Instinctively, we all know that turnovers make a huge difference in the outcome of football games and in the course of seasons. But it's a tricky thing to try to measure, particularly because some turnovers mean a lot more than others. Over the long haul, however, you would figure that the significant and insignificant turnovers would wash each other out. In the end, I think we'd all guess that teams with positive turnover differentials do better than teams with negative turnover differentials.
To try to quantify this, I got the NCAA turnover margin statistics for
2000 and 2001 for all 115 Division I-A teams (they're on ESPN's website).
These are just regular season statistics because the NCAA segregates bowl
games. But, these should give us a pretty good idea of whether there's
a consistent advantage to being positive on turnovers and how big the advantage
is. In the left hand column I indicate the turnover margin per game, i.e.,
plus one turnover per game or better, plus .5 turnovers to .99 turnovers
or better, etc. In the next column I show the win rate of the teams falling
into these groups and in the final column I compute that win rate as a
number of wins for a 12-game schedule (so 9.00 would translate into a 9-3
season).
| Turnover Margin per Game | Win Rate | Wins per 12 Games |
| 1.00 and up | .772 | 9.26 |
| .50 to .99 | .670 | 8.04 |
| .00 to .49 | .587 | 7.04 |
| -.01 to -.50 | .405 | 4.86 |
| -.51 to -.99 | .353 | 4.24 |
| -1.00 and lower | .267 | 3.20 |
The correlation is striking (and was remarkably consistent between years). Teams that achieve better than a 1 turnover per game margin over the course of the season average 9.26 wins per 12 and teams that are negative by 1 turnover or more per game average only 3.20 wins per 12.
The correlation is also remarkably linear. On average, each improvement by .5 net turnovers per game buys a team another 1.2 wins per season. In other words, assuming a 12-game season, every 5 net turnovers over the course of the season buys a team another victory.
Now, of course, coaching has something to do with turnover margins. Teaching running backs to hold the ball correctly, teaching quarterbacks not to throw over the middle late, playing a disruptive style of defense, all of these should correspond to better performance on turnovers. On the other hand, some of it is luck, and turnovers can't be completely avoided, except by an absurdly conservative offense. Some games, some years, it seems as though every time a running back drops a ball it bounces right back to him. Sometimes a pass over the middle that is thrown behind the receiver and hits his shoulder pads bounces right to a linebacker and sometimes it falls harmlessly to the ground. Consider, for instance, that Ara's teams were only, on average, +1 for the season on turnovers, Devine's averaged +1, Faust's averaged +1, Holtz's averaged +6 and Davie's averaged +4. All of them were the same coach each year, yet sometimes, for reasons a coach can do little to control, a team has a good or bad year on turnovers.
For example, Ara's 9-1 1964 team was +11, his 1966 national championship team was +8 and his 1973 national championship team was +9. On the other hand, his 8-2 1967 team was -9, his 7-2-1 1968 team was -8 and his 8-3 1972 team was -1. These, of course, were bad years by Ara's high standards. Ara's teams were so darn good, however, that they could play through some less- than-stellar turnover years and still be very good. His 10-1 1970 team was -4 and his 10-2 1974 team was -7. But even for Ara, it took the good fortune of having the ball bounce the right way a few times to close the deal on a national championship.
Devine had only one really good year on turnovers. His 1977 national championship team was +13. His teams were never better than +1 in any other year.
Faust had only one really good year on turnovers. In 1982 his team was +10 on turnovers. Although that team wound up 6-4-1 (actually a good year for Gerry), it started off as if it was going be a good team (6-1) until a tough loss sent the Irish into a tailspin.
Holtz had three years where his teams were positive by double digits on turnovers: 1988 (+12), 1989 (+14) and 1993 (+12). Not coincidentally, those were his three most successful teams. Holtz was negative on turnovers only four times: 1986, 1990, 1994 and 1996, and each season was a disappointment. His snakebit 5-6 1986 team was -1. The 1990 team (-2), though 9-3 and running over Spurrier's Gators in the Sugar Bowl, did not measure up to the 1988 and 1989 teams, the 6-5-1 1994 (-9) was a disaster by Holtz's standards and the 1996 team (-3) was a bitter ending to a legendary career. The 1996 team was actually more impressive than both the 1988 and 1993 teams in per-play yardage margins and more impressive than the 1993 team in scoring margin, but didn't even go to a bowl. In fact, if you take the "5 turnovers equals one game" formula, the 1996 team would've been 11- 0 if it could've achieved the +12 of 1988 or 1993, and would've been 10-1 if it had managed to hit Holtz's average of +6. Oh, those turnovers.
If we move on to Davie, he also had one really good year on turnovers: 2000, where the team was +14. That generated a 9-3 mark. His other 9-3 team, the 1998 team, was also pretty good: +5. The 5-7 1997 team was -4 and last year's 5-6 team was +3. This does seem to confirm the "smoke and mirrors" impression of the 2000 team. If that team had hit Davie's career average of +4 it probably would've been a 7-4 team.
To see whether these apparently random fluctuations are limited to N.D., let's consider the NCAA data. Here are the 11 teams that in 2001 had a one-turnover-per-game margin for the regular season, followed by the team's regular season record: Miami (+26, 11-0), Fresno St. (+23, 11-2), Bowling Green (+17, 8-3), Maryland (+16, 10-1), Oregon (+14, 10-1), U.S.C. (+14, 6-5), Syracuse (+15, 9-3), Iowa St. (+12, 7-4), Purdue (+12, 6-5), Ohio St. (+11, 7-4), Washington St. (+11, 9-2).
The thing that jumps out at me about this list is that, with a few exceptions, it seems like a list of those schools that were big, pleasant surprises in how well they did in 2001. Certainly Fresno St., Bowling Green, Maryland, Syracuse and Washington St. caught many people by surprise by having much better years than most thought that they would. Oregon and Iowa St. also perhaps belong on this list.
Several of them, however, faced serious exposure in bowl games: Fresno St. got beat by 6-5 M.S.U., Maryland got hammered by Florida, and Iowa St. couldn't get by a 6-5 Alabama team. Syracuse, however, looked pretty good, but its bowl opponent was 6-5 Kansas St. Washington St. won its bowl game, but it was against 6-5 Purdue, another team on the list. Moreover, even of the teams that didn't qualify as pleasant surprises, their bowl performances were generally not good. Ohio St. got beat by Lou's South Carolina team, again. Purdue has been mentioned and U.S.C. looked terrible getting beat by a 7-4 Utah squad. In fact, U.S.C. and Purdue fans should be very, very afraid: their teams last year were basically 3-8 or 4-7 teams posing as bowl teams. Ohio St. fans have no reason for comfort either; going 7-4 with a +11 on turnovers is not a sign of good things to come.
The two teams on the list that in retrospect appeared to be legitimately good teams were Miami and Oregon. My point, of course, is not that having a good turnover margin makes you a bad team. Rather, an inflated turnover margin can give an otherwise mediocre team a good record and can propel a really good team to a national championship or its threshold.
To make sure that this not somehow a phenomenon limited to 2001, let's look at the 2000 data and see if we can spot some poseurs. Here are the 2000 teams that achieved a one-turnover-per- game margin or better in the regular season and their record: Toledo (+22, 10-1), Georgia Tech (+18, 9-2), Florida (+19, 10-2), Oregon St. (+16, 10-1), Notre Dame (+14, 9-2), Cincinnati (+12, 7-4), Miami (+12, 10-1), Northwestern (+12, 8-3), Louisville (+12, 9-2), Arizona St. (+11, 6-5), Michigan (+11, 8-3) and Boston College (+11, 6-5). National champion Oklahoma was +6.
Gee, another hit parade of heartwarming stories, isn't it? Toledo, Georgia Tech, Oregon St., Notre Dame, Northwestern and Louisville would all have to fall into this category. Let's consider what happened after the regular season in 2000. Notre Dame, of course, we know got blown to bits in the Fiesta Bowl by Oregon St., and then went 5-6 this year when the Irish could only muster a +3. Now, how about those Beavers that S.I. had on the cover at the beginning of the year? When Oregon St. went to -8 on turnovers this year, they were a 5-6 team. Note, however, that if you adjust by one win for every 5 turnovers, O.S.U.'s performance is a lot easier to understand. O.S.U's two-year average on turnovers was +4. So in 2000 the Beavers were really more like an 8-3 team (which was plenty good enough to take apart N.D.'s 7-4ish team) and in 2001 they were a team that probably could've been on the cusp between 7-4 and 8-3 (which might explain, in part, why they were able to give Oregon fits in the last game of the season).
Some other ones: Northwestern's 8-3 club got squashed by Nebraska in the Alamo Bowl and when the Wildcats went to -1 on turnovers in 2001 they were 4-7, and you didn't hear about Zak Kustok every waking moment. Northwestern's falloff looks a lot less strange, however, if we think of them in 2000 as more like a 6-5 team. In 2001, if the Wildcats had gotten to their two year average of +5 on turnovers, they would've been a 5-6 team presumably.
Georgia Tech's 9-2 squad got manhandled by L.S.U. in the Peach Bowl, losing 28-14. In 2001, when the Yellow Jackets were only +1 in turnovers, they were 7-5. Again, the dropoff looks a lot less odd if you figure that both the 2000 and 2001 squads would've probably been 8-win teams if they'd met the two-year average of +6 in turnovers.
OK, how about Louisville? Louisville's 9-2 club could not get by Colorado St. in the Liberty Bowl. Louisville, however, went on to another good season (10-2 in the regular season) but was again propelled by a good turnover margin: +8.
Toledo, at 10-1 in 2000, got one of the rawest deals ever in the bowl lineups, and didn't get invited to a bowl. In 2001, they were good but not as good as 10-1, posting a 9-2 record with +3 on the turnover margin. (The Red Rockets did get to go to a bowl, however, this time going to the Music City Bowl and beating Cincinnati 23-16).
Now, to the teams that didn't really qualify as pleasant surprises, but were on the list anyway. Arizona St. (6-5 with a +11), got dominated by Boston College (another 6-5, +11 team) in the Aloha Bowl. It was Aloha for Bruce Snyder and Aloha for the Sundevils too, as this year they were 4-7 when they could only manage a -8 on turnovers. (U.S.C. fans: this is your future.) Boston College, the other side of that coin, appeared to legitimately improve in 2001. The Eagles' turnover margin was still good (+3), but their regular season record improved to 7-4.
OK, how how about Florida? In 2000, Florida was +19 on turnovers and 10-2. In the Sugar Bowl, the Gators faced a Miami team that was 10-1 and also on the list, but at a more modest +11. Miami beat up Florida 37-20. 2001's Florida team, which was 9-2, actually looked a lot more lethal than the 2000 squad (Maryland fans can attest to this), but was -4 on turnovers. Suppose that each team, Florida 2000 and 2001 had been at its 2-year average for turnovers (+7); the 2001 Florida team would've been 8-4 and the 2001 team would've been 11-0, and surely would've played Miami for the national championship. In truth, that seems like a much more realistic appraisal of the ability levels of those teams, and I think few outside of Nebraska doubt the proposition that Florida would've been a much tougher opponent for Miami in the Rose Bowl.
On to Miami. Miami is good, but you knew that already. But as good as they are, there's no doubting that being an unbelievable +26 in turnovers propelled the Hurricanes to the national championship. It's certainly not all luck; a team does not cause 45 turnovers over the course of the season by luck. In fact, Miami is the ONLY team to repeat on the list of teams that were +1 or better turnover per game. But suppose that Miami had only been the +11 that they were the previous year, not the +26. That works out to between 1 and 2 turnovers per game difference. In the Virginia Tech game that Miami won 26-24, Virginia Tech turned the ball over 4 times and Miami twice. Suppose the turnovers had been even. Would the game have come out the same way? Consider again the other side of the coin: the Gators. Florida lost to Auburn by 3 points and turned the ball over 5 times to Auburn's two. Florida lost to Tennessee by 2 points and lost the turnover battle 3 to 1. If Florida had matched its previous year's margin (a gain of about +1 per game) would it have won those games? Would it have won at least one of the games and played in the Rose Bowl? Oh, those turnovers.
So, let's end by considering Notre Dame again. Notre Dame, in reality, over the last two years has been about a .500 team. But you knew that already; we all know that there's work to be done. You'll note that Stanford does not appear on any of these lists. Stanford, despite not playing particularly well in either of the games where we probably watched them (17-13 win over N.D. and the loss in the Seattle Bowl) appears to have been for real: they were 9-2 with a +3 on the turnover margin. Moreover, there are teams on N.D.'s schedule that stand to have their balloons popped: Maryland, Purdue and U.S.C. head the list. Just off the list: Air Force was 6-6 despite being +8, Florida St. was 7-4 despite being +4 and Boston College was 7-4 despite being +3.
Oh, those turnovers.