiders College and I find them fascinating.
 |
|
and go buy their College Football Almanac |
1. I've often struggled with how to account for the loss of offensive linemen. I see a team like Oklahoma St. and the Cowboys lose 4 of their 5 starters from 2009, and all-world left tackle Russell Okung. I've been kicking around the idea, especially after reading the FO article on a drop in a college team's performance in relation to the NFL Draft, that the higher an offensive lineman is drafted, the greater the impact. Is this sort of concept presently being discussed in the FO Labs and can you discuss how FO accounts for losses along the offensive line?
Bill Connelly (BC): Those concepts (draft talent lost and returning offensive linemen) are being tracked independent of each other, but not necessarily together. The idea of draft points lost (discussed here) finds its way into our projections in a couple of different ways -- the pure sum total is used as one adjustment, as is the sum total as compared to what a team typically loses to the draft. What I mean by that last part is, USC loses a lot more "draft points" per season than, say, Utah. If they lose a boatload of players, they will not be as adversely affected as Utah would be because they are used to losing that much and have clearly succeeded at a high level (2009 aside) no matter what.
Meanwhile, starters returning on the line are accounted for in a couple different ways as well, both as a single total (0-5 returning starters) and as a part of the overall returning starters figures. It really was jarring to see how small the correlation was between returning OL starters and year-to-year change in performance. Here are the correlations per unit:
QB: 0.242
RB: 0.030
WR: 0.279
OL: 0.143
Our projections are based on the weight of the correlations, therefore returning OL's barely makes a dent, at least when compared to the overall figures. Your % of overall returning offensive starters has a 0.315 correlation. So despite our perceptions, a team like Tech, who returns a ton of experience everywhere BUT the offensive line, gets a pass from our projections.
More questions, answers and stats after the jump.
2. Texas Tech gets quite a bit of grief for having such a poor rushing attack, 115th rushing offense according to the NCAA's, but Texas Tech's Rushing S&P+ was 26th in the nation last year. Talk a bit about how there can be such a disparity between what most people look at and FO rankings.
BC: S&P+ is a per-play rating, a "tempo-adjusted" figure, if you will. That's how it's possible to have Virginia Tech and Stanford in the overall Offensive S&P+ top ten along with teams like Florida and Arkansas. it works the same way when you're talking about just runs or passes. Texas Tech clearly ran the ball less than almost everybody else in the country, therefore their per-game rankings were quite low*. But on a per-carry basis, the element of surprise was very kind to Tech over the years. Here are their Rushing S&P+ ranks for the last five years:
2005: 3rd
2006: 20th
2007: 24th
2008: 4th
2009: 26th
I always used to say that Tech had a great advantage because if they ran the ball 18 straight times, opponents would still be playing the pass on the 19th play. In their most successful seasons (2005, 2008), they ran the ball extremely well, albeit successfully. If Tommy Tuberville ends up focusing on the run more than Mike Leach (not saying he will, but it's certainly in his DNA to do so), then Tech almost certainly will not rank as high as they did in 2005 or 2008, simply because the element of surprise is gone. But when Leach chose to utilize the run, Tech was usually rather effective with it.
*The per-game rankings are one of the major reasons I wanted to dive into advanced statistics in the first place. In 2008, Tulane ranked 14th in Pass Defense (in terms of yards allowed per game), Northern Illinois fifth. Meanwhile, 11 members of the Big 12 ranked 89th or lower, including Oklahoma at 99th and Texas at 104th. Is this because Tulane and Northern Illinois had infinitely passing defenses? OF COURSE NOT. It's because teams didn't HAVE to pass on Tulane and Northern Illinois, and those that did weren't any good at it. In almost every game they played that season, Oklahoma, Texas, Texas Tech (94th) and Missouri (118th) were ahead from the opening snap. Therefore teams HAD to pass. Plus, almost every team from the Big 12 sported outstanding passing offenses. It rendered this statistic completely and totally meaningless, and yet as a Missouri fan, I cannot tell you how many times I heard "Missouri had the third-worst pass defense in the country last year, ranking 118th..." By no means was Mizzou's pass defense great in 2008 (or 2009, for that matter), but context is essential. Otherwise stats are just dishonest.
Where did the aforementioned pass defenses rank in 2008 in terms of Passing S&P+?
Oklahoma: 5th
Texas: 16th
Texas Tech: 23rd
Missouri: 78th
Which ranking looks more accurate to you?
3. FO had Texas Tech as the 28th best S&P+ and the 7th best Passing Downs S&P+ last year. There's an ongoing debate between Texas Tech fans regarding the performances of both Steven Sheffield and Taylor Potts, how do these two quarterbacks play out in these two categories?
BC: I don't track the schedule-adjusted S&P+ figure on a per-quarterback basis, but I can tell you what their unadjusted S&P figures were. Here they are per-quarterback (NOTE: sacks do indeed count as part of the passing numbers and not the rushing numbers -- I hate that they are counted as part of rushing in the NCAA stats):
Steven Sheffield: 52.1% success rate, 0.58 PPP**, 1.097 S&P
Taylor Potts: 47.4% success rate, 0.38 PPP, 0.857 S&P
Seth Doege: 40.6% success rate, 0.24 PPP, 0.645 S&P
**PPP = EqPts Per Play, an estimate of how many equivalent points the offense scores per play. S&P is a combination of success rate and PPP, like OPS is the combination of on-base percentage (an efficiency measure like success rates) and slugging (explosiveness, like PPP).
Potts' sample size was clearly the highest, and there's a chance that Sheffield's numbers would have come down if he had played more. But ... Sheffield went 23-for-32 for 234 against Nebraska. I REALLY like Sheffield. Potts is good, but SS is better according to S&P and my own eyes (granted, I did not see the OSU game).
4. Another on-going DTN debate is the effect that head coach Tommy Tuberville will have on the defense, if any at all. Do you have any numbers on how Tuberville's performed while at Auburn and have you come up with any statistics or numbers, or are you kicking any around in the aforementioned FO Labs, as to the effect of a new head coach on what was already a pretty successful program?
BC: We have defensive S&P+ numbers back to 2005. Here are Auburn's Defensive s&P+ ranks (with rushing and passing in parentheses):
2005: 34th (48th rushing, 27th passing)
2006: 20th (40th rushing, 16th passing)
2007: 8th (11th rushing, 13th passing)
2008: 32nd (26th rushing, 53rd passing)
And in their first year without Tuberville...
2009: 41st (53rd rushing, 33rd passing)
Meanwhile, here are Tech's recent ranks:
2005: 73rd (69th rushing, 79th passing)
2006: 48th (60th rushing, 43rd passing)
2007: 60th (79th rushing, 39th passing)
2008: 45th (77th rushing, 21st passing)
2009: 32nd (60th rushing, 23rd passing)
Strangely, Tech's defensive ranking in 2009 was the same as Auburn's in Tuberville's last year. Tuberville's defenses oscillated between being better against the run or the pass. I assume Tech will be strong against the run this year, though if they don't discover a pass rush and a replacement for Jamar Wall, the pass defense could likely suffer.
Of course, the other factor here is the move to the 3-4. One never knows how long it will take to make the personnel changes necessary to do well in a 3-4.
5. You mention this in FO's Big 12 Preview that the Big 12 will play out a lot like it always has in that UT and OU are both expected to finish #1 and #2. Going back to the FO article about how teams that lose talent to the NFL typically have a drop-off in production, is this a case where the Projected F/+ is so strong that's it's tough to argue against the numbers?
BC: This is where comparing draft points lost to the average draft points lost helps the teams at the top. Texas and Oklahoma are clearly used to losing top talent to the draft on a yearly basis, so the fact that they did so again this year isn't going to impact them as much as it would if Baylor were to suddenly have three players drafted in the first round or something.
6. Of the teams ranked by FO, is there a team that has out-performed Texas Tech's 5-year recruiting ranking more than Texas Tech? Also, statistically, what is the biggest reason for Texas Tech out-performing those rankings (statistically speaking, of course!)?
BC: Here are the major conference teams (i.e. not Boise State, TCU or Navy) who have the biggest difference between their 5-year recruiting rank and their 5-year performance rank:
Cincinnati: +41 (33rd in Program F/+, 74th in 5-year recruiting)
Boston College: +33 (15th, 48th)
Wake Forest: +29 (37th, 66th)
Iowa: +23 (22nd, 45th)
Texas Tech: +21 (16th, 37th)
Georgia Tech: +21 (21st, 42nd)
Connecticut: +21 (50th, 71st)
Oregon State: +20 (31st, 51st)
South Florida: +18 (34th, 52nd)
Wisconsin: +18 (28th, 46th)
West Virginia: +18 (10th, 28th)
And just for grins, here are the teams who have underachieved the most...
Notre Dame: -39 (47th, 8th)
Mississippi State: -38 (76th, 38th)
Texas A&M: -36 (55th, 19th)
Duke: -36 (98th, 62nd)
Washington: -35 (71st, 36th)
Thought I'd throw that in there, for what I think are rather obvious reasons.
I've read the preview and if you're looking to spend some money on a preview magazine, this is the place to spend your money. It's worth the $5.00 and you'll have an excellent understanding of how these advanced stats work. One more time, check out Football Outsiders College and go buy their College Football Almanac.