Week 10 Recap:
We dropped both of our plays last week, in our lowest volume week of the year. Fresno State (-12.5), our Big Play, was blown out by a Wyoming team that we rated as one of the worst in the country. Fresno wasn’t just unlucky either; they drop 10 spots to #86 in the ratings, and Wyoming rises 12 spots to #103. If the game were to be played again this week, we would still make Fresno a favorite, but only by about nine points. Our Other Play on Texas A&M also lost, and also was nowhere near covering. A&M was without its starting QB, and that a big impact on the game, one which we did not consider in our ratings. More about QB injuries and adjusting for it below.
Though we only had two official plays, we did have six leans last week, which went 3-3. Temple (+7.5) got the outright win by 10 over East Carolina, despite being dominated statistically. Temple was very lucky to recover 8 of 9 fumbles in the game. Our other two winners were easy: Georgia Tech (-3.5) won by 25 over Virginia, and Louisiana Tech (-7) blew out Western Kentucky, 59-10. We took a big underdog in Kansas (+36), but even 36 points was not enough, as Baylor won by 46. Indiana (+7), down to a true-freshman third-string QB, was putrid on offense in a 24-point loss at Michigan. And lastly, UNLV (Pick) dominated New Mexico statistically, but came up three points short, missing a field goal at the end of regulation.
Week 10 Closing Line Value:
Not as good as recent weeks, with one of our two official plays, and two of our six leans moving against us. Still, we averaged more than 1.5 points of line value on our two official plays, and one-sixth of a point of value on our leans.
- Fresno State: picked -12.5, closed -16.25, CLV +4.25
- Texas A&M: picked +33, closed -32, CLV -1.0
- *Temple: picked +7.5, closed +10, CLV -2.5
- *Kansas: picked +36, closed +34.5, CLV +1.5
- *Louisiana Tech: picked -7, closed -7.5, CLV +0.5
- *Indiana: picked +7, closed +7.5, CLV -0.5
- *Georgia Tech: picked -3.5, closed -4, CLV +0.5
- *UNLV: picked 0, closed -1.5, CLV +1.5
The fact that we’ve shown value on teams (like BYU and Indiana) who’ve lost a good starting QB to injury should not come as a big surprise, since we don’t factor injuries into our ratings. The reason we hadn’t looked at it in-depth before is that we don’t have an injury database, so we can’t tell if a QB change in the past was performance or injury-driven; however, we can make an educated guess. In a true Bayesian manner, we have information that will inform our belief about the cause of the QB change, which we can gauge through its expected impact. How many career starts did the QB being replaced have? More starts makes an injury more likely relative to ineffectiveness. How many career starts did the QB replacing him have? The impact of a QB injury will be highest for the first game, and subsequently decline as the new QB gets more playing time.
So, to summarize, we are now am able to make an empirically-based, estimate of the impact of a QB injury, tailored to the specific team/QB situation. This is a rough estimate, since it doesn’t look at specifics about the individual QBs, other than number of games started. This week, these adjustments are not reflected in our power ratings. However, our picks DO account for these changes.
Week 11 Picks:
Pick criteria is the same: Big Plays (55%+), Other Plays (54-55%), Leans (53.5-54%). Note that these estimates are conservative because lines tend to move our way. If we average even a half-point of line move value (and we’ve averaged more than that) everything bumps up at least one percentage point. To clarify, the games that I list–even leans–are all ones with edges I would play myself. Only the Big Plays and Other Plays are considered official picks, since, well, that’s how we did things last year, and those plays represent our best value.
MP Lines are calculated by taking the difference in MP power ratings, giving a bonus for teams coming off a bye, adding home field advantage, adjusting for QB injury, and further adjusting for pace of play. Lines are the Vegas Insider Consensus Lines current as of 9:05am PST. This prevents any subjective judgment on what is the consensus, and prevents the problem of using a line not widely available were I to use just one book.
Big Plays (12-9-1 YTD)
- Florida State -19.5 vs. Virginia
- Texas +3.5 vs. West Virginia
Other Plays (21-14 YTD)
- Indiana +6.5 vs. Penn State
- UCLA -4.5 at Washington
- UNLV +5.5 vs. Air Force
- Louisiana-Lafayette -16 at New Mexico State
MP Leans (11-11-1 YTD)
- Arizona -16.5 vs. Colorado
- Fresno State -2 vs. San Jose State
- Louisiana Tech -3.5 at UAB