I have attempted in other musing to confront the question of how much of a team's success is attributable to the coach's ability (aside from that of a recruiter) and how much must be chalked up to the raw talent of the team. Some of what I say here draws on those efforts. Of course, this subject is of intense concern to us now, because regardless of the outcome of this year's recruiting process, Coach Willingham will play next year almost exclusively with players recruited by Davie and mostly with Davie recruits the year after that.

My past efforts at looking at the last four decades of Notre Dame football have shown that an N.D. team needs to generate a roughly +200 point differential over the course of the season to meaningfully compete for the national championship. This phenomenon appears not to be limited to Notre Dame. Very few schools in the modern era have won national championships with smaller point differentials. The only examples that I have been able to find are Penn State's 1982 team (+199) and Colorado's 1990 team (+170) -- the latter being the team that was once beaten, once tied, beat Missouri with the help of a 5^th down play and defeated N.D. 10-9 in the Orange Bowl on a controversial clip call. Some teams have generated much larger point differentials; the largest that I have been able to find is the 1995 Nebraska team (+472) -- that team being the one that scored 62 points on Florida in the Fiesta Bowl.

The point of this investigation was to show that point differentials over the course of the season was an important measure of a team's strength. Interestingly, it confirmed the general impression of all five of N.D.'s most recent coaches. N.D. teams under both Ara and Holtz immediately improved dramatically on this measure. Thus, even though Holtz's 1986 team was only 5-6, the team went from a -4 the previous year to +80, showing that he had the team on the right track. The point differential declined the first year under Devine, Faust and Davie. In Devine's case, however, he was following Ara and managed to keep the program at a reasonably high level. The team declined further under Faust, until it eventually was negative his last year when he was fired. Davie's performance was similar, though actually worse than Faust's. In Davie's first year, the team went from a +226 under Holtz to +17, almost the mirror image of Ara's feat of taking the team from -51 to +210 in his first year.

I have also investigated yardage per play as a measure of a team's prowess on offense and defense. I focused on yardage per play, because total yardage is often a deceiving statistic. A positive turnover differential, for instance, often results in a smaller total yardage output for the game, but often a victory on the field. For instance, in the Rams' 42-17 victory over the Packers in the 2001-02 playoffs, the Packers actually outgained the Rams by about 100 yards, because the turnover differential of 8-1 had the Rams constantly playing on a short field, or scoring without any offensive plays at all.

Yards per play is a much better measure, but not perfect. One of the major problems with yards per play is that passing offenses always need to generate much larger per-play averages to be effective because such a large fraction of pass plays are incompletions. Therefore, yardage per play for rushing plays needs to be adjusted upwards to be realistic. I explained in an earlier post my formula for making this adjustment.

The table below correlates these data. The point differential is the total margin by which N.D. outscored its opposition over the course of the season, or -- in the case of a negative number, as in 2001 -- was outscored. The yards (raw) column is the yardage per play margin over the course of the season, without any adjustment. The yards (adjusted) column is the yardage with the adjustment discussed above. Normally this figure is more favorable to N.D. than the raw figure, because historically N.D. has been a running team. Finally, there is the turnover margin for the year.
 
 

Year	Record	Points	Yards (Raw)	Yards (Adjusted)	Turnover Margin	Comment
1960	2-8-0	-77	-0.530	-0.616	-16
1961	5-5-0	- 7	+0.819	+.1.607	-12
1962	5-5-0	- 33	-0.580	-0.979	+5
1963	2-7-0	- 51	-0.821	+0.021	+11	Devore interim
1964	9-1-0	+210	+2.755	+3.149	+11	Ara hired, near NC
1965	7-2-1	+197	+1.145	+2.001	+7
1966	9-0-1	+324	+2.700	+3.145	+8	NC
1967	8-2-0	+213	+1.832	+2.165	-9
1968	7-2-1	+206	+1.590	+2.491	-8
1969	8-2-1	+217	+1.695	+2.329	+5
1970	10-1-0	+242	+1.983	+2.194	-4	Near NC
1971	8-2-0	+149	+0.772	+1.434	0
1972	8-3-0	97	+1.560	+2.485	-1
1973	11-0-0	+293	+2.389	+3.619	+9	NC
1974	10-2-0	+171	+2.102	+2.670	-7
1975	8-3-0	+100	+0.522	+0.687	-1	Devine hired
1976	9-3-0	+136	+0.844	+1.241	-5
1977	11-1-0	+289	+1.730	+2.195	+13	NC
1978	9-3-0	+96	+0.808	+1.137	+1
1979	7-4-0	+46	+1.079	+0.865	-1
1980	9-2-1	+120	+0.882	+1.563	+2
1981	5-6-0	+72	+0.553	+0.560	+2	Faust hired
1982	6-4-1	+32	+0.528	+1.042	+10
1983	7-5-0	+139	+1.614	+2.008	+3
1984	7-5-0	+60	+0.134	- 0.175	-13
1985	5-6-0	- 4	+0.046	+0.169	+2
1986	5-6-0	+80	+0.691	+0.944	-1	Holtz hired
1987	8-4-0	+121	+0.737	+1.278	+9
1988	12-0-0	+237	+1.330	+2.236	+14	NC
1989	12-1-0	+238	+1.415	+2.779	+12	Near NC
1990	9-3-0	+100	+0.325	+1.325	-2
1991	10-3-0	+176	+1.217	+1.546	+5
1992	10-1-1	+256	+1.791	+2.770	+4	Near NC
1993	11-1-0	+212	+1.703	+2.842	+12	Near NC
1994	6-5-1	+62	+0.882	+1.380	-9
1995	9-3-0	+145	+0.324	+0.375	+8
1996	8-3-0	+226	+2.078	+2.678	-3
1997	7-6-0	+17	- 0.136	-0.202	+3	Davie hired
1998	9-3-0	+80	+0.836	+1.321	+5
1999	5-7-0	+17	+0.390	+0.605	-4
2000	9-3-0	+86	-0.372	+0.104	+14
2001	5-6-0	-1	-0.600	-0.051	+3	Davie fired

 

So, what are some of the more important aspects? First, adjusted yardage per play correlates in a clear fashion with the team's success. Below I match N.D.'s adjusted yardage per play and average the wins in a season, projected into a 12-game season. So, 7.0 wins would, for example, be a season of 7-5.

Yardage Wins

-.501 or less:  4.2

-.500 to .500:  6.1

.501 to 1.500:  7.7

.1.501 to 2.50:  9.3

2.501 and up:   11.0

Indeed, even casual observation shows that a substantial yardage per play margin is the surest route to dominance. The lowest figure for an N.D. national championship team in the last four decades was a still-impressive 2.195 figure for the 1977 team. Years in which N.D. got close to the national championship always featured yardage margins of over 2 yards per play, and often over 2.5. These margins, of course, are not a guarantee that N.D. will be in the hunt. Consider the 8-3 team of Holtz's final year that statistically was similar to some national championship teams (+226 point differential and a +2.678 yards-per-play differential) but suffered some unfortunate miscues (both the Air Force and USC games were particularly sickening examples of contests that should have been won) and was -3 on the turnover differential. Indeed, the notion that an over 2-yard per play advantage is the best recipe for competing for the national championship seems to apply to teams other than N.D., as both Miami and Nebraska this year were a bit over 2.5 yards (adjusted) per play to the good (Miami: +2.664 and Nebraska +3.094), though Colorado and Oregon were a weaker 1.3 yards better than their opponents (Oregon: +1.349 and CU +1.346).

The coaching averages also match our intuitions about the relative strength of the coaches at N.D. Here they are:

Ara:     +2.516

Devine: +1.281

Faust:   +0.721

Holtz:  : +1.798

Davie:   +0.229

Despite the similarities of their career records, Faust actually measures up better than Davie. Faust surely could've had a better overall record, but his poor situational coaching contributed to a 5-16-1 record in games decided by less than a touchdown. Davie's much superior turnover differential (+21 to +3 over 5 years) also allowed Davie to partially disguise some very weak teams. For instance, Davie's 9-3 2000 squad generated only a meager +.104 yardage differential, which is usually a recipe for a .500 team. In fact, no N.D. team that I studied had ever won 9 games with such a small yardage advantage. The only team to approximate it was Holtz's 9-3 1995 team, which was his weakest on this measure. Both squads were probably helped considerably by large turnover differentials.

Holtz measures about halfway between Ara and Devine. Holtz's almost 1.8 yards per play advantage for his career is extremely impressive. But it doesn't match Ara's almost unbelievable figure of over 2.5 yards per play. Another interesting question is why Devine was able to essentially match Holtz's 76% winning rate. The answer does not lie in turnovers. Devine was only +9 over 6 years, while Holtz was +49 over 11 years. Rather, it appears to have been Devine's knack for winning close games (he was 15-6-1 in games decided by 7 points or fewer compared with Holtz's 20-17).

To turn for moment to Ara and Holtz, Ara's success was in the sheer per-play dominance of his teams. He was only +11 in turnovers for his 11 seasons, but his teams were so good they didn't need big turnover differentials to win. Of course, none of this is to say that turnovers are unimportant. A large positive margin can save an otherwise mediocre season (as in 1995 or 2000); a large negative margin can turn an otherwise decent season into a mediocre one (as in 1994), can turn a strong team into a snakebit one (as in 1996 or 1986) or perhaps turn a national championship into a near national championship (as in 1970).

So what does all of this say about the relative role of coaching and recruiting in a team's ultimate success? I continue to believe that the answer is that it's about a 50-50 mix. On average, the change in the yardage per play (either direction, i.e, the absolute value) in years in which there was NOT a coaching switch is .926 yards. Thus, on average, changes in the talent of the team, the difficulty in the schedule etc. account for a little bit less than a yard per play on average. In the years since Ara was hired, the figure in years in which there was a coaching switch is 2.074 yards, i.e., a little more than double the non-coaching switch figure. You may recall as well that the point differential changes by about double the normal amount in coaching switch years. Thus, pure coaching -- conditioning of players, schemes, play calling, getting players in the correct positions, etc. -- appears to be about half the equation, and upgrading the talent the other half.

Some of this becomes more apparent if we look at the years on each side of the coaching switch divide. Ara performed the almost unbelievable feat of improving the team by better than 3 yards per play. This has never been duplicated at N.D., and it would surprise me if it has ever been duplicated in the history of college football.

Devine's hire cost the team over 2 yards per play. This says more about Ara than it does about Devine because Devine was able to keep the team at a reasonably high level, and his first team turned out to be his worst one on this measure.

Faust's hire cost the team over a yard per play. This measure is not unfair to Faust and reflected the general mediocrity of his teams.

Holtz improved the team by nearly a yard per play in his first year. This was a sign of things to come, even if the 5-6 record in 1986 didn't show it.

Davie cost the team a devastating 2.8 yards per play in his first year. This, too, was a sign of things to come. Davie was the anti-Ara.

As a detour, one piece of speculation. Had Holtz not been forced out in 1996, N.D. would've won the national championship in 1998 or, at the very least, would've played in the title game. The championship game that year was between a once-beaten FSU team and an undefeated Tennessee team. The game (the Fiesta Bowl) was won by Tennessee 23-16. Neither team was overly dominant by historical standards, each with approximately 250 point differentials for the season. Tennessee had several close escapes over the course of the year (34-33 over Syracuse, 20-17 over Florida -- in a game Florida dominated statistically, 17-9 over Auburn and 28-24 over Arkansas). FSU was blown out 24-7 by NC State and had fairly close games against Texas A&M, Maryland and Miami (who was fairly weak that year). FSU's big win was defeating Florida in Tallahassee 23-12. Clearly, therefore, an undefeated N.D. team would've been in the national championship game, presumably against Tennessee -- and a strong N.D. team might well have been able to win the game.

Why fixate on 1998? First, the schedule was relatively weak. Davie started off encouragingly by beating defending national champion Michigan, but Michigan was only 9-3 that year in the regular season, and played poorly at the beginning of the season, losing the next week at home to Syracuse. Davie's squad then got blown out in East Lansing by a Michigan State squad that only went 6-6. N.D. then righted the ship with a narrow escape at home against Purdue (31-30). They then ran the table leading up to the U.S.C. game against mostly weak opposition, ending with a 39-36 escape at home against an LSU team that finished with a losing record -- but the game ended with a knee injury to Jarious Jackson on an idiotic intentional safety. Essentially quarterback-less for the U.S.C. game, N.D. lost 10-0 to an 8-5 Trojan squad. Remember as well that the BSC's inclusion of computer rankings had not yet been introduced, so N.D.'s relatively weak schedule would not have hindered it -- essentially the polls determined it, and an 11-0 N.D. team certainly would've been ranked #1 or #2.

It is no coincidence that Davie was his most impressive statistically that year because of the weak schedule. A strong team, one of Holtz's better teams (maybe any of the teams from 1987 to 1993, perhaps the 1994 or 1996 teams as well), would not have been locked in close games with Purdue, LSU and Army. Moreover, if Holtz had stayed, Powlus surely wouldn't have stayed on for the 1997 season. Had he left, Jackson would've had a year under his belt going into the 1998 campaign, and here's to guessing that an experienced Jackson and a team led by Holtz would've not have gotten flustered by a so-so Spartan squad. Indeed, Holtz really left the team in very good shape on statistical measures, and the stars were perfectly aligned for an 11-0 regular season in 1998.

As to the future, Davie's 2001 team was almost identical statistically to Faust's 1985 team. Willingham thus faces essentially same task that Holtz faced. Willingham was able to generate a respectable per-play differential at Stanford in 2001 (a little over a yard). He'll eventually have to do better than that at N.D. to compete for national championships, but the same was true of both Holtz and Ara; they didn't dominate at their prior schools in a way that would consistently put them in the hunt. But they had the knack for getting the most out of their teams, and then won big when they tapped into the institutional advantages that N.D. offered. Willingham's record at Stanford suggests that he'll eventually be able to do the same. But nobody should underestimate the enormity of the task that lies before him. If Willingham is going to turn out to be a good coach, he'll probably take the team to the neighborhood of +1 yards per play next year. On average, a differential like that yields 7.7 wins over a 12-game season. My money is on 8-4 for next year, and if he accomplishes that, the chances are very good that we're on the right track.