College football fans have long since internalized "Turnovers = Turnaround." My ESPN colleague Phil Steele has, in his irreplaceable annual preview magazines, long professed the predictability of turnover margin -- as in, if it's too high or low, it's going to regress toward the mean. And it's going to take some of your wins with it. A lot of what I have tried to do with advanced stats is figure out why -- and how -- this happens.
After I define the luck involved in turnarounds, I'll reveal the luckiest and unluckiest offenses, defenses and teams in college football.
Let's "turnover" luck
There are two different types of regression happening within the realm of turnovers:
1. Fumble luck
Over a long enough period of time, you will recover about 50% of all fumbles -- yours and your opponent's. But in a given season, a team might recover only 25-30%, or it might recover 70-75%.
Coaches will insist there is a skill to this, and maybe there is when it comes to how many players pursue the play (and therefore have an opportunity to join in on the dogpile) or how viciously they do so. Also, the further downfield the fumble occurs, the more likely the defense is to recover it -- there are more defenders and fewer blockers around the ball.
At best, though, you control only a small number of recovery opportunities and will regress toward 50% no matter what.
2. Interception luck
This one is a little bit more vague and less intuitive. We define passes defensed as the combined number of interceptions and pass breakups a defense has. Over time, about 21-22% of your total passes defensed will be interceptions (about a 1-to-3.5 ratio of INTs to PBU).
A few factors can impact this ratio. Safeties (who are more likely to be facing the play) are going to have a more INT-friendly ratio than cornerbacks (who are more likely to have their backs turned from the ball). Plus, your choice of zone defense versus man coverage impacts the number of defenders facing the ball as well. (What a given school's scorekeeper defines as a breakup can have a slight impact, too.)
Still, despite these factors, both your offense and defense will regress or progress toward that 3.5-to-1 ratio given enough games.
Using these two guidelines, then, I have for years tracked a loose turnover luck measure, which compares your actual turnover margin to an expected turnover margin derived from the percentages above.
Turnovers = interceptions + fumbles lost
Expected turnovers = (0.21 * passes defensed) + (0.5 * fumbles)
You have at least a smidge of control over how much you fumble or how aggressively you try to defend an opponent's passing game. But the ball is pointy, and you don't have much control over what happens after it hits the ground, or after a pass is deflected into the air. This is exemplified by running some quick correlations between 2018 and 2019 turnover statistics.
Correlation between a team's 2018 and 2019 (to date) turnover margins: 0.17.
• This isn't very strong and reminds us of how random turnovers can be.
Correlation between a team's 2018 and 2019 expected turnover margins: 0.31.
• Still not incredibly strong, but clearly stronger -- you have a little bit of control over turnover chances but not much control over the turnovers themselves.
Correlation between a team's 2018 and 2019 turnover luck (the difference between margin and expected margin): -0.05.
• If my luck discussion is correct, and the bounces are random, then this correlation will be very close to 0. It is.
The gambling relevance of this idea should be pretty obvious: Turnover luck allows us to see into the future a bit. If a team is riding a particularly strong streak of turnover luck, good or bad, then its future fortunes probably aren't tied all that strongly to its win-loss record.
Last year, 10 teams enjoyed a turnover margin over plus-10. The average turnover margin between them was plus-13.9. However, their average expected turnover margin was only plus-5.9. These teams averaged about eight turnovers' worth of good fortune. These 10 teams' current average turnover margin in 2019: minus-0.1.
Similarly, seven teams suffered a turnover margin worse than minus-10 last year. Their average and expected turnover margins were the exact inverse of the teams above: minus-13.9 and minus-5.9, respectively. Their current average turnover margin for 2019: minus-0.9.
So who are 2019's luckiest and unluckiest teams thus far?
Luckiest offenses
1. SJSU (6 turnovers, 12.8 expected, +6.8)
2. Ole Miss (7 turnovers, 12.8 expected, +5.8)
3. NC State (5 turnovers, 10.4 expected, +5.4)
4. SDSU (3 turnovers, 8.2 expected, +5.2)
5. TCU (8 turnovers, 13.1 expected, +5.1)
Brent Brennan's San Jose State Spartans were so close to moving to 4-2 on Saturday. Instead, a 41-38 loss to Nevada dropped them back to .500 and relegated their odds of bowl eligibility, per SP+, to just 23%.
One of the biggest drivers of SJSU's improvement, however, hasn't necessarily been in its own control. Opponents have defensed 38 Spartans passes so far this year, which would suggest around eight interceptions thrown. Instead, SJSU has thrown just two INTs. Meanwhile, as you'll see below, SJSU defenders have 33 passes defensed (expected INTs: about 7) but picked off 13 of them. That is probably unsustainable.
Unluckiest offenses
1. Hawaii (19 turnovers, 12.0 expected, -7.0)
2. UAB (13 turnovers, 7.7 expected, -5.3)
3. South Alabama (13 turnovers, 8.1 expected, -5.0)
4. NMSU (19 turnovers, 14.1 expected, -4.9)
5. Rutgers (12 turnovers, 7.2 expected, -4.8)
Nick Rolovich's Rainbow Warriors are still serious MWC West contenders -- ESPN's FPI projections give them a 43% chance of winning the division -- but it would certainly help if the turnovers fairy backed off a little.
First, Hawaii has lost nine of its 11 fumbles on the season; now, 11 fumbles is far too many (only 16 FBS teams have fumbled more), but the Warriors have still lost three or four more of them than should have been expected.
Second, although opponents have defensed 30 passes (expected INTs: about 6.3), they've held on to 10 of them. Cole McDonald throws very catchable balls, apparently: He's completing 67% of his passes, but including INTs, less than 30% of his passes have actually hit the ground. In Hawaii's two losses, its TO margin is minus-6. UH plays with a slight (and attractive) recklessness, but it has still been unlucky in the turnover column.
Luckiest defenses
1. SJSU (18 takeaways, 12.7 expected, +5.4)
2. UMass (12 takeaways, 6.7 expected, +5.3)
3. USF (15 takeaways, 9.8 expected, +5.2)
4. Texas Tech (12 takeaways, 7.0 expected, +5.0)
5. FAU (13 takeaways, 8.3 expected, +4.7)
It really isn't a good sign for SJSU that it's had the luckiest offense and defense so far (I'll give you one hint who's atop the overall list below), but let's pause for a minute to admire UMass' exploits. Walt Bell's Minutemen are 1-6 and dead last in SP+, saved from winlessness by a home tossup win over third-from-last Akron. In their six losses -- all to teams ranked worse than 90th in SP+ -- they've been outscored by an average of 53-18.
And they've been lucky.
Their worst-in-FBS defense has averaged nearly two takeaways per game, but that has required them to intercept 28% of all passes defensed and recover five of their opponents' six fumbles. They're allowing 49.9 points per game, and their success has been inflated.
Unluckiest defenses
1. Arizona State (7 takeaways, 14.6 expected, -7.6)
2. UCF (8 takeaways, 13.8 expected, -5.8)
3. NC State (5 takeaways, 9.3 expected, -4.3)
4. Minnesota (9 takeaways, 13.3 expected, -4.3)
5. Pitt (8 takeaways, 12.3 expected, -4.3)
FPI gives Arizona State only a 7% chance of winning the Pac-12 South at the moment -- the Sun Devils' loss at Colorado back in September, combined with a likely loss at Utah this coming week, will probably hold them back. But while freshman quarterback Jayden Daniels is drawing a lot of the headlines during ASU's 5-1 start, it's the defense that has driven the success.
The Sun Devils are 58th in offensive SP+ but 26th in defensive SP+, and they'd perhaps be even higher if they'd gotten a few more bounces. They've recovered only five of opponents' 14 fumbles and held on to only two of 35 passes defensed. If their luck begins to progress in Salt Lake City, an upset could be in the works.
Best and worst overall 2019 turnover luck
Here are the 11 teams with a turnover margin at least five turnovers better than expected. Not all of them have done well this year, but their win total has still potentially been a bit inflated.
1. SJSU (+12 margin, -0.1 expected, +12.1)
2. FAU (+8 margin, -0.2 expected, +8.2)
3. Illinois (+5 margin, -2.8 expected, +7.8)
4. Texas Tech (+4 margin, -3.7 expected, +7.7)
5. Western Michigan (+5 margin, -1.0 expected, +6.0)
6. Boston College (+6 margin, +0.4 expected, +5.6)
7. Texas (+7 margin, +1.5 expected, +5.5)
8. Kent State (+2 margin, -3.2 expected, +5.2)
9. BYU (+3 margin, -2.2 expected, +5.2)
10. Wake Forest (+2 margin, -3.0 expected, +5.0)
11. Washington (+6 margin, +1.1 expected, +5.0)
Here are the 12 teams with a turnover margin at least five turnovers worse than expected:
1. Hawaii (-12 margin, -2.8 expected, -9.2)
2. NMSU (-15 margin, -7.5 expected, -7.5)
3. Arizona State (+3 margin, +9.7 expected, -6.7)
4. USC (-7 margin, -0.5 expected, -6.5)
5. New Mexico (-10 margin, -3.5 expected, -6.5)
6. Utah State (-3 margin, +3.4 expected, -6.4)
7. West Virginia (-4 margin, +2.3 expected, -6.3)
8. Air Force (-3 margin, +2.9 expected, -5.9)
9. Rutgers (-7 margin, -1.4 expected, -5.6)
10. Oklahoma State (-7 margin, -1.4 expected, -5.6)
11. Marshall (-3 margin, +2.4 expected, -5.4)
12. UAB (-2 margin, +3.4 expected, -5.4)
SJSU and BYU are evidently sucking up all of the West's turnover luck, leaving Hawaii, ASU, the New Mexico schools, etc., high and dry.
Remember, although we see lists like this and start declaring that certain teams are "due" a regression or progression in luck, "due" isn't really a stat thing. Technically, Hawaii could keep right on throwing extremely catchable interceptions until the end of time.
Still, the correlations at the top of this piece tell us that one year's luck doesn't correlate with another's, and a half-season's luck doesn't correlate with another half-season's, either. However you're feeling about Hawaii or ASU, for instance, you should maybe feel even better because they've managed to go a combined 9-3 despite the god of turnovers putting a thumb on the scales.