Well, a bad week was bound to happen. Brian at Advanced NFL Stats did a good job of getting in front of things by reminding everyone that the model isn’t supposed to be right all the time; if it thinks that a team should win 80% of the time, that means they should lose 20% of the time. And, to the extent that misery loves company, Brian’s picks also didn’t do very well this week. I didn’t agree with all of his picks, but I did think that Oakland, Dallas, Houston, and Detroit in particular surprised by doing better or worse than expected. Here’s the post mortem:
Over/under: There was one -115 game, Buffalo-Jacksonville, that both models hit. Mario was 7-6 in the rest of the games for a total of $121.31. Luigi was 8-5 for a total of $314.22. Mario continues to bring up the rear on the season at 26-33 for -$840 while Luigi is 31-28 for $120. Hooray in the black!
Moneyline: Mario liked the Rams, Chiefs, Redskins, Broncos, Cardinals, Titans, Eagles, and Vikings. That means it went 4-4; at $100 a game that’s an overall outcome of $375. Luigi had the same picks except it liked the Bucs and not the Redskins, so it also went 4-4 but for $510. The models seem to only pick underdogs from the moneyline, so they can afford to be wrong (or break even) in terms of number of games and still win money. On the season Mario moves to 11-6 for $1375 and Luigi is 11-7 for $1410.
Spread: Sadness rains as Mario only picks 5 games correctly out of 14; Luigi did much better with 7 out of 12(skipping the Dallas-Tennessee and Minnesota-New Jersey games). Mario goes to 27-29-1 and Luigi 29-22-1.
Simmons: Bill went 9-5 with his picks. Mario only went 4-10, missing an extra game because Bill’s line had Buffalo at +1 instead of -1.5 as it was on Bodog on Sunday. Luigi went 9-5 (the NO pick in the Simmons section is due to a typo in the line; Luigi went with Arizona just like for the spread). So Bill goes to 40-33-3, Mario to 27-32-1, Luigi to 33-26-1, and I go to 35-38-3.
Given how things are going, I’m making the call to switch to Luigi as my preferred model. I had liked Luigi for the over/under and picking the winner, which it is still doing well at, and I liked Mario for the spread because in the past two years it had a higher profit. But, in terms of games correct it was about even with Luigi, so there’s some chance that it happened to pick an unusual number of upsets or the like. So next week I’ll still present all the same predictions for both models, but for my challenge with Bill Simmons I’ll be using the Luigi picks. I think I need to tweak the moneyline one more time as well, so I’ll update that next time.