Monday, September 30, 2013

How high will the Jaguars-Broncos point spread go?

Barry Petchesky of Deadspin today speculated that the week 6 matchup between the Jaguars and the Broncos could have the largest point spread in NFL history. According to a tweet from RJ Bell, Jay Kornegay, director of the Las Vegas Hilton sportsbook, puts the point spread at 28 points.

According to my rankings (as of September 30*), that number seems a bit high. The Broncos' GPF (Generic Points Favored) is 8.7 points. The Jaguars' GPF is -11.5 points. Factoring in a generic home field advantage of 2.5 would imply a point spread of 22.5. According to, there have been two games since 1989 with a point spread at least that high.

Early week 6 lines should be available soon, so we'll see where they start the spread and how it eventually closes.

* There was an error in the rankings table that has since been corrected. The original version of this post had a predicted point spread of 21.5.

Sunday, September 29, 2013

College Football Rankings Now Available

College football rankings have been available for about a week now, but I just hadn't gotten around to announcing it. Like my NFL rankings, these will be updated daily. As a reminder, these rankings are based on the betting point spreads and totals, with the goal of reverse engineering an implied betting market power ranking. See this post for more background, or the Methodology page for a more general overview.

The rankings tend to be a bit noisy early in the season. I'm not sure if Oregon would really be favored over Alabama on a neutral site, but that's what the numbers are saying as of today.

Sunday, September 22, 2013

NFL Rankings Now Updated Daily

The daily version of my NFL rankings is now up and running. These rankings check the latest point spreads each morning and update accordingly, with the explicit goal of deriving what the betting market thinks is each team's true strength. If my model was perfect, you could duplicate the point spread of any matchup by taking the difference of the two team's GPF (Generic Points Favored) and adding 2.5 points for home field advantage.

For additional background, refer to these posts:

NFL Rankings - Now Updated Daily
NFL Rankings - Strength of Schedule
NFL Rankings - Projected Wins

Also, my backend database for these rankings is a a Google Docs spreadsheet, which doesn't always cooperate. If anybody is having trouble getting the rankings to load, feel free to shoot me an email, or leave a comment below.

Saturday, September 21, 2013

NFL Week 3 Rankings

Russell Wilson with the SeahawksHere are the week 3 NFL betting market rankings. The daily version of these should be up and running soon, as I now have enough weeks of lines and totals to have a properly connected dataset. I still plan on posting a weekly version as a standalone blog post so that past results are a bit easier to view.

These rankings attempt to figure out what the market really "thinks" about each team by reverse engineering an implied power ranking from the Vegas point spread. See here and here for more background.

Wednesday, September 11, 2013

NFL Week 2 Rankings

Deux jaguars du Pérou
Deux jaguars du Pérou
Here is your weekly check in on how the betting market ranks the NFL teams. With opening lines available for week 3 NFL games, I can now dispense with the Cantor Gaming lines to fill in the gaps in my regression analysis. The rankings below take the point spreads that Vegas sets for each NFL game and uses them to reverse engineer an implied team ranking. See here and here for more background.

The Jaguars Limbo Under a Low Bar

The Jaguars (somehow) found a way to be even worse than expected. A four point underdog against the Chiefs (at home, no less), and they lose by 26. Prior to their week 1 loss, the Jaguars were a 3 point underdog on the road against the Raiders in week 2. The market has since corrected, and the Jaguars are now a six point underdog, although part of that movement could be due to a perceived better than expected showing by the Raiders against the Colts.

Monday, September 9, 2013

US Open Mens Final - Win Probability Graph

Here is the win probability graph for Rafael Nadal's four set victory over Novak Djokovic in the US Open Mens Final (pretty much identical to the live version). See Jeff Sackmann's recap for a detailed breakdown of the match. I don't have much to add, other than to point out that the Mens Final and Womens Final had nearly equivalent numbers for the Excitement Index and Comeback Factor.

Novak Djokovic2641 Excitement4.3
Rafael Nadal6366 Comeback0.6

Point Probabilities (Model Inputs)

US Open Mens Final - Live Win Probability

I make no guarantees that this will work, but I will attempt to show a live win probability of today's US Open Mens Final at this page. I currently have the Women's Final below as a placeholder, but once the Djokovic-Nadal match starts, the graph and table below should start updating. But if it breaks, I probably won't be able to fix it.

The probabilities are pre-calibrated to the match betting odds (Djokovic is a slight underdog with a 41.9% win probability). They are also based on a best of 5 sets, in case you were wondering why these differ from my official graph of the Womens final.

This is not set up to auto-refresh, but just hit the "Load Data" to get the latest. If the graph goes blank or looks funny, just wait a few seconds and click again.

Sunday, September 8, 2013

Name Change

Ron Artest 2011
Ron Metta
When I started this blog a year and a half ago, I began with a fairly narrow scope: analysis and exploration of sports betting markets. Although my interests were broader than just the money bet on the game, my feeling at the time was that most of the interesting work in fundamental sports data analysis had already been done, and done much better than I could have hoped to do on my own. And while that is still largely the case, for better or worse, my focus on this blog has expanded beyond just "following the money".

So, as part of this mission creep, I've decided to rename this site. For one, I was looking for a name less, um, bureaucratic than Betting Market Analytics. And with fewer letters (so many letters in that url, especially with the Picking a website name in 2013 must be akin to picking out a personalized license plate: all the good ones are taken, forcing you into immediate compromise and uncomfortable questions ("Does anybody ever actually visit a .co domain?", "Can I use a 3 for an E?","What if I add a silent K?", etc.). While the pickings were slim, I was able to snag a domain name I was happy with:

That's not a word

True, inpredictable is not a word, but it's close enough. Inpredictable, of course, is a mangling of a real word: unpredictable. And if I've learned anything over my career in crunching numbers and making predictions, it's that numbers and analysis can make you smarter, but you're still going to be pretty dumb. The good news is, most everybody else is just as dumb (fans, pundits, former players, even your bookie). After all, "It's tough making predictions, especially about the future".

The "in" in "inpredictable" can refer to in-game win probabilities, a recent obsession of mine, and one I intend to be a core feature of this site (I've already built models for the NBA and tennis, and made some furtive stabs at an NFL model).

Team Rankings and Other Features

My betting market team rankings I produce for various sports (NFL, NCAAF, NCAAB, NBA, and MLB) will continue to be shared here. The same goes for various other features, like the Turnover Index. I also plan on some upgrades to the format and layout of the site that should be rolling out soon.

All links referring to bettingmarketanalytics pages will automatically redirect to inpredictable, but it's probably a good idea to update your bookmarks.

At its core, this will remain a site devoted to analysis and understanding of the sporting (and sometimes betting) world. Hope you enjoy.

US Open Womens Final - Win Probability Graph

Here is the win probability graph for the US Open Womens Final, featuring Serena Williams against Victoria Azarenka. The match looked to be over mid way through the second set, with Serena leading 4-2 in the second set and threatening to go up 5-2, having two break points to give. Williams' win probability at that point was 99.5%. But Azarenka was able to hold serve in that game and ultimately force a tiebreaker. After winning the second set, Azarenka's win probability got as high as 40%, but Williams took control after that, rolling to a 6-1 victory in the final deciding set.

See here and here for more background on the win probability graphs. Here is Jeff Sackmann's detailed breakdown of the match.

Serena Williams766 Excitement4.3
Victoria Azarenka671 Comeback0.7

Point Probabilities (Model Inputs)