Vegas vs. FPI: Week 1 disagreements
Seth Walder
ESPN PLUS ($ MATERIAL)Seth Walder
Las Vegas and FPI both try to predict the future -- often in lockstep, occasionally at odds.
In this weekly article, we'll examine the latter scenario. The instances when, for one reason or another, our predictive metric for college football has a substantially different view on the game than the betting market does.
The parameters: We're looking at four games per week where the Vegas line differs by five or more points from FPI's predicted points margin. That's the computer element. Then I come in and try not to screw it up by picking the best bet in those four games.
What you do with this information is up to you. Just reading because you're curious? Cool. Plan on driving through New Jersey and making your wife pull over on the side of the road before crossing out of the state -- that's definitely a hypothetical, certainly not me last weekend -- so you can open up a particular app? Do as you wish. But you should know that FPI was not specifically designed to beat the spread. In the interest of record keeping, we will note that, if measuring against the spread, this article last year would have gone 23-21.
Now, to 2018 and Week 1.
Note: All odds are via Westgate Las Vegas SuperBook as of Aug. 29.
Michigan Wolverines at Notre Dame Fighting Irish (PK)In this weekly article, we'll examine the latter scenario. The instances when, for one reason or another, our predictive metric for college football has a substantially different view on the game than the betting market does.
The parameters: We're looking at four games per week where the Vegas line differs by five or more points from FPI's predicted points margin. That's the computer element. Then I come in and try not to screw it up by picking the best bet in those four games.
What you do with this information is up to you. Just reading because you're curious? Cool. Plan on driving through New Jersey and making your wife pull over on the side of the road before crossing out of the state -- that's definitely a hypothetical, certainly not me last weekend -- so you can open up a particular app? Do as you wish. But you should know that FPI was not specifically designed to beat the spread. In the interest of record keeping, we will note that, if measuring against the spread, this article last year would have gone 23-21.
Now, to 2018 and Week 1.
Note: All odds are via Westgate Las Vegas SuperBook as of Aug. 29.
FPI: Notre Dame by 6.5 points
Game: Saturday at 7:30 p.m. ET
FPI may be a kindergarten-aged system, but it is riding for Notre Dame this season like it's already a grandparent. In the very beginning, the books were on board, too. According to VegasInsider.com's consensus lines, this game opened Michigan -1.5 and hit Michigan -2.5 before the pendulum started to swing toward South Bend, especially after Wolverines receiver Tarik Black broke his foot over the weekend.
There are two reasonable critical takes on this pick:
1. FPI knows that the Notre Dame offense lost some talent, but it doesn't precisely know that two of the missing starters are Quenton Nelson and Mike McGlinchey, who both went in the top 10 of the NFL draft.
2. FPI knows Michigan has brought in a transfer quarterback, but it might not realize how much of an upgrade Shea Patterson is over John O'Korn & Co.
Does that all equal a 6.5 gap against the spread?
Odds are, no. First, FPI has shown over the long term that losing one or two star players doesn't dramatically alter the mean expectation for a team. First-round picks are removed from college football without fail every season, and it does not in general trip the system up. As for quarterback, it is the most important position on the field, but the reason Michigan struggled so much on offense last season was not only because of the play under center. Should the Wolverines have added an other-worldly talent at quarterback, it would be worth changing our view of them, but Patterson ranked 60th in Total QBR last season. That's not nearly good enough to tear up FPI. But Notre Dame is definitely the favored team here.
FPI lean: Notre Dame
FPI: West Virginia by 3.1 points
Game: Saturday at 3:30 p.m. ET in Charlotte
Will Grier may be great. Will Grier may be a Heisman candidate. But Will Grier does not play defense.
Though last year we debunked the idea that Big 12's defense on the whole is a complete atrocity, it's still not that great. West Virginia -- who some people (but not FPI) consider to be a or the main threat to Oklahoma in the conference -- has the third-worst defense in the Big 12. That group (57th in defensive efficiency, our play-level descriptive statistic that adjusts for opponent and garbage time) is only bringing back five starters.
Grier played most of last season for West Virginia. Sure, he threw up some monster raw numbers, but the Mountaineers only ranked 27th in offensive efficiency. It kind of feels like that should have been higher given the coronation that Grier has received this offseason. The West Virginia quarterback is talented, but let's not forget he finished behind John Wolford and Brandon Wimbush in Total QBR last season.
In other words, West Virginia does not deserve this much respect on a neutral field against an SEC team (unless that SEC team is Vanderbilt).
FPI lean: Tennessee +9.5
FPI: UCF by 15.5 points
Game: Thursday at 7 p.m. ETFPI ranks the Huskies as the 120th best team in the FBS. That's probably not where the disagreement on this line lies.
For the exact same reason that FPI likes Memphis and not UCF to win the AAC, it's down on the Knights in their first game. Yes, UCF went undefeated. Yes, McKenzie Milton is back. But this is not one of college football's great powerhouse programs. Two seasons ago, it went winless. We have to expect regression.
If our rating of UConn is correct, a 23.5-point spread would require UCF being of the same quality as teams like South Carolina, Missouri or TCU. Last season, it was better than teams like that. But there's little reason to expect, with a new coach and plenty of new faces starting, that the Knights really ought to be at that level.
FPI lean: UConn +23.5
FPI: Duke by 22.4 points
Game: Friday at 7 p.m. ETA season ago the difference between these two teams was not significant, but they are moving in opposite directions. In 2017, the Blue Devils ranked 41st in total efficiency while Army ranked 57th. Army was fueled by its offense, and Duke by its defense.
But while the Blue Devils are returning eight starters from its superior side of the ball, just three offensive starters are returning for the Black Knights. That's a major difference that is the key driver of why FPI considers these two squads now worlds apart. Add in better recruiting from Duke over the past four seasons and it's hard to imagine Army replacing its lost starters with near the talent that Duke will for its far fewer open spots.
FPI lean: Duke -14
