VN: Stats vs. Eyeballs After Five Weeks

VN: Stats vs. Eyeballs After Five Weeks
VN: Stats vs. Eyeballs After Five Weeks
Photo: USA Today Sports Images

by Bill Connelly

Through five weeks, the F/+ rankings are a (hopefully delicious) mole recipe with too many ingredients to count. There isn't nearly enough connectivity to let opponent adjustments stand on their own, which means that raw data (for my S&P+ ratings) and preseason projections (for both S&P+ and FEI) are still playing a role in the numbers.

We have a lot of factors propping up the overall rankings, things that won't necessarily be used for the end-of-season rankings. Still, we do a pretty good job of approximating what will become the hierarchy for the rest of the season. The correlation between last season's Week 6 F/+ ratings and the end-of-season ratings was a solid 0.884.

So yeah, we're most of the way there when it comes to figuring out who's good, who's bad, who's really good, etc. But we're not all the way there. Of the top 25 teams in last season's Week 6 rankings, 13 did end up in at least the final top 25, but Florida (No. 48 at the end of the year) was still a top-five team (preseason projections were still playing a role, and injuries hadn't completely taken effect yet) and Washington (eventually No. 36) was still in the top 15. And on the flipside, eventual No. 4 Auburn was still laboring at 48th while eventual No. 6 Michigan State was still hanging out at 17th.

While it's safe to assume that most of the teams the F/+ rankings currently see as very good are indeed just that, we know the season is still waiting to unveil a few twists and turns. So as we approach the season's midpoint, I thought it would be fun to compare stats to eyeballs and note some of the teams I feel are prime candidates to make a move (for better or worse) between now and the end of the season.

(Side note: my eyeballs are almost always wrong.)

Be skeptical

No. 5 Ole Miss

I love, love Ole Miss' defense, and I don't say that simply because I'm terrified of it. The Rebels play angry on that side of the ball; they're second in the country in Defensive S&P+ at the moment, and I think it's justified. Their raw numbers are almost untouchable -- they're one of just three defenses that are in the top 10 in both raw Success Rate (seventh) and IsoPPP (first). No, they haven't played good offenses yet, but the ones they have played certainly looked much worse than normal against the Rebels. Memphis gained 469 yards (5.8 per play) and scored 35 points against UCLA, then gained 104 yards (1.7 per play) and scored three points against Ole Miss. Vanderbilt gained 379 yards (6.9 per play) and scored 34 points against South Carolina but gained 167 (3.3 per play) and scored three points against Ole Miss. So from an opponent-adjusted perspective, the Rebs are faring just fine so far.

Unfortunately, I've also seen Ole Miss' offense play.

Now, it's not that the Rebels' offense is bad; it very much is not. The Rebels are 12th in Offensive S&P and have averaged at least 6.2 yards per play in three of four games, and even though it sometimes takes them a while to get rolling, they have been outstanding in the second half this year. Quarterback Bo Wallace has a 68 percent completion rate, six touchdowns, and six interceptions in the first half this year; he has a 77 percent completion rate, five touchdowns, and no picks in the second half. His fourth-quarter passer rating thus far: 297.1. (Yes, that's in just 14 passes over two games.)

Still, he has six first-half interceptions. He looked like a nightmare for most of three quarters against Boise State, and while he was fine late, his passer rating ws just 122.8 (60 percent, two touchdowns, two interceptions) last week against Memphis. When he finds his rhythm, he's outstanding. But when he loses it, it takes him a while to find it again.

Oh yeah, and Ole Miss is currently 89th in Rushing S&P.

So what do you figure the odds are of Wallace catching fire and Ole Miss living up to what are some pretty lofty projections against Alabama this weekend? Yeah, same here.

If Wallace looks good -- good enough to upset the Tide or come close -- then the sky is the limit for Hugh Freeze's Rebels this year. But expecting a one-dimensional offense to disturb Alabama too much is ... usually an expectation that isn't met. Barring an epic run of injuries, Ole Miss isn't going to pull a Florida-esque fall. I certainly see the Rebels as a top-15 or top-20 team. But top-five status (and "projected to beat Bama" status) makes me nervous.

No. 14 Louisville

S&P+ is typically quite volatile at this time of year (my own ratings are why Ole Miss is so damn high, too) and one team benefiting from that is Louisville, which has had, from a raw data standpoint, one of the most successful defenses in the country so far. The Cardinals rank fourth in Defensive S&P+, and while that wouldn't be the biggest reach in the world -- they ranked 10th in Defensive F/+ last year, after all, and new defensive coordinator Todd Grantham did have some success at Georgia before a 2013 fade -- they have gotten what are probably a couple of artificial boosts.

First, Louisville has played Wake Forest. As long as raw data plays any role, even a minor one, in the ratings, playing Wake Forest is going to be good for you. The Ville held the Deacs to 100 total yards (1.8 per play) last Saturday, and while that's certainly not bad ... well, UL-Monroe held Wake to 94 yards (1.9 per play) in the season opener. Wake's offense is just about as young as an offense can possibly be, so it makes sense that the Deacs are struggling, but Louisville will likely sink a bit in the ratings when opponent adjustments are taken fully into account.

Louisville also gets a boost from having played Miami in Week 1. The Hurricanes started a true freshman, Brad Kaaya, at quarterback; while he's learning quickly (he has produced at least a 133.3 passer rating in each of the last four games, looked outstanding against Arkansas State, and displayed nice, mature control in a win over Duke), he still looked like a first-time starting quarterback against Lousiville. If the Cardinals were to face Kaaya again in the ACC title game (they won't, but go with me), I would expect him to perform quite a bit better than 17-for-29 for 174 yards, a touchdown, and two picks.

The Louisville offense is a work in progress in Bobby Petrino's first season back in the land of Papa John's. The Cardinals are showing quality big-play potential (27th in IsoPPP) and absolutely no efficiency whatsoever (102nd in Success Rate). When their defensive numbers come back down to earth a bit, F/+ probably won't see them as a top-15 team anymore.

No. 28 Kansas State

Both S&P+ and FEI are zeroing in on KSU as a team that ranks between about 25th and 30th. Usually when they agree, there's reason for it. Still, the Wildcats' 2014 performance strikes me far more as that of a top-20 team. The offense has thoroughly dismantled bad defenses -- 8.1 yards per play against UTEP (despite basically taking the fourth quarter off), 6.7 yards per play against Iowa State -- and despite replacing quite a few pieces from last year, the defense has yet to allow 5.0 yards per play in a game and shut down Auburn's running game in a way I didn't think was possible.

[ad placeholder 3]

The Wildcats are missing a second strong receiving threat to take pressure off of star Tyler Lockett (who has been battling a nagging hamstring issue but still had 10 combined catches for 220 yards against Iowa State and UTEP), and their running backs are very much just so-so. That's not going to change. But the offense is showing quality efficiency, and the defense looks excellent, and while there were some self-destructive tendencies against Auburn ... well, Auburn's off the schedule now. Trips to Oklahoma and Baylor probably aren't going to go well, but I still trust Kansas State to handle just about anybody on the schedule that isn't currently in the F/+ top 10.

No. 34 Boise State

I'm not completely sure why S&P+ likes Boise State. The defense is certainly pretty good, allowing just 4.2 yards per play in the last three weeks. And the offense has at least had its moments, torching a decent Colorado State defense. But they also committed seven turnovers against Air Force. Quarterback Grant Hedrick threw four interceptions, backup Ryan Finley threw a fifth, and Jay Ajayi lost two fumbles, as Jay Ajayi tends to do. Hedrick now has a five-to-nine TD-to-INT ratio. That's ... suboptimal. Congrats on the 70 percent completion rate, but that's negated a bit when one of every six incompletions is an interception.

One of six!

Sounds about right

Florida State

I've been getting yelled at by Florida State fans for being unfair and having expectations of the Noles that were just too high heading into the year. But they have disappointed me this year; projected No. 1 by both the F/+ rankings and my own brain, they have been sleepy on offense, inconsistent up front, and glitchier than expected on defense.

They've got enough talent that they've survived without a loss, and there's a pretty good chance that they eventually find fourth and fifth gear after spending quite a bit of September in second and third. But until they hit those higher gears, we don't know for sure that they will. For now, I have no problem with them being outside of the F/+ top four (even though No. 2 Oregon will now drop after Thursday's loss and No. 5 Ole Miss makes me nervous).

Mississippi State

I regret saying this already, but while Ole Miss strikes me as incredibly unstable, No. 7 Mississippi State looks the part to me. Like their in-state mates, Mississippi State's quarterback also makes me nervous -- three or four times per game, Dak Prescott will make some crazy video game play that either works beautifully or fails miserably (he ducked out of about 17 tackles and threw a touchdown pass against LSU; he also lost a fumble that was returned for a touchdown), and the Bulldogs' fate is tied to the fate of those instances.

Still, the defense is mostly sturdy (barring about 10 or 15 minutes against Alabama-Birmingham and LSU), and wacky plays or not, A) the offense has averaged at least 7.1 yards per play in three of four games, and B) Prescott has only thrown two interceptions in 96 passes. That's not bad for a gunslinger. (It's also a little bit lucky, of course. Opponents have picked off two of 22 defensed passes, and averages suggest that should result in closer to four or five picks.)

My eyeballs might be misleading me on this one, but I loved what I saw out of the Bulldogs against LSU, and if they can survive Texas A&M this weekend without suspended starting center Dillon Day, they could make noise in the West. Lord knows they've got the schedule for it.

[ad placeholder 4]

This week at SB Nation

Michigan's upcoming coaching search is perilous, but necessary

The Numerical, Week 5: We're still waiting for excellence, Florida State
Week 5 advanced box scores
Missouri 21, South Carolina 20: Beyond the box score

Projecting the SEC's 2014 conference race using advanced stats
Updated Big Ten football projections

Is Oklahoma-TCU or Baylor-Texas more likely to be an upset?
Week 6 F/+ picks

Arizona's Oregon upset again proves Rich Rodriguez is a master builder


1 comment, Last at 03 Oct 2014, 7:28pm