clock menu more-arrow no yes

Filed under:

How FEI and S+P (sort of) see the game: Oregon at MSU 2015

New, 35 comments

Early on we have virtually no useful data about how two teams are going to play - but that never stopped us from making a whole lot of vaguely informed guesses into what the game is going to be like. How bad was our defense? How neat are the new S+P numbers? Let's get it on for the first advanced stats article of the season.

Steve Dykes/Getty Images

FEI is the Fremeau Efficiency Index, created by Brian Fremeau. Brian Fremeau is an author at Football Outsiders,ESPN and BCFToys. FEI is an advanced statistical measure for college football that tracks drive efficiency instead of per-play success.

S+P is created by Bill Connelly. Bill Connelly is an author at SBNation, RockMNationFootball Study Hall and Football Outsiders. S+P is an advanced statistical measure which combines success rate, explosiveness per play and opponent adjustments.

New! If you're curious what these numbers mean, here's a glossary, stolen happily from our friends at Roll Bama Roll.

  • FEI: The Fremeau Efficiency Index, an overall team quality metric that is drive-based and opponent-adjusted. For a more detailed discussion of FEI, check out the PTN primer.
  • OFEI: The offensive component of FEI.
  • DFEI: The defensive component of FEI.
  • FPA: FEI Field Position Advantage, a measure of how much field position value a team earned against its opponents.
  • STE: FEI Special Teams Efficiency, a composite measure of a team's efficiency in all facets of special teams (kicking, punting, and returning), based on points per game.
  • S&P+:S&P+ is primarily play-based and consists the Five Factors: efficiency, explosiveness, finishing drives, field position, and turnovers (measured largely by sack rate). This is then adjusted for opponent strength.
  • Second Order Wins: Defined here, this is how many wins a team would have expected to have won if you just take how well they actually did in a game.
  • S&P+: The overall S+P rating, given as both a percentage and as a margin above the average points scored. IE, a team with a +10.0 S+P would likely score 10 points more than a team with a 0 S+P.
  • OS&P+: The offensive component of S&P+.
  • DS&P+: The defensive component of S&P+.
  • Weighted S+P: This weights more recent games more heavily, giving a bit more value to teams that do better as the season goes on (or teams that get devastated by early injuries after looking great).
  • F/+: The F/+ combined ratings combine FEI and S&P+ into one metric that serves as Football Outsiders' official rankings for college football. For a more detailed discussion of F/+, check out the PTN primer.
  • Off. F/+: The offensive component of F/+.
  • Def. F/+: The defensive component of F/+.
  • ST F/+: The special teams component of F/+.
  • line yards are in general how many yards the line is responsible for compared to how many yards the running backs get by themselves.
  • Opportunity rate is how often a team gains at least 5 yards.
  • Power rate is how often a run play converts 2 or fewer yards on 3rd and 4thd own.
  • Stuff rate is how often a running play gets zero or less yards.
  • Sack rate is adjusted to opponent and is how often the team is sacked

So S+P has changed?

S+P has gone some major revisions over the offseason, and likely very much for the better. They become significantly better retrodictively (how the stats demonstrate why a team won after the fact) - over 80% - and become better predictively against the spread as well (about 54%, up from 51-52%). Here's a big ol primer if you're curious about these numbers.

This also makes S+P end up taking quite a lot about what FEI used to do. S+P is about things like success rate (efficiency) and explosiveness - but it's also about finishing drives (what FEI basically measures) and field position, and even cares a lot more about turnovers now. This means that this year, the F+ ratings might, well, not be actually as predictive or useful as just straight up S+P. I'll continue to do the comparisons, but note that FEI didn't do great at predicting games; it ended up at about 35% ATS in the bowl games, for example, whereas S+P got it right about 64% of the time.

But early on, the stats suck, right?

Well, yeah. FEI throws out the first Oregon game completely as it's against an FCS team - so we have absolutely no data other than preseason projection data for FEI for Oregon. That's...awesome. S+P doesn't have enough per-play data to make any real weighted or adjusted data, and even if they did we're talking about a sampling of how Oregon did against a very good FCS team and how MSU did against a directional state team. Not exactly great sampling.

The good news, as I said last year - these teams are remarkably consistent from a year to year basis. The bad news: I was totally wrong about MSU being consistent from year to year, as you'll see. There are significantly bigger changes from year to year than there were in 2013; MSU loses both coaches on their defense and a number of impact players - and most importantly almost their entire offense save Connor Cook. Oregon, meanwhile, loses a number of great players on the offensive line and some 3-star QB no one's ever heard of who can't get anything through tight windows.

How S+P sees the game with 2015 stats:


OVERALL When Oregon has the ball When Michigan State has the ball
Category Oregon

Michigan State

UO Off MSU Def UO Def MSU Off
F/+ Rk 3 (59.8%) 6(48.2%)



S&P Percentile 97.4 94.6



2nd Order Wins 0.9 (-0.1) 0.9(-0.1)



S&P+ 3 (25.1) 9 (20.9) 1 (61.7) 36 (20.5) 95 (36.6) 16 (41.4)

Not what you'd call the most impressive set of stats. It'll get better as the season goes on. The important stuff is the following: Oregon didn't do anything to make anyone think that they were particularly bad (unlike, say, Auburn, which fell 5 spots) and actually improved slightly in S+P's view. Oregon's offense looked as good as ever. Michigan State's offense looked quite good too. MSU's defense looked a smidgen off.

Oregon's defense was also apparently playing. Quite possibly they were playing a football-like sport. By the defensive S+P stats, Oregon's defense compares to teams such as Tennessee (88th), Rutgers (97th), Akron (94th) - and almost every team in Ohio not pegged with a 'state' moniker. In fact, Oregon currently shows as having the worst defense in the PAC-12. It also shows as having a worse defense than Western Michigan (70th). This is also one of the biggest differences in performance that I could find; Oregon's offense has a 94 ranking difference to Oregon's defense. Only Texas Tech (7th best offense, 114th best defense) was more different with that high of an offense, and Oregon's offense has a 15 point advantage on that Texas Tech team.

Based on this incredibly small sample size, we can see this becoming a crazy shootout. MSU has at least some semblance of a defense - but it's hard to say how good that really is when going up against the best offense in the country (suuuuure). Michigan State's offense is good but Oregon's defense is horrible, and if MSU can put up 37 points against the 70th ranked defense, clearly they'll put up even more here, right?

Well, probably. Maybe.

How S+P sees the game with 2014 stats:


OVERALL When Oregon has the ball When Michigan State has the ball
Category Oregon

Michigan State

UO Off MSU Def UO Def MSU Off
F/+ Rk 3( 59.8%) 6 (48.2%)



S&P+ 3 (23.7) 11 (18.6) 2 (46.1) 22 (21.4) 28 (22.4) 64 (98.3)
S&P+ Percentile 97.9% 94.5%



Weighted S&P+ 3(23.3) 12(14.6)



2nd Order Wins 12.9(-0.1) 9.9(-1.1)



Rushing S&P+

5 133.5) 23 (123.7) 52 (105.6) 17 (122.7)
Passing S&P+

4 (142.6) 16(120.0) 36 (109.9) 6 (136.7)
Std. Downs S&P+

5 (129.5) 32 (109.7) 27 (111.8) 13 (123.1)
Pass. Downs S&P+

3 (154.2) 5 (142.7) 62 (101.4) 5 (141.1)
Success Rate+

8 (124.7) 3 (133.0) 49 (104.4) 13 (120.3)
IsoPPP+

2 (152.8) 41 (110.2) 27 (116.7) 6 (138.7)

What's interesting to me is that MSU is known as being this incredible defensive team - but they didn't play like it last year. They were actually very similar in style to Oregon, and won due to their great offense and good defense. In particular, Sparty is better at getting explosive plays and not as good at getting sustained drives, they're good in standard downs but better in passing downs. And also surprisingly - just like Oregon - they were better passing the ball than running it. The notion that MSU is some ground and pound team seems to be taken from the same place that their dominant defense comes from. At least last year - and what I've seen of this year - it's not about running the ball. It's about passing well and explosively.

Similarity scores for Michigan State

Last year, Michigan State was most similar offensively to UCLA (8th), though UCLA is not a great comparison, as they ran better and passed worse than MSU did. MSU was better than Florida State (16th), which...snigger...um...yeah. Those two teams are otherwise somewhat comparable in ability, save that MSU is better at explosiveness and not as good at efficiency.

UCLA was also a good choice for comparison on defense (25th), though Utah(30th) is another decent choice. MSU was better at almost everything compared to UCLA save one - explosiveness. Which is likely the real key to the game, just as it was last year.

Similarity scores for Oregon

Michigan State played three powerhouse offenses: Ohio State (1st), Oregon (2nd) and Baylor (7th). Yikes. In each of those games they gave up over 40 points, so at least that's consistent. Of those, Oregon compares closest  to Ohio State, though was not nearly as good at running the ball.

Defensively Oregon wasn't that close to any of MSU's opponents, but the closest was Michigan (18th). The next closest was Purdue (53rd). Michigan was significantly better at stopping the run but about the same on passing defense.

Oregon's offense vs Michigan State's defense

Yeah, who knows, right?

Comparing last year's results we would expect that Oregon would have something of their hands full. MSU matches up very well against Oregon, and has a better pass defense than run defense (though not by much). The one big factor (as mentioned above) is that MSU gave up a lot of explosive plays - and only one other team in the nation was as good as Oregon was in getting big plays last year (that would be Ohio State, if you were curious). I wouldn't expect sustained run offense success (Western Michigan gained 18 net yards running) but the passing should be quite open. That assumes a lot though - how well Oregon's line holds up, how well our receivers do, and the big question - how well Vernon Adams does against a legitimately good defense in his 4th week in the Oregon system.

I just don't know. My suspicion is that the inability to rely on the running game is going to be problematic for Oregon, and Adams isn't going to be able to make up for it. I also suspect that the incredible set of receivers Oregon has will make MSU occasionally look really, really stupid. I wouldn't expect a lot of sustained drives, but I could easily see Oregon get 5 scoring plays over 20+ yards each. Then again, I don't think Adams has seen an away crowd anything like Michigan State. It's very hard to figure out what a player will be like in their first national spotlight. Let's hope he's up for it.

Oregon's defense vs Michigan State's offense

This probably isn't going to be very pretty.

Ready to see a lot of 3rd and 8s converted? Just like the OSU game, or just like the FSU game, or just like, well, the game against the Eastern Washington Eagles? Yeah...that's where MSU shines. MSU does great on passing downs, and Oregon is not good at all. While Oregon doesn't have much in the way of advantage...well, anywhere, they are likely to limit explosive plays and do decently on standard downs - but good luck getting MSU off the field. As almost every preview of the defense goes, the key for Oregon is going to be generating turnovers and using offensive pressure to force teams into 4th down stops. MSU is a bad matchup for Oregon and that's going to be tough.

And that's assuming the Oregon of 2014, where we had Ifo making absurd catches. We don't have him any more, and chances are good that we'll not have as many turnovers as we did in that game.

Almost nothing frustrates me more than seeing Oregon force a 3rd and 12 or 3rd and 15 - and that team converting it. My suspicion is that this season, at least early, I'm going to be a very frustrated guy.

How FEI sees the game:

Since this is early, only the numbers that are boldly italicized are from this year's projections. Everything else is from last year.


OVERALL When Oregon has the ball When Michigan State has the ball
Category Oregon

Michigan State

UO Off MSU Def UO Def MSU Off
F/+ Rk 3 (59.8%) 6 (48.2%)



FEI Rk .246 (3) .193(6) 3 (.747) 40 (-.212) 14(-.441) 14 (.399)
Field Position 7 (.550) 10 (.545)



Raw Efficiency 3 (.262) 7 (.200) 1 (.872) 24 (-.267) 59 (-.056) 19(.364)
First Down rate

1 (.830) 5 (.533) 101 (.724) 5 (.772)
Available Yards rate

1 (.664) 10 (.353) 83 (.485) 13 (.551)
Explosive Drives

5 (.259) 87 (.161) 59 (.125) 34(.169)
Methodical Drives

31 (.163) 1 (.051) 114 (.184) 84 (.118)
Value Drives

1 (.643) 23 (.309) 74 (.396) 10 (.518)

Congratulations, Oregon: you are the 2015 FEI National Champions! This is about as legitimate as at least one of UW's claimed championships, so there ya go.

Thanks to a harder than expected schedule and the utter obliteration of teams like Stanford, UCLA and FSU, Oregon stayed up in FEI ranks despite losing by 3 scores in the NCG. And look at those offensive numbers. Wow.

Similarity scores for Michigan State

So here's how I really start to wonder about FEI. Who was the best offense in the nation last year by FEI standards? Georgia Tech. Next up was Auburn. You have to go to 7th to find Ohio State. This is a case of FEI absolutely punishing the hell out of a team for having a weak schedule as well as not doing well early on against teams they should have dominated.

But even with that caveat, Michigan State still shows up at 14th. They're not as good as OSU (thank goodness) or, oddly, UCLA (um) but are much better than Arizona (23rd) or Cal (32nd).

2013 had Michigan State's defense as one of the best in the nation. 2014 had them plummet, and boy did FEI not like them. They were fine at shutting down poor offenses, but struggled heavily against strong offenses, being gashed for 41, 46 and 49 points. Even Oregon managed to hold OSU to fewer points. By FEI standards MSU was closest to UCLA (47th), though UCLA had a significantly harder schedule beefing up their value.

Similarity scores for Oregon

Oregon is, surprising no one, closest to Ohio State (7th)...again, seriously? FEI? Come on. Given that OSU and Oregon have similar styles of play they compare similarly too, especially in explosiveness and getting value on drives.

Oregon last year was rated stupid high by FEI due to their ability to inflict grievous damage via turnovers, and as a result they compare pretty closely to Penn State (9th) and Wisconsin (23rd). Oregon's defense is not likely quite as good as that this year. Maybe.

Oregon's offense vs Michigan State's defense

Oregon's advantage is enormous as far as FEI is concerned from last year. It's still hard to say how well Vernon Adams will do, but chances are that it's MSU's response to losing Pat Narduzzi as defensive coach that will dictate how the game is really changed. Their line did great at stopping the run - but they weren't trying to stop a Freeman. Oregon has even better weapons than they did at WR, and there were significantly bad matchups there all day. If Vernon Adams can get enough time I think he will be able to get some very big plays. And boy, was MSU bad at giving up big plays.

That's really tough though, as that was a characteristic of Narduzzi's defense. This year? They didn't give up a whole lot of huge plays (only two that I counted) but gave up a ton of yards. Oregon isn't typically that good at doing methodical drives; we'll see if the MSU defense has really changed that much. At the very least this is somewhat encouraging: Western Michigan threw 50 times, completing 33 of them. I'll happily take a 67% completion rate.

Oregon's defense vs. Michigan State's offense

As far as last year goes, it doesn't get much closer than this. Oregon's specialty on defense is stopping long plays (or at least was...oy) and forcing teams to make mistakes. Connor Cook was less than 50% passing and looked shaky all day. MSU also lives and dies by big plays, and Oregon was able to stop that last year. I'm less confident in our defense this year, but I'm also much less confident in MSU's skill players this year. MSU replaced almost their entire offense and there is going to likely take some time to gel.

At the same time - Oregon's defense needs to get  it together, and fast.

Special Teams

Category

UO(13-1)

MSU(11-2)

EDGE

Special Teams Efficiency
17 (1.435)
18 (1.432)
EVEN
Field Goal Efficiency
46 (.109)
96 (-.250)
OREGON
Punt returns vs. punt efficiency
12 (.090)
21 (-224)
EVEN
Kickoff returns vs. kickoff efficiency
88 (-.182)
20 (-.250)
MICHIGAN STATE
Punting vs. punt return efficiency
49 (-.129)
72 (-.105)
Oregon
Kickoff vs. Kickoff return efficiency
24 (-.236)
15 (.011)
EVEN
Opponent Field Goal Efficiency
71 (.069)
40 (-.157)
Ohio State

Last year, MSU was pretty awesome at a whole lot of special teams. This year? Eeeeh. MSU gave up 170 yards on two kickoffs to Western Michigan and 220 total return yards on the day. Oregon in theory could take advantage of this - but sadly, Oregon was a lot better at punt returns than kickoff coverage. That could be different this year in theory (and we did look decent against Eastern Washington) but chances are good that Oregon won't get that much advantage in kickoff returns. I can hope that Byron Marshall makes them pay - but chances are good that that won't happen.

So what does this all mean?

This is what I said for the NCG:

Oregon has some advantages, including some intangibles. They're facing a quarterback on their third start in the biggest game in the history of college football. Oregon's offense is incredibly good, and Ohio State really hasn't faced anything nearly as good. Ohio State does turn the ball over (22 turnovers on the season) - which isn't nearly as many as FSU, but it's still up there. Oregon has played a harder schedule and dominated that schedule.

But boy, Ohio State's offense is incredible.

if Oregon wins, it's going to be because they will be able to limit the big plays by Ohio State. Oregon will have contained, somewhat, Devin Smith. Oregon will have not let Ezekiel Elliott make some backbreaking giant runs, though he is awfully good at them. Oregon will not have done a lot of mistakes of their own, and Oregon might even have a defensive turnover score or some low-percentage play, and will definitely have won the turnover battle - possibly needing to win that decisively. And somehow, Marcus Mariota will escape sack after sack after sack and get positive yardage. It likely has Cardale Jones somehow get stopped a bit or look like a guy in his third start. Erick Dargan will have another crazy day where he's around the ball and gets a turnover or two. That's what an Oregon win looks like.

That's also not the most likely scenario. Ohio State's offense, on paper, is an incredibly consistent, efficient killing machine that is better than anything Urban Meyer has ever wielded. Oregon's defense is at best opportunistic, but that only goes so far. Ohio State's defense - particularly their front 7 - is disciplined, fast, and very good at causing havoc and disruption, and will likely pressure Mariota all day. While I expect Oregon to be able to run quite a bit, Oregon will likely have enough failures in the passing game to not score points - and against OSU, I don't see that going well.

The numbers don't see Oregon winning, much less covering. They see a fairly close Ohio State win, with a ton of points by both sides. If you like, you can say that FEI is way better than S+P and will predict an Oregon win. That's possible, but chances are not great for that.

Sadly, Oregon did almost everything they needed to do on defense - get turnovers, win the turnover battle decisively, cause some havoc - and it didn't matter, because Oregon just couldn't score enough points and Mariota couldn't make enough plays. Cardale Jones didn't  come down to earth. Ezekiel Elliott didn't get stopped or even slowed down. The real hero of that game was Joey Bosa.

But hey, the numbers were pretty good in predicting that. Or at least S+P was. FEI was ridiculously wrong, absurdly so, and continues to be pants-off wrong.

What does that mean for this week? Not a lot of good. S+P being more right has a reeling, horrible looking Oregon defense going up against a very good offense that is good in precisely the way that Oregon is bad. Oregon's offense looks incredible and is going up against a suspect defense, but that difference isn't quite as much as the glaring badness of Oregon's defense. Oregon doesn't have major special teams advantages, Oregon is very far on the road, and MSU has been looking forward to this game since last year. Vernon Adams is going to be a good QB for Oregon and Oregon has more firepower than anyone in the NCAA save Ohio State, but it's still Adams' second FBS start, on the road 2000 miles away against a top ten school.

The game started as a push but had money flowing in to MSU, driving the line to MSU being favored by 4 points. I suspect that Oregon doesn't win this game...and sadly, doesn't even cover. MSU by like 10. But I'm not really super sold on that. I'd not be surprised by almost any result save a very low scoring one. There should be a lot of fireworks in this game, and it should be a lot of fun to watch - sort of like the USC-Oregon game in 2012, except with Oregon on the losing side.

If Oregon wins this, I'll say it now: Oregon is going to play in the playoffs barring injury.