Arturo avatar

3/4 Visual Guide

All Data for the post is through 3/12/2014

We're back with more visuals! You might note that we made some upgrades. The logos are smaller, and we converted the offense and defense ratings to point margin. As always, we can't do this without your awesome feedback, so please it coming.

Season Ratings

The Season Ratings graph shows the value of each team's offense and defense to their overall adjusted point margin per game. It's expressed in points per game, with offense on the side and defense on the bottom. Greater than zero numbers mean better than league average offensive/defensive unit. The higher the better the offense, the more to the right the better the defense.

Adjusted Win% vs Actual Win%

The Adjusted Win% versus Actual Win% graph intends to show how much the schedule affects the actual win total of each team. A team on the center black line represents a team whose record actually reflects the strength of that team. Teams above that line are stronger than their current record. Teams below that line are weaker than their record. 

 

Pace

The Pace graph shows the typical number of offensive possesions in a game for each team. Faster Pace (to the right) means more points and larger point margins per game. Slower pace (to the left) means fewer points and smaller point margins.

 

Strength of Schedule

The Strength of Schedule graph shows how a team's schedule and opponents affect its actual point margin per game. Teams with a negative point margin (to the left of the vertical axis) have played a tougher schedule. Teams with a positive point margin (the right of the vertical axis) have played an easier schedule.

 
As before, consider this a continuing proof of concept of some of the tools that I'm building. Thoughtful discussion and criticism is welcomed, but please send all flames to /dev/null.

 

You know how when you do a regression with a cloud of data that has zero correlation and one distant outlier, it can mess up the regression and show a high correlation because of the outlier? Well I'd like to see pace vs. win% without the 6ers. Their so far down and to the right (it's adj. W/L, so their even decently below MIL)!
And why don't you guys just use adjusted W/L or something similar for the power rankings, their way too subject to recency bias based on whatever formula that is. Which one correlates better with future performance? I bet it's the unweighted one.
One other note, has anyone else noticed how these guys' early season graphs tended to show a non-1 slope for the W/L vs. adj. W/L; the winning teams were called lucky and the losing teams unlucky, but now it looks like a pretty 1 to 1 relationship (probably because they have a regression to the mean constant in there whose impact becomes weaker as the season progresses. Someone calculated the constants for W/L and point differential to prove how much better point differential is, maybe a Wages commenter, but I gave up after some cursory searching of the interwebs).
Suggestion (this is a pipedream maybe): if there was some way to make it so that the data "played", that would be awesome. Like Gapminder, where the unlucky / lucky graph showed after every game the movement.

That would make it easy to show, say, the Pacers fans that called you haters how unlikely they were to maintain their early season for, because watch 1996, and see how, eventually, most teams regress back towards the line.
Very nice work. Is there some way to make icons see-through? It took me a while to find the Spurs in the season ratings because the Clippers logo completely covers it. I do love all the little tools build in you can play with though.
I agree with SVG that what the 76ers are doing is reprehensible. Playing at that helter-skelter pace with that bad a team is a very clear attempt to lose ALL OF THE GAMES.

16.5 wins might already be out of reach since they have to win 2 of their last 16 games and Boston is the only team they play in the home stretch that isn't scrambling for playoff seeding.

The Philly over might turn out to be one of the worst gambling losses ever- given that they had to go 2-33 to beat it.

Sign in to write a comment.