We continued our recent streak of dominance last week, going 2-0 on Big Plays, 3-0-1 on Other Plays, and 2-1 on MP Leans. After our dreadful start, this means we have fought back to the 50% mark for all picks.
I’ve gotten some questions recently about what the threshold is for the different pick categories. The quick-and-easy answer: Big Plays have a 55%+ chance of covering; Other Plays are 54%-55%; MP Leans are 53.5%-54%. The way we’ve determined picks has evolved a bit as the season has gone on. The Massey-Peabody line is linearly determined. This means that just because the M-P line for a game is -7.5 does not mean that a team has a 50% chance of covering a line of -7.5, because score differential is discrete, and not normally distributed. In this example, I would expect a score differential of 7 from the favorite’s perspective 6.1% of the time, while 8 and 9 should occur only 1.9% and 1.0%, respectively. So the mean and median score differential can differ.
The way I’ve dealt with this is to use a statistical technique that enables me to estimate the probability mass function for score differential. Based on this, I empirically find that a 7.5-point favorite according to the M-P line should only cover a line of -7.5 45.3% of the time. However, this assumes that the Massey-Peabody line is all that matters, when in reality the market contains lots of information not accounted for by our line. So I need to factor in the market numbers as well. Empirically, I’ve determined the optimal weight that should be given to my line relative to the closing line (for the stat police: this is done out-of-sample), but we know that the closing line is way more efficient than the line mid-week. This means I’m being conservative, and probably regressing my cover rate a little too strongly back to 50%.
So the point here is that while our lines are entirely quantitatively-determined, and our edges are also quantitatively driven, my predicted cover rates are very conservative.
On to this week’s picks. As usual, we don’t account for injuries or matchup factors. Lines widely available as of 12:15pm EDT on Wednesday.
Big Plays (14-8 YTD)
- Cincinnati -2.5 at Memphis (note: this game is played Wednesday night!)
- Notre Dame -16.5 vs. Navy
- Southern Mississippi +31 at Marshall
Other Plays (15-23-1 YTD)
- USC +5 at Oregon State
- Minnesota +9.5 at Indiana
- North Carolina -3.5 at N.C. State
- Texas -28 vs. Kansas
MP Leans (11-9 YTD)
- Northern Illinois -23.5 at Massachusetts
- Temple +13.5 at Rutgers
- Penn State -10 vs. Illinois
- San Diego State -14.5 vs. New Mexico