My wish list had everything but one item checked off after Monday, as the picks went pretty well but the Lions lost. But we’ll save some of that for tomorrow with the season predictions. Overall it seemed appropriate that a Bond movie opened this weekend, as a number of quarterbacks across the league were left shaken, not stirred (it’s a concussion joke!). But let’s get to the models and how they did.
We always start with the over/under to get the depressing stuff out of the way. Luigi was 2-12, Yoshi 1 6-8, and Yoshi 2 3-11. That puts the season records at 48-77-5, 61-64-5, and 53-72-5.
When I made the picks, I pointed out a few games that were all expected to be lopsided; three teams were favored by more than 10. As it turns out, none of them covered (in fact, one won in overtime and the other tied). My general perspective is that as much as we may think that the Chiefs and Jaguars of the world are awful football teams, it’s just really hard for one NFL team to be expected to beat another by more than 10. Just think about the randomness involved in one game, or even how often a team gets a garbage score at the end. I’ve been tracking spreads for two seasons plus the current one now, so I thought I would look at a couple things with double-digit games.
First, teams winning by double digits isn’t really rare. In the 2010 season it happened 121 times out of 252 games including the playoffs (but ignoring week 1), 119 times in 2011, and 61 times out of 130 games so far this year. So you can expect that about half the time, one team or the other is going to win by double digits. But how often is a team favored by double digits? In 2010 it was only 28 games; it happened 45 times in 2011, and it’s happened 12 times this year. That’s only about 13% of games, or less than one in six. Apparently Vegas appreciates the fact that there’s a difference between what actually happens and what can reasonably be expected to happen. And what happens when a team is favored by double digits? If Vegas was about right, then we should see that the favorite covered around half of those 85 games. In 2010 favorites covered 11 of the 28 games; in 2011 they covered 20 of 45 games; and so far in 2012 they’ve covered 4 of 12 games. Across two and a half seasons, that’s 35 out of 85, or 41%. That isn’t statistically different from 50%, and there are a couple of pushes counted as a failure to cover, so Vegas probably isn’t systematically off on these games. But at least numerically it seems like big favorites fail to cover as often as you might think.
Picking outright winners went well, as Luigi went 8-5-1, Yoshi 1 10-3-1, and Yoshi 2 8-5-1. The ‘push’ was of course due to the 49ers and Rams, a game where neither team apparently wanted to win. There were stupid penalties, missed field goals, back-up quarterbacks… everything you could want if you enjoy watching others suffer. Anyway, the season records for winners are now 77-52-1 for all three models. Picking moneyline winners didn’t go quite as well, thanks in part to the Rams getting a tie instead of a win and the Chiefs bungling their chance against the Steelers as well. Either one would have put the models in the black, but instead they were in the red. Luigi was 4-9-1 (counting the Rams as a push, which I’m not sure is correct), Yoshi 1 5-8-1, and Yoshi 2 4-7-1 for season records of 36-82-1, 43-78-1, and 40-77-1.
The spread went more successfully. Luigi was 7-7, Yoshi 1 10-4, and Yoshi 2 7-7 picking all games. The season records are 54-71-5, 68-57-5, and 57-68-5. That isn’t great (at least for Luigi and Yoshi 2), but you don’t need to pick every game. In the games where they were a bit more confident, Luigi was 7-3, Yoshi 1 9-3, and Yoshi 2 6-4. That puts their season records at 48-49-5, 54-45-5, and 46-49-5. The consensus picks did similarly well; all three models agreed on seven games and they went 6-1 in those games. The consensus season record is 34-28-2.
How about the Hilton SuperContest? For that I use the five games each model thinks are farthest away from the Hilton’s spreads. Luigi would have gone with Tampa -3 (win), Buffalo +11 (win), Kansas City +12.5 (win), Minnesota +2.5 (win), and Miami -6 (loss). That’s a total of 4-1 for a season record of 23-20-2. Yoshi 1 went with KC +12.5, Rams +11 (win), Jets +6.5 (loss), and out of a slew of similar options Buffalo +11 and New Orleans +2.5 (win). That’s also 4-1 for a season record of 21-23-1. Yoshi 2 went with KC +12.5, Buffalo +11, Rams +11, Tampa -3, and Minnesota +2.5 for a perfect 5-0 week. Yoshi 2’s season total is 23-20-2. It’s probably a case of too little too late since the models are all about 7 games out of first, which is a fair amount with 7 weeks left to go.
Wrapping up, we have Bill Simmons’ picks. Bill finally coughed up a crappy week, going 4-10. I look forward to reading what his excuse is. Luigi was 8-6, Yoshi 1 9-5, and Yoshi 2 8-6. That puts the season records at 67-60-3, 57-70-3, 64-63-3, and 56-71-3. Being 10 games behind Bill (taking Luigi as my main model) with 7 weeks to go isn’t great, but at least we pick every game, so there are opportunities to make up ground. We’ll see if he goes into a late-season swoon.
Tomorrow is power rankings and season predictions, and I’ll try to remember to address the mystery of the Tampa Bay Buccaneers.