Compu-Picks 2012 Analysis: Week Three Games

Mr Pac Ten
Posted Sep 13, 2012

2012 Compu-Picks Analysis: What results to date can tell us about key week three games

One new piece of analysis I'm doing this year is looking at some of key matchups and what Compu-Picks thinks the results of these matchups will say about the rest of the season. For previous weeks I've looked at what key games can tell us about future win expectations, but this time I'm going to take a different tack in my analysis.

With two weeks of results in the books, we've learned a fair amount about a number of teams. Given what we've learned, how should we adjust our projections? This analysis attempts to do that, adjusting the original preseason projections given win/loss information (as in previous articles, ONLY wins and losses are considered; in-game information such as margin or other statistics are not considered).

First up is Florida at Tennessee. I start with the preseason spread (as can be seen here) and current spread for comparison purposes, then look at the original Compu-Picks preseason projection of Florida's probability of winning this game. I then show the adjusted probability given each successive piece of information. As you can see, Tennessee's win over NC St and Florida's win over Texas A&M were both important data points, each shifting the win probability by nearly 4%. The various other data points moved things a bit, but none by very much.

Of course, this doesn't take into account Florida's struggles against both Bowling Green and Texas A&M, so the final 79% number is almost certainly overstated, but it's hard to see it being overstated by so much that Tennessee should be a favorite, much less a full field goal favorite. Tennessee hasn't really proven anything yet, beating up a (projected to be) mediocre NC St team (and it's not like squeaking by UConn is impressive for the Pack) and a AA opponent, so there doesn't seem to be much reason for their power rating to move very much.

And on the flip side, Florida looked quite poor against Bowling Green, but it's hardly unusual for a power team to not really show against a weaker opponent. And while they certainly struggled at Texas A&M, they still won a key game on the road, which is a decent achievement until proven otherwise (that's A&M's only result so far, so we can't say much about them yet). So it doesn't seem that Florida's power rating should really move very much either, which is consistent with Compu-Picks' calculations.

In other words, Florida +3 looks like a very good pick given what we've seen from each team so far. Of course, this doesn't factor in the Jelani Jenkins injury, but it's hard to see his absence as worth a field goal, much less enough to make the Vols the projected winner, much less enough to suggest it should be a full field goal line.

One final note: I've also factored in the results from Florida and Tennessee's opponents to date. Since this is based on 5,001 simulations, and the effects of opponent results tends to be small (barring major upsets shaking up the numbers), there is noise in these adjustments. NC St beating UConn should have lowered Florida's odds of winning (since it helps Tennessee's rating); showing a small bump instead is almost certainly system noise.

Preseason Line for Florida at Tennessee -5.0
Current Line for Florida at Tennessee 3.0
Original Compu-Picks Probability of Florida win over Tennessee 78.14%
Probability Given Florida Win over Bowling Green 79.15%
Probability Given Tennessee Win over N. Carolina St. 75.48%
Probability Given Florida Win over Texas A&M 79.24%
Probability Given Tennessee Win over AA 79.15%
Probability Given Bowling Green Win over Idaho 79.46%
Probability Given N. Carolina St. Win over Connecticut 79.51%

Next up is the other key game of the weekend, Notre Dame at Michigan St. The same types of calculations occur here as we saw for Florida-Tennessee. Here again the system disagrees with the line movement compared to preseason; while it agrees that Michigan St has achieved a bit more, and their odds have improved, it's not by that much. A six point line generally represents more than a 60% chance of winning, typically on the order of a 2/3 chance. So Compu-Picks thinks there's some, though not huge, value here. And given that Michigan St had about as tough a time with Boise as Notre Dame did with Purdue (and they both blew out their other opponent), there's nothing in the in-game results to suggest that the 42.27% estimate has missed anything important.

Preseason Line for Notre Dame at Michigan St. 2.5
Current Line for Notre Dame at Michigan St. 6.0
Original Compu-Picks Probability of Notre Dame win over Michigan St. 46.29%
Probability Given Notre Dame Win over Navy 48.89%
Probability Given Michigan St. Win over Boise St. 42.48%
Probability Given Notre Dame Win over Purdue 45.48%
Probability Given Michigan St. Win over C. Michigan 42.81%
Probability Given Purdue Win over AA 42.53%
Probability Given C. Michigan Win over AA 42.27%

Thursday night's game between Rutgers and USF could be interesting, though the books don't seem to think so, with a 7.5 point line (and it was 10 earlier in the week). Have the books seen something Compu-Picks missed?

It's certainly possible, though if anything what Compu-Picks has missed (USF's lucky escape at Nevada) points more towards Rutgers than the simple W/L records suggest. Of course, it could be that Nevada is actually good (Compu-Picks doesn't think so), but even if that's true, 7.5 points is a lot for a game that Compu-Picks pegged preseason as a tossup and hasn't seen reason to adjust the projection very much.

Preseason Line for Rutgers at South Florida n/a
Current Line for Rutgers at South Florida 7.5
Original Compu-Picks Probability of Rutgers win over South Florida 51.09%
Probability Given Rutgers Win over Tulane 53.61%
Probability Given South Florida Win over AA 53.15%
Probability Given Rutgers Win over AA 53.42%
Probability Given South Florida Win over Nevada 50.68%
Probability Given Tulsa Win over Tulane 50.51%
Probability Given Nevada Win over California 47.21%

A game that's lost a little of its luster is USC-Stanford. As was the case with Notre Dame - Michigan St, both teams had a blowout win and a struggle, though obviously Stanford's struggle was FAR closer than USC's. The system only considers W/L, so this isn't factored into its numbers, though obviously we can do while interpreting its projections.

It's interesting that the line has moved towards Stanford despite the extreme struggles in week one. The real question here, though, is whether you believe the system's preseason take (solid lean to Stanford) or Vegas's preseason take (USC as a double-digit favorite). There's been enough up and down for both to think that preseason pegs for this game shouldn't much change, so it's really all about how you called it. Personally, I'd lean USC in this one, and actually suspect a blowout, but we'll see. It's obvious what Compu-Picks thinks.

Preseason Line for USC at Stanford -11.5
Current Line for USC at Stanford -8.5
Original Compu-Picks Probability of USC win over Stanford 40.21%
Probability Given USC Win over Hawaii 40.85%
Probability Given Stanford Win over San Jose St. 40.62%
Probability Given USC Win over Syracuse 42.35%
Probability Given Stanford Win over Duke 40.75%
Probability Given San Jose St. Win over AA 39.77%
Probability Given Northwestern Win over Syracuse 39.07%
Probability Given Duke Win over FIU 39.17%

A game that's gotten a bit more interesting is ASU-Missouri. Preseason, Compu-Picks projected this as an ugly blowout. Now, after ASU beat Illinois and Mizzou lost to Georgia, it's a bit closer, though still projected as Mizzou winning pretty easily.

Here again, it's worth noting that neither ASU's win nor Mizzou's loss were close, though Mizzou was very much in it until the awful fake punt run on 4th and 11 swung momentum and things got out of hand. So it's reasonable to think that this number is too low. That said, Compu-Picks thinks Mizzou should be a much more solid favorite, and I'm somewhat inclined to agree.

Preseason Line for Arizona St. at Missouri n/a
Current Line for Arizona St. at Missouri 6.5
Original Compu-Picks Probability of Arizona St. win over Missouri 4.64%
Probability Given Arizona St. Win over AA 5.29%
Probability Given Missouri Win over AA 5.22%
Probability Given Arizona St. Win over Illinois 7.02%
Probability Given Missouri Loss to Georgia 8.56%
Probability Given Illinois Win over W. Michigan 9.34%
Probability Given Georgia Win over Buffalo 9.28%

Alabama at Arkansas had originally been considered a key game, but ULM put that thought to bed. Bama started out a mild favorite, but after last week's key upset and the injury to Hogs QB Wilson, the line has moved enormously. Is this fair? Compu-Picks says yes. While it's tougher to accurate translate high spreads to win probabilities (and there's no $ line out yet), the system definitely agrees with making a major adjustment to the number based on what we've seen so far. In fact, you can make a reasonable case that the line is still a bit too light, especially if Wilson is in fact out.

Preseason Line for Alabama at Arkansas -5.0
Current Line for Alabama at Arkansas -20.5
Original Compu-Picks Probability of Alabama win over Arkansas 79.78%
Probability Given Alabama Win over Michigan 84.76%
Probability Given Arkansas Win over AA 84.71%
Probability Given Alabama Win over W. Kentucky 84.72%
Probability Given Arkansas Loss to La.-Monroe 93.51%
Probability Given W. Kentucky Win over AA 93.68%

Another game that no longer seems as compelling as it did preseason is BYU-Utah. Without a preseason line, we can't analyze line movement, though it's fairly safe to say that Utah's loss to Utah St has swung this line by a lot, a take Compu-Picks agrees with. In fact, calling it 67% odds for BYU suggests that 4 points is too low of a line.

That said, having actually watched last week's Utah-Utah St game, the adjustment may well be an overreaction. That was a weird game, with an atrocious game plan by Utah, a fairly fluky blocked punt by Utah St, some unlucky fumble bounces early on (depriving Utah of some potentially huge momentum changers), and a blown call on the last play of the game (missing pass interference on 4th down in overtime). So for me personally, this would be a stayaway, though Compu-Picks would suggest otherwise.

Preseason Line for BYU at Utah n/a
Current Line for BYU at Utah -4.0
Original Compu-Picks Probability of BYU win over Utah 45.03%
Probability Given BYU Win over Washington St. 46.53%
Probability Given Utah Win over AA 46.43%
Probability Given BYU Win over AA 46.50%
Probability Given Utah Loss to Utah St. 66.46%
Probability Given Utah St. Win over AA 66.11%
Probability Given AA Win over Washington St. 67.61%

This is one of those "full disclosure" type games. Preseason, Compu-Picks thought UNC would be a train wreck, and this table definitely reflects this projection. In fact, they've played better than what their record shows, blasting their AA opponent and losing a squeaker to Wake. That said, 3 points still seems very light given both the preseason projections and the fact that even if it was a good game, UNC still did lose to a pretty mediocre Wake squad.

Preseason Line for North Carolina at Louisville n/a
Current Line for North Carolina at Louisville 3.0
Original Compu-Picks Probability of North Carolina win over Louisville 12.76%
Probability Given North Carolina Win over AA 13.95%
Probability Given Louisville Win over Kentucky 11.48%
Probability Given North Carolina Loss to Wake Forest 7.81%
Probability Given Louisville Win over AA 7.55%
Probability Given Wake Forest Win over AA 7.63%
Probability Given Kentucky Win over Kent St. 6.89%

When you talk about a game losing its luster, even Bama - Arkansas may not hit the level that Virginia Tech - Pitt has sunk to. While there are reasons to think that Pitt isn't really as bad as they've looked, they're 0-2 including a AA loss, neither game was anything like a nailbiter, and there really doesn't seem to be any strong reason to think this shouldn't be an ugly blowout. Based on Compu-Picks' numbers, 10 points seems way too low.

Preseason Line for Virginia Tech at Pittsburgh n/a
Current Line for Virginia Tech at Pittsburgh -10.0
Original Compu-Picks Probability of Virginia Tech win over Pittsburgh 63.31%
Probability Given Virginia Tech Win over Georgia Tech 69.80%
Probability Given Pittsburgh Loss to AA 95.24%
Probability Given Virginia Tech Win over AA 95.16%
Probability Given Pittsburgh Loss to Cincinnati 96.36%
Probability Given Georgia Tech Win over AA 96.30%

There are a few important notes and caveats I need to make about this model:

1) Compu-Picks does not endorse implicitly or explicitly any form of illegal gambling. Compu-Picks is intended to be used for entertainment purposes only.

2) No guarantee or warranty is offered or implied by Compu-Picks for any information provided and/or predictions made.

3) This preseason model is primarily based on the main compu-picks model. Essentially, it attempts to predict how well a team will rate given its rating history, as well as a number of other data points, such as returning starters, draft talent lost, turnovers, recruiting, etc. This means, among other things, that the rankings are power rankings based on how good a team projects to be, as opposed to a more cynical (though accurate) model that attempts to project how the BCS will rank a team by making adjustments to favor those with easy schedules and punish those with tough schedules.

2012 Compu-Picks Blog

Questions, comments or suggestions? Email me at

Follow cfn_ms on Twitter

Related Stories
Picking the Pac: Week Three
 -by  Sep 12, 2012
Notebook: Is Alabama's OL tipping plays?
 -by  Sep 12, 2012
Saturday Suggests Offense vs. Defense
 -by  Sep 13, 2012