The most revealing and relevant statistics in football are per-play averages. A college team is involved in at least 1500 plays over the course of the season on offense and defense. Special teams plays account for fewer than 200 plays most years. Thus, the surest route to dominance is to be better per play than the opposition. If a team goes through the year averaging 5 yards per play and holds the opposition to 4 yards per play, chances are that this team will prove to be a pretty good team.
Even just raw per play statistics, however, are skewed. In particular they are skewed in favor of teams that have a lot of pass attempts. One reason is that sacks are counted as runs and sack yardage is counted against rushing totals. A clearer picture would actually emerge if sacks were counted as pass attempts and sack yardage counted against passing totals. (The NFL does the latter.) Moreover, passing is an inherently riskier proposition. There is a vastly greater chance of a zero gain (incompletion) passing than there is running. Moreover, a team is between 2 and 3 times as likely to turn the ball over on a passing play than a running play. That ratio would probably be more like 4 to 1 if fumbles on sacks were counted as turnovers in the passing game. So, in rough terms, 4 yards rushing is worth about 7 passing. And, in fact, the NCAA average is about 3.85 per rush and 6.9 per pass attempt.
Thus, to get a clear picture, passing yardage and rushing yardage need to be normalized to fit the overall per play average of about 5.2 yards. This adjustment avoids skewing in favor of pass-happy teams. And, N.D. has always been a rush prone team, and last year was no exception. N.D. ran well over 60% of the time while the NCAA average is about 55%. Even with sacks included, N.D. had fewer pass attempts than rushes just by tailbacks.
Adjusted, then, N.D. averaged 4.988 yards per play on offense and gave up 4.222 yards per play on defense, for a per-play margin of +0.766 yards per play. This, as we'll see, is a decent but not overwhelming margin. It also indicates that N.D. was a somewhat below-average offensive team (though not a bottom-ten team by any means ) but a terrific defensive team. Remember, the adjusted per play average is about 5.2 yards per play. So, N.D.'s defense was a full yard better than average and N.D.'s offense was roughly 0.2 yards worse than average.
So, what does a per-play margin of +0.766 yards mean? First, it means that N.D. was hugely improved from 2000 and 2001. Here are the 2000, 2001 and 2002 adjusted per-play averages:
The 2001 team was slightly worse than the opposition per-play, which is completely consistent with that squad's 5-6 record. In fact, this was the second time in 5 years that one of Davie's teams was actually negative on a per-play basis. The other team was the 7-6 1997 squad which was -0.202. The 2001 squad was ever so slightly positive, but the margin was so thin that the 2001 team was probably deserving of only a 6-5 record or so based solely on its per-play advantage. The explanation for that team's 9-2 regular season mark lies largely in its +14 turnover margin. Every +5 or so on turnovers through a season buys a team another win. In 2001 and 2002, N.D. was only +3 and +5 on turnovers, suggesting a record much closer to what the per-play averages indicate.
Really, as it turns out, Davie's best team was his 1998 team which was about +1.3 yards per play. There were several reasons for this decent margin. One is that because of the leftover recruits from Holtz (particularly Jarious Jackson and Autry Denson, both of whom had good years) Davie had a more talented and complete team. The other is that the 1998 team faced the weakest schedule in modern N.D. history. Most rating services rank N.D.'s 2002 schedule as having been between the 5th and 10th hardest, but the 1998 team faced about the 50th hardest schedule. The 1998 team beat only two teams with winning records (Michigan and Purdue), lost to two mediocre teams (MSU 6-6 and USC 8-5) and many of the teams it beat were down that year. Stanford (3-8), Boston College (4-7), L.S.U. (4-7), Army (3-8), Navy (3-8), Baylor (2-9) and A.SU. (5-6) gave N.D. 7 wins over teams with losing records and 4 wins over teams that were 3-8 or worse.
In 2002 by contrast, all three of N.D.'s losses were to teams that had 9 wins or more, N.D. beat four teams that won 9 games or more (Michigan, Maryland, F.S.U. and Pittsburgh), beat two more teams that finished with winning records (Air Force and Purdue) and only four of its wins came against teams who finished with losing records: MSU (4-8), Stanford (2-9), Navy (2-10) and Rutgers (1-11).
One further little illustration as to how much progress was made. Here's a pop quiz: In 5 years of games away from N.D. stadium, how many teams did Davie beat that had winning records at the end of the year? Here's the answer: two. N.D. defeated L.S.U. (9-3) in 1997 and West Virginia (7-5) in 2000. In 2002, N.D. had three such wins: Maryland (11-3), Air Force (8-5) and F.S.U. (9-5).
It is clear that real, underlying improvements have been made. The 2002 team was considerably better than any Davie team.
Now, that's not to say that a team with a +0.766 per play margin should expect to win 10 games every year. In that sense, N.D. did have some good fortune. Historically, N.D. teams between a +0.5 and +1.5 yards per play have won just under two-thirds of their games. So, under normal circumstances, an N.D. team with a per-play margin of roughly +0.8 yards per play that played a 13-game schedule should expect to be 8-5. If you grant N.D. another win based upon its +5 turnover margin, the Irish would figure to be a 9-4 team. So going 10-3 involved a bit of good fortune, but not an enormous amount.
Now, how far is there to go? The answer is that there's a ways to go.
National-championship-caliber teams are generally an adjusted +2 yards per play better than the opposition. Both Miami and Ohio State were between 2 and 3 yards better than the opposition in 2002. Miami in 2001 was more than 3 yards per play positive. Consider the per-play margins of N.D.'s national championship and near-national championship teams since 1960: 1964 (+3.149), 1966 (+3.145), 1970 (+2.194), 1973 (+3.169), 1977 (+2.195), 1988 (+2.236), 1989 (+2.779), 1992 (+2.770) and 1993 (+2.842). The thinnest margins that have ever allowed N.D. to win a major bowl were 1978 (+1.137) and 1991 (+1.546), though the 1995 squad (+0.375) nearly won the Orange Bowl.
What sort of improvement curve is possible? Interestingly, a fair amount can be gleaned by looking at how a coach affects a team in certain key statistical categories his first year. Let's look at how N.D. coaches have affected their teams in the first year in win total, scoring margin and per-play averages.
WillinghamWin change: +5
Scoring change: +74
Per-play change: +0.817
DavieWin change: -1
Scoring change: -209
Per-play change: -2.880
HoltzWin change: 0
Scoring change: +84
Per-play change: +0.775
FaustWin change: -4
Scoring change: -48
Per-play change: -1.004
DevineWin change: -2
Scoring change -71
Per-play change: -1.580
AraWin change: +7
Scoring change: +261
Per-play change: + 3.128
Getting the team moving the right direction at all the first year has been a certain sign of success at N.D. Since Rockne, only 5 coaches have improved or held steady the win total and improved the scoring margin. Those five coaches are Layden, Leahy, Ara, Holtz and Willingham.
Taking the team the wrong direction the first year indicates that the coach is probably not as good as his predecessor. In Devine's case this was understandable because there weren't very many coaches as good as Ara and although Devine's 8-3 1975 team wasn't nearly as good as Ara's last couple of teams, it was a decent team. Devine turned out to be a very good coach, he just wasn't as good as Ara.
The amount of damage that Davie did in one year, however, simply boggles the mind. He was almost unbelievably lucky to have cost the team only one of its win total (8-3 to 7-6), but Davie's 1997 team was, in rough terms, almost 3 touchdowns per game and 3 yards per play worse than Holtz's 1996 team. In fact, for his entire 5 years Davie averaged only about +0.2 yards per play. By contrast, Ara was +2.5, Holtz was +1.8 and Devine was +1.3. Even Faust was almost a half yard better than Davie, and Faust played much tougher schedules.
In the modern era, with scholarship limits and more parity it's probably not possible to equal Ara's performance. But Holtz provides an interesting analogy.
In many respects, Holtz's and Willingham's first year performances were very similar. Both took over 5-6 teams, though Faust's 5-6 team from 1985 was actually a bit better than Davie's last team. Faust's last team had a per-play margin of +0.169 while Davie's 2001 team was, as noted above, slightly negative.
Here was Holtz's progression beginning with Faust's last year:1985: +0.169
Interestingly, the 1986 team is a very close cousin statistically to the 2002 team, except that the 1986 was almost unbelievably unlucky. The 1986 team was -1 on turnovers, but really its 5-6 record was due to going 1-5 in games decided by 7 points or fewer. By contrast, the 2002 team was 5-1 in such games. The 1986 team looks much more like a 7-4 or 8-3 team statistically, which is what it would've been had it finished 3-3 or 4-2 in those close games.
But the interesting thing is that both Holtz and Willingham were able to make almost exactly the same incremental improvement their first years (Willingham made slightly more headway on per-play and Holtz made slightly more headway in scoring margin). Holtz was making progress from the beginning, it's just that his unlucky 5-6 and 8-4 records his first two years disguised how much headway had been made.
Fortunes evened out a little then in 1988. In both scoring margin and per-play margin, the 1988 team (though a terrific team that thoroughly deserved the national championship) was not as impressive as the 1989 team. Parenthetically, in the current system, the 1989 team, despite its road loss to Miami, would've been the national champion. Based upon schedule strength and the probable computer rankings (both of which would have hugely favored N.D. over once-beaten Miami), the national championship game would have been N.D. versus Colorado, which was won by the Irish 21-6.
In any event, N.D.'s season last year has none of the look of a fluke. Though N.D. was a bit fortunate to win 10 games, 8 or 9 wins was a very likely outcome, and anything worse than that would've required the sort of bad fortune that Holtz endured his first year.