Box Score Geeking: 04/11/2013 Preview

Introducing our new daily feature, Box Score Geeking. Every day we'll take you through the results of the previous night's games and prepare you for the night to come. Right now we'll just focus on today's previews -- we'll catch up on previous results later.

Expect the look of these posts to change as we improve our site features. For now, we're going with basic tables:

Date: 4/11/2013
Home (H): Cleveland Cavaliers
Away (A): Minnesota Timberwolves
Win Probability (H): 47%
Box Score Geeks PM: Wolves by 1
Vegas PM: Cavs by 1.5
Home (H): Los Angeles Clippers
Away (A): Houston Rockets
Win Probability (H): 47%
Box Score Geeks PM: Rockets by 0.9
Vegas PM: Clippers by 5
Home (H): Memphis Grizzlies
Away (A): Boston Celtics
Win Probability (H): 78%
Box Score Geeks PM: Grizzlies by 8.8
Vegas PM: Grizzlies by 12
Home (H): Philadelphia 76ers
Away (A): Golden State Warriors
Win Probability (H): 54%
Box Score Geeks PM: Sixers by 1.1
Vegas PM: Warriors by 6.5

Our most likely upset of the night would have to be the 76ers over the Warriors. Vegas -- as represented by Vegas Insider Consensus lines -- expects the Warriors to win by 6.5 pts. The Box Score Geeks' expected margin is Sixers by 1.1, with the Sixers having a 54% chance of winning at home.

Blowout watch is in effect for Celtics at Grizzlies -- BSG's expected margin is 8.8, and the Vegas Consensus is 12.

We expect Wolves at Cavs, Rockets at Clippers, and Warriors at Sixers to be close games that could go either way.

I know Devin wrote this because he wrote the date wrong.
4 de Noviembre del 2013
What's the problem Kevin? :-)
I actually waffle between the two most common methods, but doesn't it just make more sense to start with the smallest unit -- days -- and increase as you move from left to right?
oh how i pine for sensical date keeping and the metric system...
anyway, devin needs a writer-cartoon and info on The GEEKS page, which his profile erroneously says is there, because teamwork.
ISO_8601 - yyyy-mm-dd
those are some terrible tables. lol.
The fact that your lines are so far off from Vegas show the weakness of Wins Produced.
Vegas was close on the rockets-clippers and warrior-sixers game while bsg was close in grizzlies-Celtics and timberwolves-cavaliers, its 2-2 actually
First off, I am astounded at your ability to predict total season wins and completely agree with the methodology you use to do so.

However, for individual games you need to add additional variables, most notably: game-fixing bias. Without that, your predictions will be off.

The NBA - as an entity operating in an entertainment market - has the right to fix it games. Hence, it is incentivized to fix games that bring in more $. To do this, they typically assign certain referees to certain games. Unless you can account for the referee, then your findings will be inconclusive.

Note: accounting for referees is extremely time intensive. To do so, you have to create a database of every single referee, every single game that they've refereed, and then draw conclusions from the database. Once you have certain conclusions drawn, you can visit http://www.covers.com/pageLoader/pageLoader.aspx?page=/data/nba/referees/current_assignments.html to see the Ref Assignments for the day, and adjust your lines accordingly.

Other factors that will influence the outcome of an individual game: rest, rivalry, home-court advantage, incentive to win, trends, and streaks. That's just a few off the top of my head.
As far at the date is concerned I have settled on the format:
5-Nov-2013 or just 5 Nov 2013
for the same reason as Devin but also to avoid ambiguity. This also makes searching on a date easier because the smallest unit is first. However because the months are not in alphabetical order if sort order is a major concern then it is better to use the ISO or reverse ISO standard:
yyyy-mm-dd or dd-mm-yyyy.
"The fact that your lines are so far off from Vegas show the weakness of Wins Produced."

Right, because it is IMPOSSIBLE that either a) Vegas might be wrong about some things or that b) Vegas might not give a flying fuck about getting the number of wins correct, and only cares about splitting bets on both sides of a line to make money on the rake.

Both of those things are so ludicrous that only a moron might consider them as a possibility, so our statistics MUST be weak.
Shut the fuck up Max. Just... ugh. Nobody's been able to demonstrate consistent ref effects (even though plenty of people would be quite interested to see it), and while I don't think that the referee system (or WP) is without flaw, it's quite minimal.
I know the model doesn't like the Warriors as much as I do, but how could it have given you 76ers by anything? Was it just weighting MCW's recent performance really heavily, since it's the only NBA sample we have for him?

Also, WHOOOOOOOOOOOOOO CURRY TRIPLE DOUBLE AND IGGY BANGING IN 7 TREYS!
You guys should put up a tracker showing your MSE on games to date vs Vegas line. You know, for science :-)
Patrick, you seem to be particularly naive about how the sports betting world works. My comments were mostly directed at the lines posted for games yesterday, but since you brought up win totals lets talk about that.

If I had a metric that I really thought was the closest to the truth (better than anything else out there) and it said that there were three bets with a > 98% confidence, I would mortgage my house, max my credit cards out, and borrow money from friends so I could put every last dollar I had on those bets. It would be the safest investment in the history of mankind.

Sportsbooks are the most consistently accurate predictor of game lines and season win totals. The minute you start significantly deviating, you're probably wrong, and if you're not - you could easily make millions of dollars. Who know, maybe I'm really the moron here, and the WP crew will be swimming in their pool money soon.
@DooDoo_Jump: Feel free to look into Bob Voulgaris or any other top NBA bettors. All of them calculate in referee bias. And if you don't believe in referee bias then watch a quick video on how the Lakers made the playoffs last year.

@Everyone Trash-Talking Wins Produced:
Wins Produced is the most logical statistic metric that also happens to have insanely high correlation to actual wins in a season. However, on a game by game basis it's less accurate since it doesn't calculate for game by game variables. To calculate for these variables these guys would have to make it a full-time job (reading up on players, watching post-game interviews to study psyche, calculating trends, referee bias, etc.)

Also, don't trash talk a metric after a 4 game sample.
I agree that you should track your predictions vs. Vegas for the season, I would be interested to see how WP stacks up against Vegas odds for a full season.
"And if you don't believe in referee bias then watch a quick video on how the Lakers made the playoffs last year."
so spot on it made me laugh... also, wish this remained/hope this remains a daily feature.
To everyone knocking them for not betting all of their money on their predictions: those predictions are made without knowledge of trades, injuries, or quality knowledge about freshman and sophomore players.

While the guys here might do well to temper their seemingly absolute belief in WP, they aren't claiming that it's foolproof.

And even if you had what you believed to be a 95% probability of winning, you'd be foolish to put all your money into it.
So... what happened to the daily aspect?

DooDoo_Jump,

Re: “While the guys here might do well to temper their seemingly absolute belief in WP, they aren't claiming that it's foolproof. “

My problem is that they give lip service to the notion that it is not foolproof but in practice they do not hesitate to make judgments about players, coaches, owners, etc, as if WP was a fact instead of just a calculation.

Whenever I refer to, for example, WP48 on another website I always just say something like: “for a different perspective, here is an advanced stats view of ...{whatever is being discussed}.” I never make or imply that WP48 is the “true answer”.

DooDoo_Jump,

Re: “While the guys here might do well to temper their seemingly absolute belief in WP, they aren't claiming that it's foolproof. “

My problem is that they give lip service to the notion that it is not foolproof but in practice they do not hesitate to make judgments about players, coaches, owners, etc, as if WP was a fact instead of just a calculation.

Whenever I refer to, for example, WP48 on another website I always just say something like: “for a different perspective, here is an advanced stats view of ....{whatever is being discussed}.” I never make or imply that WP48 is the “true answer”.
I'm with you there. I think they've been getting a little better about it here, but there's still that tone problem, and the complete disinterest in exploring alternative explanations for anything.
it hurts us if our favorite player/team is trashed not thee wp ;)

Sign in to write a comment.