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. I've done this a couple times already (the latest was a look at key week two matchups), and this specific article represents a different take on this analysis.
Where the previous articles focused more on game by game projections, with the ultimate win total as a bottom-line sum, this article will instead focus solely on the win total projections and not at all on specific matchup projections.
Let's start with UCLA. As you can see in the below table, the original 5Dimes line for UCLA was 6.5, while the original Compu-Picks projection was 8.33. This meant that Compu-Picks was substantially more optimistic about UCLA's chances in 2012 than the books. You can then see the additional impact of UCLA's two wins, bumping the projection from 8.33 to 8.76 and then to 9.66 (the larger bump was because the Nebraska game was originally considered much more in doubt than Rice, so it had a more meaningful impact).
In addition, you can see the impact of Rice's win over Kansas, further bumping UCLA's projection, though only a little (and the reverse happens given ASU's win over Illinois, since that result makes ASU more likely to be good enough to beat UCLA). This is partially because how well an opponent does in a given game is less meaningful than how you do, and it's also partially because with two results in the books, there's much less variation left in the projections.
To illustrate this, consider that before the season started, 0-12 and 12-0 were both in play (though of course both were very unlikely). With two wins in the books, 0-12 is obviously off the table, and similarly, disastrous 2, 3, or 4 win seasons are almost certainly off the table as well. The reverse would have been true had UCLA lost their first two games, and had they split, it would have been a decent indicator that a more middling type season was most likely.
While of course anything can happen, the fact remains that we've learned a good deal about this team in two weeks, and it makes sense that projections start to coalesce.
| 5Dimes Line for UCLA || 6.50 |
| Original Compu-Picks Projection || 8.33 |
| Given Win over Rice || 8.76 |
| Given Win over Nebraska || 9.66 |
| Given Rice Win over Kansas || 9.74 |
| Given ASU Win over Illinois || 9.67 |
A team that moved in the opposite direction is Oklahoma St. Here, Compu-Picks was less optimistic than the 5Dimes line, but not by terribly much. Their AA win very slightly boosted their win projection (since it was a virtual lock as a win, it really didn't tell us much that they indeed matched expectations), but the Arizona loss drastically dropped the projection. There's a mild bump from the fact that Arizona beat Toledo (the better Arizona does, the more likely it is that OK St simply lost to a good team, which is less damaging to their future prospects), and a material drop from the fact that Iowa St beat Iowa (since that makes the Cyclones more likely to be a dangerous threat).
| 5Dimes Line for Oklahoma St. || 8.50 |
| Original Compu-Picks Projection || 8.03 |
| Given Win over AA || 8.05 |
| Given Loss to Arizona || 5.61 |
| Given Arizona Win over Toledo || 5.79 |
| Given Iowa St Win over Iowa || 5.46 |
One team that has defied Compu-Picks' projections is Arizona. While 5Dimes had the Wildcats at 5.5 wins, giving them a solid shot at bowl eligibility, Compu-Picks was substantially more pessimistic, showing a 3.64 win expectation. As seen in the previous week two article, the Toledo win boosted the Wildcats' win expectation by about a full win (partially because a loss was considered possible, partially because the win boosted projections for future games).
Beating Oklahoma St, meanwhile, boosted the win expecation by nearly two full games. This is because it was considered a pretty unlikely outcome; in this case, the direct impact was over 0.8, and then the rest came from boosting the win probabilities in the games still on the schedule.
Looking at the other two entries on the table, as we've seen in the above examples, it's difficult for other teams' results (either previous opponents or future opponents) to substantially move the dial much compared to Arizona's own performances, though it does have a minor impact.
| 5Dimes Line for Arizona || 5.50 |
| Original Compu-Picks Projection || 3.64 |
| Given Win over Toledo || 4.67 |
| Given Win over Oklahoma St || 6.43 |
| Given Toledo Win over Wyoming || 6.44 |
| Given Oregon St Win over Wisconsin || 6.38 |
Colorado St is an interesting case. One step forward, two steps back. They were given a 6.5 win total by 5Dimes, and Compu-Picks was solidly under that projection, a bit less than 5. Beating Colorado gave them a nice boost, bumping the win expecation to nearly 6.2, but they gave it all back and then some by gagging to a AA opponent. And when Colorado also gagged to a AA opponent, it made Colorado St's week one upset worth a lot less, since it strongly suggested that Colorado is simply a terrible football team. Given all of that data, factoring in Wyoming's loss to Toledo didn't really move the dial too much for the Rams' expectations.
| 5Dimes Line for Colorado St. || 6.50 |
| Original Compu-Picks Projection || 4.87 |
| Given Win over Colorado || 6.18 |
| Given Loss to AA || 4.15 |
| Given Colorado Loss to AA || 3.94 |
| Given Wyoming Loss to Toledo || 3.97 |
Cincy is a team where opponents' results have a bit more of an impact; this is both because Pitt's AA loss was so majorly unlikely that is drastically changed the situation and because Cincy only has had one game to begin with, their win over Pitt. Once the Panthers made it fairly clearly that they were just a bad football team by gagging to a AA opponent, Cincy's win suddenly became not worth very much (of course, Pitt's loss came before Cincy's win, but I'm trying to show teams' actual results first, and opponents' results second, to create a consistency of process).
| 5Dimes Line for Cincinnati || 7.50 |
| Original Compu-Picks Projection || 7.89 |
| Given Win over Pitt || 8.97 |
| Given Pitt Loss to AA || 8.21 |
| Given Uconn Loss to NC St || 8.35 |
Ohio is a team that Compu-Picks started out pessimistic on, but once they scored the win over Penn St, Compu-Picks' projections taking into account that win (not the actual stats or margin, just the fact that they won) puts it right around the original 5Dimes win total.
| 5Dimes Line for Ohio || 9.50 |
| Original Compu-Picks Projection || 8.20 |
| Given Win over Penn St || 9.51 |
| Given Win over NM St || 9.57 |
| Given Penn St Loss at Virginia || 9.52 |
| Given Bowling Green Win over Idaho || 9.51 |
Another team that surprised Compu-Picks is Northwestern. Originally projected to be a 5-win team, Northwestern scored key wins over Syracuse and Vanderbilt, bumping their win projection to nearly 8. Of course,
| 5Dimes Line for Northwestern || 6.50 |
| Original Compu-Picks Projection || 5.15 |
| Given Win Over Syracuse || 6.32 |
| Given Win over Vanderbilt || 7.84 |
| Given Vanderbilt Loss to South Carolina || 7.81 |
| Given Nebraska Loss to UCLA || 7.82 |
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 firstname.lastname@example.org