We’re one game into the preseason and that means it’s time for a lot of forward-looking predictive pieces in the next few weeks, as talking heads and the “stats guys” battle for prediction supremacy.
We’ve already discussed Hollinger’s player profiles and PER projections, and Basketball Prospectus’ SCHOENE projections are still a few days from publication (if you’re a rich yuppy type who can afford paywalled content).
Today, though, we were treated to some statistical analysis from Wages of Win. WoW is an analytical site that I enjoy reading, however, I’m unsure if there has ever been any research done into the efficacy of different projection systems. This has been done in baseball for individual player projection systems, but I’m ignorant to whether it has been done for basketball at the league-projection level and, if so, how WoW stacked up. With that said, they’re currently available (and shocking), so let’s take a look.
Their methods are a lot to explain, so if you’re interested, head over there and have a look. If not, I’ll summarize by saying that they’ve found that empirically, only the top-6 players on a team matter for playoff performance, but quality depth can lead you to the playoffs.
Based on eight different categories (Wins Produced, Point Margin, Offensive Wins Produced, Defensive Wins Produced, and each of those four also on a per-minute basis), they rank every player in the NBA similar to the ESPN NBA Rank program we looked at earlier in the offseason. They then take the individual player rankings to create team rankings (more on that in a bit).
The Raptors’ roster is distributed similarly to what we saw with NBA Rank, with a few key differences…no wait, some seriously major differences. A starting lineup of Calderon-Lowry-Fields-Davis-Johnson with Wright, McGuire, and Gray as the first guys off the bench? What-what-what?
The first major difference is that Kyle Lowry ranked as the #21 player in the NBA with an average rank across the eight categories of 44.5. Basically, because Lowry is in the second tier in just about every area, he ranks as a top-tier player overall.
Ed Davis also makes a massive leap, finding himself at #42 overall thanks to some solid defensive and per-minute contributions. This one was a bit of a shocker since, while I’m a fan, I think you’d be hard pressed to find anyone other than Arturo from WoW who believes Ed Davis could be a top-2 player on an NBA team, now or in the future (although to be fair, I doubt he thinks this as well). In a similar vein, Amir Johnson ranks at #60, showing that the Raptors big-man rotation, while overrated here, is potentially underrated on the whole.
Jose Calderon clocks in at #51, posting upper-crust numbers in all non-defensive categories, including a #6 ranking in overall offense. Landry Fields is at #53, and then suspected bench-dwellers Chris Wright, Dominic McGuire, and Aaron Gray check in at 101, 136, and 152. They don’t rank beyond #180, since the purpose of their piece is to look at the “top-6” tier of players.
I dug deeper using the Google Docs link he posted, and the players rank as follows:
I swear to Shuttlesworth that I didn’t post this just to troll DeRozan and Bargnani comments. But have at it…Jamaal Magloire is better than Andrea Bargnani. Boom.
Now…I know what you’re thinking. This does not appear to be an intuitive analysis, and you’re probably right given that I find averaging a player’s categorical ranks to create a full ranking isn’t necessarily the best way to rank value. Also, there should be checks in place that say “when a borderline training camp invitee narrowly misses the top-100, we should re-evaluate our formulas.” However, let’s push forward.
Raptors as a Team
The analysis then begins to evaluate teams based on the number of players they have in the “top-6” range (#1-180), the “bench” range (#181-330), and the “dregs” range (worse than #330). With 8 players in the top-6 category, the Raptors are among the best teams in terms of having playoff-caliber players.
Using an arbitrary weighting system based on player rank, the article ranks the Raptors 4th based on high-end roster composition. This, once again, should have stuck out as an issue with the system, since not even the most optimistic among us would argue that the Raptors are the 4th best in the NBA at…well, anything, probably.
They then use an algorithm to project minutes, and then apply those minute distributions to their simulation engine to create win projections. Note: These do not include rookies…I strongly disagree with not even trying to project them or using some sort of baseline in the interim, but I digress.
The Key Result
The Raptors will win 44.5 games and finish 4th in the Eastern Conference. No, seriously. Depending on how they alter their model, it could be as low as 43.5 or as high as 45.4, but they’re expected to get the 4th seed in an average scenario.
The author indicates that he plans to add rookies and Euro League players into his analysis before the season starts, and I’m interested to see how that changes things in the win projections.
As it is, you’d almost have to argue the current methodology is flawed – in general, you shouldn’t make conclusions about a model’s efficacy based on results you didn’t expect, but between the high win total and the bizarre rankings of some players (seriously, Andrea is the 2nd-worst player in the NBA? Come on) you have to question it a bit.
With that said…he admits it’s incomplete, and I don’t believe his main point is to accurately predict win totals. If you take the analysis as more of a touching-base with roster composition, you can use it as a means to identify teams that could be undervalued in the public eye due to their strength coming from depth or all-around adequacy rather than top-heavy offensive dynamite. The fact that the Pacers, Timberwolves, and Nuggets all also do well, and are likewise teams built with depth, speaks to that notion.
So no, I don’t think Chris Wright is a top-100 player in the NBA, and I don’t think the Raptors will win 45 games and have home court advantage in the playoffs. I do, however, think that our own “inflated” expectations of this being an adequate team and a borderline-playoff team, are not as ridiculous as some make them out to be, however flawed this particular analysis may be.