What Turnovers Actually Cost: PPA/Play Before and After
PPA (predicted points added) is CFBD’s play-level efficiency model — the same kind of metric most sites, including this one, call “EPA.” We took three Week 2 classics — Michigan-Oklahoma, Alabama-Kentucky, and Texas-Ohio State — and measured each team’s PPA per play two ways: with every turnover included, and with the turnovers stripped out. The gap between those two numbers is exactly how much a giveaway (or a takeaway) is worth beyond what the scoreboard shows.
Michigan 17, Oklahoma 10
vs
Oklahoma turned it over twice; Michigan turned it over zero times — but Michigan still fumbled twice and recovered both themselves, and those plays carry real PPA value even when possession never changes hands.
| Qtr / Clock | Team | What happened | PPA on the play |
|---|---|---|---|
| Q2, 13:32 | Oklahoma | Mateer completes to Livingstone for 8 yards, fumbles, recovered by Michigan (Bowles) | -3.34 |
| Q4, 8:49 | Oklahoma | Mateer intercepted by J.Hill, returned 24 yards | +0.23 |
MichiganOklahomaAlabama 45, Kentucky 17
vs
Six total turnovers in this one — Alabama gave it away three times, Kentucky three times, and two of the six went the distance for defensive touchdowns.
| Qtr / Clock | Team | What happened | PPA on the play |
|---|---|---|---|
| Q1, 10:15 | Alabama | Russell intercepted by Humphrey-Grace, returned 2 yardsDefensive TDBack-breaker | -6.61 |
| Q1, 7:17 | Alabama | Russell sacked, fumbles, recovered by Kentucky (C.Works) | -0.96 |
| Q1, 6:44 | Kentucky | Minchey intercepted by L.Metz, returned 34 yardsDefensive TDBack-breaker | -7.23 |
| Q2, 0:55 | Alabama | Russell intercepted by J.Castell | +0.02 |
| Q3, 5:47 | Kentucky | Minchey sacked, fumbles, recovered by Alabama (I.Faga) | -0.51 |
| Q4, 2:42 | Kentucky | Patterson rushes for 3 yards, fumbles, recovered by Alabama (I.Taylor) | -1.66 |
AlabamaKentuckyTexas 24, Ohio State 23
vs
Texas turned it over twice, Ohio State once — on the final play of the game, with the outcome already decided.
| Qtr / Clock | Team | What happened | PPA on the play |
|---|---|---|---|
| Q1, 14:49 | Texas | Manning completes to R.Brown, fumbles, recovered by Ohio State (J.Timmons) | -0.73 |
| Q1, 10:15 | Texas | Manning intercepted by J.McClain, returned 7 yards | -2.05 |
| Q4, 0:17 | Ohio State | Sayin intercepted by G.Littleton, final play of the game | +0.14 |
TexasOhio StateTurnovers moved every offense’s number — just not by the same amount
The size of the swing depends on two things: how many turnovers a team had, and how costly each one was. Kentucky and Alabama each gave the ball away three times, but Kentucky’s turnovers — two defensive touchdowns against them — were far more damaging than Alabama’s, so their PPA/play swung more even though the raw turnover count was identical. Ohio State’s single turnover came on a meaningless final snap and barely moved the needle at all.
“PPA/play without turnovers” is useful for isolating how a team executed on the plays that stayed alive, but it can flatter an offense that got bailed out by turnover luck, or undersell one that was otherwise playing well before the ball came loose. The figures on PRIME’s Advanced page already include every turnover, for exactly this reason.
More on how PRIME builds its numbers
For the mechanism behind why early-season ratings shouldn’t be fully trusted yet, see our Week 4 networks piece. For the full breakdown of every rating on the site, see Methodology.
Turnover counts are CFBD’s official box score (/games/teams, turnovers category) for gameIds 401856679, 401856674, and 401856682 (Week 2, 2026); each was matched to its play-by-play row by hand. PPA/play figures use CFBD’s play-level predicted-points-added model and match the live figures on PRIME’s Advanced page.