Comments for Skeptical Sports Analysis

Comment on Graphs of the Day: Bird vs. Bron by benjaminmorris

benjaminmorris — Fri, 18 May 2012 08:01:53 +0000

That’s a good question, the post-season is obv different. Though, despite failing to win a championship, Cleveland was actually a pretty strong post-season team. I don’t think there’s enough information at this point to say for sure.

Comment on Graphs of the Day: Bird vs. Bron by Matt

Matt — Thu, 17 May 2012 17:19:03 +0000

Based on your research would you say this version of the Heat or the Cavs with Lebron would me more favored to win a championship? I think this is the second time you have mentioned that you think the Cavs were better than the Heat are now, but would the qualities that make them a better regular season team also make them more likely to win a championship? Kind of going off your “multiple primary options” tangent.

Also I am a big fan of the frequency of posts, keep up the great work!

Comment on Stat Geek Smackdown 2012, Round 1: Odds and Ends by yariv

yariv — Wed, 16 May 2012 14:59:55 +0000

Actually, the team whose city name is first alphabetically in the playoffs gives you 0 of the last 28 champions. The reason is that the Celtics only won championships when the Hawks made the playoffs in this period.

Comment on The Case Against the Case for Dennis Rodman: Initial Volleys by Jacob

Jacob — Tue, 15 May 2012 02:14:11 +0000

Love your site. You are awesome for doing this. My respect for your work only continues to increase as your personal respect and standards for your work becomes even more evident.

Comment on Stat Geek Smackdown 2012, Round 1: Odds and Ends by benjaminmorris

benjaminmorris — Mon, 14 May 2012 23:54:22 +0000

Yes, the NBA version is out of the field.

Comment on Stat Geek Smackdown 2012, Round 1: Odds and Ends by Matt

Matt — Mon, 14 May 2012 22:42:06 +0000

Or maybe I misread your post — are you saying that it predicts the result out of the entire field, or that it predicts the the winner out of the two teams in the Finals? I tend to think it’s the former now re-reading it. If so, carry on…just ignore me.

Comment on Stat Geek Smackdown 2012, Round 1: Odds and Ends by Matt

Matt — Mon, 14 May 2012 22:37:54 +0000

Hello Benjamin,

I have a simpler model that also predicts 18 of the last 28 champions: pick the team whose city name is first alphabetically.

I say this not to be a jerk or to say that you’re not onto something (I tend to think that you are), but to make the point that we are just talking about a sample of 28 here. I think it’s incredibly difficult to draw many concrete conclusions regarding what makes a champion because we just lack a big enough sample.

Thanks,
Matt

Comment on Championship Experience Matters! (Un-Sexy Version) by Crow

Crow — Sat, 12 May 2012 14:53:21 +0000

I’d be interested in knowing regular season and past playoff volatility of performance. Are champs and repeat champs high or low on past playoff volatility? Is there playoff performance volatility / regular season performance volatility notably different than other contenders?

Comment on Stat Geek Smackdown 2012, Round 1: Odds and Ends by Picking the Games for Round 2 of the 2012 NBA Playoffs | The Wages of Wins Journal

Sat, 12 May 2012 12:00:54 +0000

[...] important note is that experience in the playoffs matter. Some of my competitors have pointed this out this year. I have been pointing this out for a while. Basically, playoff vets get favorable calls. To account [...]

Comment on Championship Experience Matters! (Un-Sexy Version) by Westy

Westy — Sat, 12 May 2012 04:26:46 +0000

* “less than” 50%…

Comment on Championship Experience Matters! (Un-Sexy Version) by Westy

Westy — Sat, 12 May 2012 04:24:14 +0000

But then, as per ElGee’s point below, don’t most seasons result in the best team (albeit having the best odds) having a 50% shot?

Comment on Championship Experience Matters! (Super Bowl Edition) by Frank DuPont

Frank DuPont — Sat, 12 May 2012 00:56:40 +0000

One other thing is that beating recent champions is more attractive for regular season opponents. Recent champs always get the opponent’s best game. But in the playoffs opponent effort is going to ge generally even throughout.

Comment on Championship Experience Matters! (Un-Sexy Version) by ElGee

ElGee — Fri, 11 May 2012 01:49:07 +0000

Hi Benjamin — I made the comment about the test period affecting the results on Abbot’s post. Sample size IS an issue here ITO of placing a lot of stock into the predictive power of this method. Obviously, I can’t really display what I’m saying unless we come back to this post in 40 or 50 years, but let’s use your other point, about “ease” of winning a 7-game series, to illustrate this.

It’s not that easy, depending on the absolute difference between the teams. A huge difference – sure it’s easy. But I don’t see a lot of evidence for huge differences in many series. To wit:

Let’s suppose a team is closer and closer in quality to its opponent as the PS unfolds, and their chance of winning a game is:

1st round: 90% every game
2nd round: 70% every game
3rd round: 60% every game
4th round: 55% every game

(Obviously this method ignores home-court, but there’s no need to overly complicate this to make a point.) With such an edge throughout the playoffs, the odds the favored team wins all 4 series are 37.6% (99.7%, 87.4%, 71%, 60.8% per series). One in three times, we’d expect a team with such comfortable margins to win out.

Based on game-by-game analysis, and quarter-by-quarter variance, I think 90% for a first-round opponent is quite high. But perhaps the opponent in the Finals isn’t so close to our team. If the distribution looks like this:

1st Round: 75% every game
2nd Round: 70% …
3rd Round: 65% …
4th Round: 65% …

The odds of winning all 4 series are 52%. Basically, when a team has no one really to close to it, the odds of winning the title are going to be strong (better than 50%). But that’s being liberal by assuming no matchup at any point is a challenge.

The point: It’s actually quite easy to lose in a 7-game series with the high-variance nature of basketball. (I wrote about this last week: http://www.backpicks.com/2012/05/01/how-often-does-the-worse-team-win-sample-size-and-variance-in-a-7-game-series/ ) I’m using liberal numbers here, but in a 28 set sample, it’s not hard to have results that say “40% of the time the best team won” or “60% of the time, the best team won.”

Now, when you say 67-80 data supports your theory, you lose me. The more series there are, the less likely it is the best team wins. What am I missing there?

Comment on Championship Experience Matters! (Un-Sexy Version) by benjaminmorris

benjaminmorris — Thu, 10 May 2012 23:05:24 +0000

No, it doesn’t work that way. The 10 game “margin of error” of an 82 game season technically only applies to a 41 win team. The further you get from .500, the more the range skews. A rough way to approximate is to look at the equivalent percentage change between each % and the maximum and minimum possible. So, e.g., 50% +/- 10% is similar to 75% +5% to -15% (20% in either direction), though note that the likelihood of each value in the range is not equal. It’s similar to what I did for adjusted win % differential in the Rodman series.

Comment on Championship Experience Matters! (Un-Sexy Version) by amobogio basketball-jones

amobogio basketball-jones — Thu, 10 May 2012 20:51:40 +0000

Championship experience drives the model because the very nature and rules of the game change in the playoffs. I would propose that there are more than sufficient differences between the regular season and the play-offs to account for the lack of correlation between them (minutes management, time off and especially refereeing — regular season fouls become playoff play-ons, regular season flagrant fouls are just good hard playoff fouls (see Bynums takedown of McGee the other night night)).

The culture of the NBA worships hard physical playoff basketball – which was why the Nash led Suns never had a prayer of winning a championship.

Additionally as much as I love the game, the NBA focus on championships makes an unfair playing field – in what other sport would [the league allow players as valuable as Boris Diaw and Stephen Jackson to end up on San Antonio? It’s a front runner league, no question…

Comment on Championship Experience Matters! (Un-Sexy Version) by Mike

Mike — Thu, 10 May 2012 20:38:34 +0000

Have you or someone else written elsewhere about the +/- 10 games regular-season idea? I’m weak with statistics, but this seems to me to suggest that there’s a 5% chance, for example, that the 1995-96 Bulls could have gone undefeated. Or if the season had played out 100,000,000 times they would likely have gone undefeated in 5,000,000 of them. And won only 62 games in 5,000,000 others. Is this a correct interpretation? I’d love to read more. Thanks!!!

Comment on Championship Experience Matters! (Un-Sexy Version) by benjaminmorris

benjaminmorris — Thu, 10 May 2012 06:13:48 +0000

This question is actually a bit more complicated than you might think. In any given series between “moderately” different teams, the better might only win ~70-75% of the time. But the best team will be more than moderately better than some of its early-round opponents, and may not even have to play the actual 2nd best team, since that team will only be a “moderate” favorite against the next level of opponents, and so on. Ultimately, you don’t need a huge edge on the field to be a favorite to win the championship.

Comment on Championship Experience Matters! (Un-Sexy Version) by Westy

Westy — Thu, 10 May 2012 03:51:33 +0000

But what level confidence does a 7 game series provide that the “moderately” better team won?

Comment on Championship Experience Matters! (Un-Sexy Version) by Frank DuPont

Frank DuPont — Thu, 10 May 2012 00:33:10 +0000

One other thought that might be a corollary to Hollinger’s thought – maybe non-championship teams value the regular season more because they don’t know that they are expending valuable energy when they win those regular season games.

Winning games is harder/more tiring than losing games, so in addition to the idea that some teams can just turn it on at playoff time, it might be the case that some teams are resting during the regular season.

I suppose it would be fairly easy to figure out whether this is the case or not by looking at minutes played for the championship teams that fit the 5+5 model.

Comment on Stat Geek Smackdown 2012, Round 1: Odds and Ends by benjaminmorris

benjaminmorris — Wed, 09 May 2012 20:25:00 +0000

Congrats! This is the first non-spam comment that akismet has ever flagged as spam (that I know of). I think it doesn’t like the word “poker.”

Anyway, sure, email it to me. Though you might want to DM me on Twitter or let me know here when you do.