Beware: Weirdness afoot.
Tennessee opened as a 10.5-point favorite over the UAB Blazers this week, and since then it’s bounced around between 9.5 and 12.5. Currently, it’s Vols -12. So . . . will the Vols cover against UAB?
The SPM’s performance so far in 2019
After three straight weeks of being over 55% (and notching nearly 64% last week), the SPM fell victim to the turf monster last week and went 25-30 (45.45%) overall. Over our confidence threshold, it went 9-11 (45%), and in our pre-defined favorites range, it went 6-5 (54.55%). This week, we tracked both Vegas and non-Vegas lines, and essentially, there was only a one-game difference between those two sets of data.
For the season, the SPM now sits at 223-217 (50.68%) overall, 99-80 (55.31%) over the confidence threshold, and 55-32 (63.22%) in our favorites range.
I’m officially now regretting my early-season decision to include the overall SPM stuff with the weekly Tennessee statsy preview because the SPM is TERRIBLE at Vols games. It again got the Tennessee game wrong last week, the third week in a row it’s whiffed on the Vols. And it’s going to make itself look downright stupid this week. I consoling myself with constant reminders that the Vols have everybody else confused, too, including Vegas.
So what does the SPM have to say about the Tennessee-UAB game this week? Ahem . . .
Okay, first, the SPM’s opinion on the Vols-Blazers game this weekend should be mostly ignored. The only reason I’m even posting this is (1) in the interest of full disclosure and to avoid allegations of cherry-picking, and (2) because of that word “mostly.”
Without going through the full analysis, I’m just going to tell you that the SPM has UAB winning this game by 16.2 points. This actually provides an excellent illustration of why you can’t just assume that the machine gets more reliable the more it diverges from the Vegas line. The truth is that there is a sweet spot. The results generally get more reliable the more they diverge from the Vegas line, but only up to a point. If they diverge too much, it’s more an indication that it is not reliable because something is very, very wrong.
So what’s wrong? This is a whole different neighborhood, man.
Those of you who’ve been reading for some time know that the SPM is a comps system. While other systems rank teams based on how good they are and then make predictions based on those relative rankings, the SPM uses comparables to make predictions, and the comps are just no good this week. To UAB, the Vols defense looks most like Rice and Old Dominion and the offense looks like Western Kentucky and UTSA. To drive the point home a bit more, from Tennessee’s perspective, UAB’s defense looks like Florida and Alabama and its offense looks like South Carolina and Florida.
As odd as that looks, there’s good reason for it. UAB’s stats put them right alongside those teams in the national rankings. The problem is that those stats have been compiled against nothing but terrible competition. As I said in that earlier post, Tennessee strength of schedule is currently ranked 12th, and UAB’s is ranked 130th. And as @Alyas Grey said in this terrific comment, wins and losses aren’t the only issues. You can’t compare Manhattan commercial real estate to a duplex in Podunk, Alabama. Well, you can, but you’ll end up with UAB winning by 16 points.
Configuring the SPM to account for that will have to be an offseason project. For now, we’ll just have to mostly ignore it on those rare occasions when it spits out something funny.
That word “mostly”
But I say “mostly ignore” because all of that might also be hinting at something we’re all missing. There are two sides to this problem, and our tendency is to intensely scrutinize only one side of that coin while ignoring the other.
If you don’t know how good a team is because they haven’t played anybody, you can’t just conclude that they are no good. If you don’t know how good they are, you don’t know how good they are. They could be worse than they look, but they could also be better than you’re giving them credit for.
The Blazers’ 6-1 record so far in 2019 is not unexpected. After finishing our magazine this spring, I wondered whether the Vols’ toughest non-conference opponent would be BYU or UAB. So, while we’re dismissing their schedule, we’d do well to give some weight to their resume.
SPM Final Estimates
For the record (and in case you skipped everything else, I’m putting it here: This is likely to be really wrong):
SPM Final estimated score: UAB 30, Tennessee 13.8
SPM Final estimated spread: UAB -16.2
SPM Confidence level (difference from the opening spread): 26.7
One more time with feeling: That 26.7 “confidence” level is so high that it signals serious wonkiness, that you should not be confident in it. Yes, I should rename it something other than “confidence level.”
What do we do about all of this? The Blazers haven’t ventured out of Podunk so far this year, but they did head to the big city once last year when they played Texas A&M at College Station. Despite finishing the 2018 season 11-3, they lost to the Aggies 41-20. The SPM says that if those two teams played today, the score would be Blazers 25.1, Aggies 24.8. Occam says that’s wrong by three touchdowns.
If the SPM is similarly wrong about the Vols-Blazers, that would make it Vols by 5. Combine that with the fact that the SPM has been wrong on the Vols for three weeks in a row now, presumably due to the fact that it doesn’t recognize that they’re improving every week, and I’d bump that up another touchdown. Let’s say Vols by 12, maybe.
So, my eyeball-adjusted prediction is Tennessee 28, UAB 17. I think the Vols cover the opening spread of Tennessee -10.5, but I don’t think they cover the current spread of Tennessee -12.
Other predictions from other systems
With the current spread at Tennessee -12 and an over/under of 48, that translates to something like Tennessee 30, UAB 18.
ESPN’s FPI gives the Vols an 88.3% chance of winning.
Bill Connelly’s SP+ likes the Vols 27-18, and gives them a 70% chance of winning. That’s picking the Vols not to cover.
SP+ broke even last week. According to Bill’s Twitter timeline, SP+ went 27-27-1 (50%). It’s still at an impressive 55% for the season.
As I said earlier, our SPM is now 223-217 (50.68%) on the season and 55-32 (63.22%) in our favored range.
The SPM does not like the Vols to cover this week but lacks credibility on the Vols generally right now and in this game in particular. Using my own eyes, I like the Vols by 11, which means I think they’ll cover the opening spread of -10.5 but not the current spread of -12.
What do y’all think?