In past discussions, I evaluated the cumulative scoring margin of N.D. teams over the course of year as a way of attempting to assess each team's dominance. This yielded some interesting results, because it showed that for a team (and really any school's team that plays N.D's sort of schedule) a 200-point margin over the course of the season was a bare minimum to having a chance of sniffing around the edges of the national championship derby and that closer to 250 was a level at which a team had a good chance to win the NC. Davie's two best years - 9-win seasons in 1998 and 2000 - produced only about 80-point margins, suggesting that those teams were nowhere close to being in the hunt, a fact that was confirmed by their bowl performances. Davie's other years were much worse, and in 2001 N.D. was actually outscored by one point over the course of the season, something that had happened only one other time (Gerry Faust's last season) since Hugh Devore turned over the reins to Ara. It also showed the coaches of Holtz's and Ara's caliber were able to make roughly half the improvement needed in the first season to get their teams in the hunt for the national championship. In Holtz's case, this is disguised somewhat because his snakebit 5-6 1986 team was a huge statistical improvement over Faust's last year, going from -4 to +80. Because a coach can do little to help the raw talent of his team in the first year (and only slightly more the second), this suggests that coaching - in the sense of play calling, conditioning, game planning, placing players in the appropriate positions and the like - plays a significant but not determinative role in the team's success.
This is, I think, more than an academic or purely retrospective question. For present purposes, it goes to the issue of what are reasonable expectations of Coach Willingham in his first and second years.
Other posters made the useful suggestion that looking at yardage indicators might shed some light on this issue. I agreed that this might be a useful avenue of inquiry, but it turns out to be much harder to do in a meaningful way.
First, total yardage is often a misleading statistic. Good kick coverage, good return play and takeaways are all the enemies of accumulating yardage, but the friend of winning football games. Consider: Team A kicks off to Team B. Team B takes a touchback and then moves the ball to midfield. Team B then punts to Team A, whereupon Team A (see Tim Brown in your mind's eye) returns it for a touchdown. Team A then kicks off again to Team B and Team B again takes a touchback. Team B now moves the ball to Team A's 10-yard line and then throws an interception that is returned for a touchdown (in your mind's eye, see the final play of the 1st half at USC in 1988). Team A now leads 14-0 but trails in total yardage 100 to 0. In fact, in that previously referenced 1988 game against then-#1 USC, N.D. was outgained by about 100 yards or so, yet won 27-10. The great LA sports columnist Jim Murray wrote the next day that the game felt like it was 57-10, and he was right. So, let's throw out total yardage as a meaningful indicator.
A better start is to look at yards per play, both gained on offense and given up on defense. Although that doesn't tell you everything you need to know about a football team's performance (it still omits turnovers, kick coverage and returns, and so on), it does at least tell you how efficiently a team is performing on offense when it runs a play and how efficiently it is allowing the opposing team to operate.
Yards per play is a better start, but is still misleading because of certain realities of football and the differences between passing and running. Taking all Division I-A football teams for the last two years (114 teams last year, 115 this), the median yards per play is 5.27. Median yards per rush is 3.85 and median yards per pass is 6.96. Interestingly, this means that about 55% of the plays in Division I-A college football are runs and about 45% are passes.
If yards per play were a major determinant, why do teams ever run, let alone run more often than pass? Even N.D.'s woeful 5.12 yard per pass (the worst in Division I-A football) would be an excellent rushing number. The answer lies in the scattering of the yardage gains. Most college teams complete between 50% and 60% of their passes. This means that nearly half the time the gain is zero (or worse, if the ball is intercepted). Running is much more reliable. If a team averages 4.5 yards per carry, well over 90% of those runs gain at least a yard. Moreover, lost fumbles occur with less than half the frequency per running play as interceptions occur per passing play. (N.D. this year was fairly typical - 12 lost fumbles in 530 running plays; 11 interceptions in 218 passing attempts). Purely and simply, assuming the same yardage per play on average, running the ball is much more effective at getting first downs and moving the ball down the field for scores.
To visualize this, assume you have two teams, one that runs almost every down for 5 yards per carry and another that passes almost every down for five yards per pass. The first team would have an offense that looks like Nebraska's this year (47 rushing TDs) and the other would like the 2001 N.D. team being forced to throw on every down (4 passing TDs). It doesn't take much imagination to see the first team winning 11 games and the other losing 11.
Therefore, unless you make an adjustment to control for these differences between rushing and passing, you will always understate the value of rushing offense and overstate the value of passing offense, and vice-versa on defense. A reasonable way to do this is to multiply both rushing and passing yards by a factor that will bring them to the median for total yards. In other words, multiply the average rush of 3.85 yards so that it looks like the average gain of 5.27 yards and multiply the average pass of 6.96 yards so that it looks like an average gain of 5.27 yards. Thus, the adjustment co-efficient for rushing yards is 1.369 and the adjustment co-efficient for passing yards is .757. For the hypothetical average offense (3.85 yards per rush, 6.96 per pass, and a 55-45 run-pass ratio) the adjustment does not affect yards per play. For run-oriented offenses it improves the yards per play and for pass-oriented offenses it decreases those numbers.
What does all this show about N.D.'s performance in recent times? I could only get detailed statistics back to 1995, but that turns out to be a good a place to start as any. 1995 and 1996 were Holtz's last two years, and he had good, but not extraordinary teams. The 1995 squad went 9-2 and then lost 31-26 to FSU in the Orange Bowl; the 1996 squad went 8-3 and did not accept a bowl invitation. In 1997, of course, Davie took over, so we have the complete data set for his teams.
So, what does it show? For each year I report the average gain per rush, the average gain per pass, the average gain per play (raw) and the adjusted gain per play using the formula set forth above. I also show the same numbers for ND's opponents in their games, and then calculate a raw and adjusted per play differential. To my mind, the adjusted differential is the most important number. If that number is positive by some decent margin, that indicates that ND's offense was operating considerably more efficiently than the defense was allowing the opposition to function. If that number is negative, that would indicate that the team was being dominated on the field. A number close to zero means that the team is in for dogfight after dogfight each week, and can finish well above .500 only with much superior turnover statistics or better kick coverage and return play. I also show the team's turnover margin, something I come back to later.
1995 (Holtz: 9-3)
Turnovers: +8
ND yds/rush: 4.659 Opp. yds/rush: 4.527
ND yds/pass: 8.362 Opp. yds/pass: 7.134
ND yds/play: 5.865 (raw) Opp. yds/play: 5.509
ND yds/play: 6.364 (adjusted) Opp. yds/play: 5.989
Differential: +.324 (raw) Differential: +.375 (adjusted)
1996 (Holtz: 8-3)
Turnovers - 3
ND yds/rush: 5.229 Opp. yds/rush: 3.114
ND yds/pass: 8.524 Opp. yds/pass: 5.671
ND yds/play: 6.238 (raw) Opp. yds/play: 4.160
ND yds/play: 6.943 (adjusted) Opp. yds/play: 4.275
Differential: + 2.078 (raw) Differential: + 2.678 (adjusted)
1997 (Davie 7-6)
Turnovers + 3
ND yds/rush: 4.064 Opp. yds/rush: 4.488
ND yds/pass: 6.839 Opp. yds/pass: 6.402
ND yds/play: 5.130 (raw)
Opp. yds/play: 5.266
ND yds/play: 5.415 (adjusted) Opp. yds/play: 5.617
Differential: - .136 (raw) Differential: - .202 (adjusted)
1998 (Davie 9-3)
Turnovers + 5
ND yds/rush: 4.620 Opp. yds/rush: 3.587
ND yds/pass: 8.611 Opp. yds/pass: 6.258
ND yds/play: 5.832 (raw) Opp. yds/play: 4.996
ND yds/play: 6.384 (adjusted) Opp. yds/play: 5.063
Differential: +.836 (raw) Differential: + 1.321 (adjusted)
1999 (Davie 5-7)
Turnovers - 4
ND yds/rush: 4.180 Opp. yds/rush: 3.986
ND yds/pass: 8.608 Opp. yds/pass: 7.120
ND yds/play: 5.904 (raw) Opp. yds/play: 5.514
ND yds/play: 6.030 (adjusted) Opp. yds/play: 5.425
Differential: +.390 (raw) Differential: + .605 (adjusted)
2000 (Davie 9-3)
Turnovers + 14
ND yds/rush: 4.275 Opp. yds/rush: 3.724
ND yds/pass: 6.833 Opp. yds/pass: 7.722
ND yds/play: 5.000 (raw) Opp. yds/play: 5.372
ND yds/play: 5.510 (adjusted) Opp. yds/play: 5.406
Differential: - .372 (raw) Differential: + .104 (adjusted)
2001 (Davie 5-6)
Turnovers + 3
ND yds/rush: 3.906 Opp. yds/rush: 3.498
ND yds/pass: 5.124 Opp. yds/pass: 6.931
ND yds/play: 4.261 (raw) Opp. yds/play: 4.861
ND yds/play: 4.920 (adjusted) Opp. yds/play: 4.971
Differential: - .600 (raw) Differential: -.051 (adjusted)
Holtz Average:
ND yds/play: 6.654 (adjusted) Opp. yds/play 5.132 (adjusted) Differential: +1.522
Davie Average:
ND yds/play: 5.541 (adjusted) Opp. yds/play: 5.283 (adjusted) Differential: + .158
It is easy to see that there was an immediate fall-off in the performance of the team, mostly on the offensive side of the ball, as soon as Davie took over. It would be hard to attribute this all to a drop-off in talent between the 1996 and 1997 years. Powlus, for instance, returned at quarterback, and while he has deservedly taken his lumps for less-than-clutch play, he averaged over 8.5 yards per play in 1996, and this plummeted to under 7 in 1997. Powlus in 1996 was actually a quite impressive 133 of 232 (about 57%) with 12 TDs and only 4 interceptions for 1972 yards. His numbers under Davie in 1997 were not as good, but still reasonably good, and Jarious Jackson was reasonably good in spot duty. The 1997 team also returned from 1996 its top running back (over 1200 yards) Autry Denson, one of its top receivers Malcomb Johnson, and had other receivers of high ability levels, including Bobby Brown, Raki Nelson and Jabari Hollaway.
Strangely for a defensive coordinator taking over the head job, the defense also got worse, and immediately so. The 1997 defense was almost 1.5 yards per play worse than the 1996 defense. There were some substantial graduation losses, such as Renaldo Wynn, but nothing so devastating that it would account for such a drop in performance.
Overall, Davie's performance was dismal. Twice Davie had negative adjusted differentials, suggesting that the team could not win without a large turnover margin or excellent kick return play. This is borne out by the numbers; with the exception of 1998, if the turnovers were close to even for the year, or even negative as they were in 1997, the team sat right about .500. The 2000 team won 9 games largely on the huge turnover differential. Given that on a per-play basis that squad was almost dead even with the opposition, it probably would have been a 6-5 or 5-6 team had the turnovers been close to even.
Compare this with Holtz's performance. Holtz's relatively weak 1995 team was able to garner a modest .3 yards per play advantage, but that coupled with a decent turnover differential was enough to win nine games. The 1996 team actually dominated teams on both sides of the ball, but a combination of losing the turnover battle and kicking woes (8 missed extra points, including a crucial one against USC that cost ND a BCS bowl bid) meant that it won only eight games. But note the "only" here. That team had over a two-yard-per-play advantage on the opposition, which allowed it to survive these misfortunes and still be a good team. Had this combination of circumstances hit one of Davie's squads it probably would've meant a 3-8 campaign.
The only squad of Davie's that actually came close to dominating on both sides of the ball was the 1998 team. A quick review of the schedule shows that this was one of the weakest faced by N.D. in recent years. Aside from Navy, Baylor and Army presented very weak opponents (though the game with Army turned out to be close). Apparently more difficult opponents LSU and Michigan were both down that year; LSU actually had a losing record that year, although the game was, again, close. Our two reasonably difficult road opponents, MSU and USC, both defeated us -- MSU in a blowout. Essentially, an easy schedule and some heroic play by Jarious Jackson made that team look much better than it was. Thus, although that was Davie's best year, one could easily see one of Holtz's better squads running the table or going 10-1.
In the end, Davie simply could not achieve - with the exception of 1998 -- the kinds of differentials necessary for consistent victories, unless he could generate turnovers or some other advantage. His inability to consistently outplay the opposition on both sides of the ball left him with teams that were headed every year toward .500, and could be bailed out only by a huge turnover margin, as in 2000, or a weak schedule, as in 1998.
So where does this leave us? The bad news is that I can see no evidence that there is a NC caliber team just waiting to be dusted off by Coach Willingham. The rot, so to speak, has gotten pretty deep. Even if we assume that a good deal of the ineffectiveness on offense is due to Davie's and his coaches' ineptitude, it seems hard to believe that good coaching will cause it to just snap back into place and that ND will begin to generate the one-yard or better per play differentials that are necessary for consistent victories.
Now to what I hope is the good news. The defense is actually pretty good -- it's not an illusion. Remember that about 5.3 yards per play is the national average, so holding the opposition under 5 yards, as they did, is a good effort. Coach Willingham was clearly right to retain Coach Mattison in some capacity. His unit probably saved us from a 2-9 or 3-8 season. Willingham's expertise appears to be on the offensive side of the ball, which is where we are in desperate need of help.
Moreover, if you look at Stanford's numbers for this year, they are pretty good on both sides of the ball, against a schedule that was statistically harder than ND's this year.
2001 Stanford
Turnovers + 3
ST yds/rush: 4.352 Opp. yds/rush: 3.526
ST yds/pass: 8.301 Opp. yds/pass: 6.982
ST yds/play: 5.913 (raw) Opp. yds/play: 5.457
ST yds/play: 6.081 (adjusted) Opp. yds/play: 5.080
Differential: + .337 (raw) Differential: + 1.001 (adjusted)
Stanford's ine wins this year were not a fluke. They had the same slight turnover margin as N.D., but managed to generate reasonable advantages on both sides of the ball. They also played very stout run defense, forcing their opponents to throw the ball about 100 times more than they did over the course of the season. Stanford also ran and threw the ball effectively, leading to one of the most effective and balanced offenses in the country.
Assuming we play 12 regular-season games next year, 10 or more wins would be a miracle given how far this team has to come. Nine is probably the high side of realistic. My money is on eight. If we do that, we're clearly on the right track.
Evaluating the "Final Four" Teams in 2001
Here is information on what would've been this year's final 4 apparently, if we had a playoff.
Miami
UM yds/rush: 5.334 Opp. yds/rush: 3.120
UM yds/pass: 8.094 Opp. yds/pass: 5.241
UM yds/play: 6.556 (raw) Opp. yds/play: 3.799
UM yds/play: 6.779 (adjusted) Opp. yds/play: 4.115
Differential: + 2.757 (raw) Differential: + 2.664 (adjusted)
Oregon
OU yds/rush: 5.172 Opp. yds/rush: 3.492
OU yds/pass: 7.675 Opp. yds/pass: 7.278
OU yds/play: 6.286 (raw) Opp. yds/play: 5.499
OU yds/play: 6.516 (adjusted) Opp. yds/play: 5.167
Differential: + .787(raw) Differential: + 1.349 (adjusted)
Colorado
CU yds/rush: 4.769 Opp. yds/rush: 3.815
CU yds/pass: 7.945 Opp. yds/pass: 6.335
CU yds/play: 5.884 (raw) Opp. yds/play: 5.117
CU yds/play: 6.348 (adjusted) Opp. yds/play: 5.002
Differential: + .767(raw) Differential: + 1.346 (adjusted)
Nebraska
NU yds/rush: 5.619 Opp. yds/rush: 3.356
NU yds/pass: 8.231 Opp. yds/pass: 5.172
NU yds/play: 6.216 (raw) Opp. yds/play: 4.239
NU yds/play: 7.358 (adjusted) Opp. yds/play: 4.264
Differential: + 1.977(raw) Differential: + 3.094 (adjusted)
These are just the regular-season statistics; I didn't bother to add in bowl games (though I did for my calculations of ND above). Miami and Nebraska both rolled up large margins (actually pretty similar to the 1996 ND team), although Miami did it against a considerably harder schedule. Nebraska also benefited from 8 home games and, again, mostly weak competition. In the regular season, its only games against good teams turned out to be Oklahoma and CU. The other apparent serious obstacles on its schedule, ND and KSU, were a combined 11-12.
Oregon and CU were less impressive, rolling up less than yard per play average raw, and a bit over a yard adjusted. That roughly one-yard margin appears to be enough to get you in the hunt if you mostly play clutch (as in Oregon's case) or come on strong at the end (as in CU's case).