Chris Paul and Deron Williams... Again

I've written at long lengths in the past about the Paul/Williams debate. Another post on the subject would pretty much be taking the already dead horse, running it over with a dump truck, incinerating it, cryogenically freezing it, tossing it in a black hole, and watching it be exuded as Hawking Radiation. In other words, unnecessary.

So why exactly am I writing about this? Well, Basketball John at SLC Dunk did a post over the weekend on the whole debate. I normally wouldn't respond to a Williams > Paul post, but Basketball John does a great job over at the site, and I enjoy reading SLC Dunk. So I figure I'll explain why his post is misleading.

He makes three major points: (1) CP3 and DWill are only separated by 0.5 assists per game, (2) CP3 and DWill have been headed in opposite directions the last two months, and (3) DWill scored 30 points in 5 straight games, and thus we "[could] see what he had in him."

(1) Only a 0.5 Assist Separation?

If you've been reading this blog for any length of time, you know how much I despise per game stats. Per game stats are among the most misleading statistics you can possibly cite. Why? All games are not created equal. Different games have differing amounts of possessions.

Specifically, the New Orleans Hornets play at an absurdly slow rate (87.8 possessions per game), and the Jazz play at a much crisper rate (92.2 possessions per game).

Assists per possession is the stat we should be looking at. In essence, Paul is being penalized for having fewer opportunities per game if we go by assists per game. Chris Paul averages 0.546 assists per possession; Williams averages 0.471. That's a far bigger disparity than assists per game indicates. If Williams played the same number of possessions as Chris Paul, he'd only average 9.7 assists per game.

(2) Paul Going Down, Williams Going Up?

Nope. This is what we refer to as "statistical aberration." Also known as: stuff happens. A two month sample is subject to a ridiculous amount of statistical variation. Not to mention that Paul recently hurt his groin, and still doesn't look to be at full health.

Add in the fact that Paul started off the year extraordinarily hot- there had to be some regression to the mean. Eschewing a holistic look at Paul and Williams' careers in favor of "nifty graphs" of the last two months is statistically unsound. Those graphs are neither nifty nor do they actually show anything of value.

(3) We've Now Seen What Deron Williams "Could" Do

This is probably the most interesting argument. Here's the flaw: a player's value is not measured by his points per game. The graph illustrates Williams' rising points per game values. A close look at his season indicates that DWill is shooting significantly worse from the field, his already bad rebounding rates are even lower, his  turnover rate has become a little bit better but is still poor. His increased PPG value is more a result of his increased field goals attempted, rather than an increase in efficiency.

And Some Other Stuff...

In the past, Williams has only ever had one advantage on Paul- his shooting. This year, Paul is out eFG%'ing him by an entire 20 points. By the same logic I used to defend Paul's sinking stats, I will say that Williams eFG% should definitely regress to the mean (and become far better). But still... Paul is pretty much at the shooting level that Williams was at his career best (53% vs. 54%). Even if we assume that Williams recovers to last year's levels, shooting is all tied up.

Add on to that: (1) Paul is much better at getting to the line, (2) Paul is much better at converting from the line, (3) Paul is approaching double the rebound rate of Williams, (4) Paul's steals rate is more than double that of Williams', (5) Paul turns the ball over about 3 times fewer per 100 possessions, (6) Paul is accountable for a higher percentage of his team's offense (usage%), even with Boozer out for SLC, (7) Paul plays more minutes per game, (8) Paul is required to create his own shot twice as often as Williams (measured by assisted on %), (9) Paul is significantly better at finishing near the hoop (+8%).

Given the above paragraph, the reason people think Paul is better than Williams is clear. It isn't that they "don't watch Deron play enough." They think Paul is better because he is, plainly, better.

One last point: size. This is the last difference Utah fans cling to, now that the shooting percentages have evened out. Here's the thing: Size inherently means nothing. Size only has meaning when it translates to actual differences on the court.

When we can't see the effects of size difference on the court, size by itself has no meaning. That's what is happening with Paul and Williams. DWill is clearly much more imposing physically. But it doesn't translate in the slightest onto the court, offensively. It doesn't translate at all in terms of rebounding. Emerging defensive stats (adjusted +/-, Basketball Prospectus' DMult) are beginning to prove more and more that Paul is one of the premier defenders in the NBA, regardless of his rep as a "gambler." Both of those stats rate Williams extremely poorly.

Still, "is [comparing the two] totally out of line?  Are we comparing Rafael Araújo to Dwight Howard?" wonders Basketball John.

Nope, not at all. According to PER, it's more like comparing Dwight Howard to Andrew Bogut.