Week 12: Spread Picks, Results and Analysis

by thesanction1

I’d never been more convinced in NFL karma than when the Broncos jumped out to a 24-0 lead over the Patriots this weekend. Finally Peyton Manning would dominate Young Tom.  Finally Bellicheck would get his comeupins. Finally the world would be right-side up, and my inner Raider could celebrate the Patriots being almost as embarrassed and humiliated as the Raiders were by the Tuck Rule fiasco. And it would come on the week when the algorithm – which has thousands of variables, none of them having to do with this quite relevant history – picked the freaking Patriots plus 2.5 points as the most confident pick of the week. Outside of a Bronco victory by such a narrow margin that the Pats could still cover, I was destined to be disappointed in one way or the other.

It turns out, as so often is the case in sports, my heart went wrong where my computer-assisted head prevailed. The Patriots would rally against 3 early turnovers, against what amounted to career suicide (at least for the near-term future) for Steven Ridley, against a 24-0 deficit, to claim victory 34-31 in overtime. The game, which started out so badly, so lopsided, that many probably went to bed in the first half, ultimately lived up to every expectation. Brady dominated Manning statistically. Bellicheck solidified his super-evil-genius by deciding to take the wind over the ball – kicking to Peyton Manning at the start the overtime period.  Facing one of the all-time greatest quarterbacks ever, Bellicheck threw up the middle finger. Come get me. Incredible. Brilliant. I hate him. Manning couldn’t keep the nose of the ball down in that wind, and once it got a hold of his often loose-spiraled passes all was lost. Peyton couldn’t perform with his typical laser-like precision. He was cold, he was uncomfortable, and he was human. Bellicheck knew it, and he wanted everyone else to see it too.

The game was so close that during the final 5 minutes of overtime I genuinely thought it might end in a tie. This would have been ideal; the machine gets the pick right against the spread, and I can save face with the Peyton-Brady discussion my old man and I would have in the days to come. Neither team could move the ball in the overtime period, and after another three and out by the Patriots it seemed like the offenses were running out of steam.  Then something insane happened. On the ensuing punt, the Broncos replaced young speedster Trindon Holliday with the more sure-handed (and ex-Patriot) Wes Welker to make absolutely certain that the punt wasn’t muffed.

The punt was muffed.

The Patriots took possession in Bronco territory and the rest is history. Total insanity. I’m no conspiracy theorist, but it wouldn’t totally surprise me if Bill ‘Lucifer’ Bellichek had schemed up the entire plot years ago and planted Wes at Denver in an epic long-con, for this one incredible payoff. Either way, this game proves a vital point for anyone trying to predict the outcome of sporting events: the future is more insane than you can imagine.  The odds of the Patriots overcoming a 24-0 deficit to force an overtime period against the league-leading Denver offense were nill.  The odds of Wes Welker muffing the game-losing punt were even lower. There were a handful of moments throughout this game that could have gone a different way and changed the outcome of the matchup. No one can predict such things, which is why they play the game.

The algorithm attempts to distill all the fuzziness and uncertainty of competition into something quantifiable, to bring a bit more clarity to the opacity of sport.  Let’s see how it did.

Overall performance – week: 9 of 14 (64%), season: 14 of 26 (54%)

Here is the performance if we ignored the absolute size of the Cover.Index (SCI), which is a proxy for the confidence of the pick, and simply picked every game:


Sticking with my one-week tradition, I’d like to include the probability that this level of performance or better for the season is a product of chance, by using our coin flipping analogy and my favorite stats site www.WolframAlpha.com. This will give you some idea of how likely it is that the machine is just ‘getting lucky’.

The probability of going 14 of 26 or better using a coin to pick games is 42.2%. So, good week?  Sure. On the right track? Seems so! Betting the farm next week? I wouldn’t advise it.

SCI greater than 1.0 – week: 5 of 9 (56%), season: 10 of 18 (54%)

Here is the performance if we restrict our selections to those games in which the SCI is greater than 1.0:


The probability of going 10 of 18 using a coin is 40.7%.

SCI greater than 2.0 – week: 2 of 4 (50%), season: 5 of 9 (56%)

Here is the performance if we restrict our selections to those games in which the SCI is greater than 2.0:


The probability of going 5 of 9 using a coin is 50%.  Think about that – if you pick 9 games at random, half the time you are going to win more than you lose. This is the lens through which all betting outcomes should be viewed and the reason why any legitimate edge should be played out over the longer term.

SCI greater than 3.0 – week: 2 of 3 (67%), season: 4 of 6 (67%)

Here is the performance if we restrict our selections to those games in which the SCI is greater than 3.0:


The probability of going 4 of 6 flipping a coin is 34.3%.  To date our most confident picks are also the least likely to have been a product of chance – but a sample size of 6 definitely doesn’t qualify for any measure of significance.

Once again we see the highest SCI games being our most performant for the season.  This is a good sign as it matches with the way we expect the algorithm to work.  Last week the more volatile, less confident games swung against us. This week they went our way, contributing to a very solid overall performance of 64% win percentage against the spread. The picks with SCI greater than 1 but less than 3 underperformed the picks with a less than 1 SCI, which is not expected to happen week-in and week-out, but is definitely inside the realm of the statistically feasible.  It’s the kind of outcome I was happy to not have happen with my week one picks, because it makes explaining the SCI system slightly more complicated.

All in all, another solid week for the machine.  I am going to make an effort to have picks ready prior to the three Thanksgiving games this week, because turkey always taste better when your winning a few bets.

If anyone made money on these picks, or just has an awesome Sunday story, I’d love to hear about it in the comments. Check back tomorrow for the fantasy QB analysis, hope everyone has a great week!