clock menu more-arrow no yes mobile

Filed under:

Wages of Wins are not Pelican Fans, Part I


Last week, Arturo Galletti, of the Wages of Wins Journal, predicted the Pelicans will finish the 2013-14 season in the neighborhood of 29 wins - and the 26th worst record in the NBA. Yes, that's just two more wins than last season's total, despite the revamped roster.

Is this cause for panic? Probably not, as it won't be the first time their model is likely undervaluing a franchise by a sizable margin.

In advance, I'd like to apologize to all those who may be offended, but statisticians do not always get it right. In my experiences, this has proven true with future projections, especially when talented, but non-Lebron-James-types switch teams, an entire roster is drastically overhauled and/or there is a significant challenge or change in a team's strategic approach.

Today, we'll focus on some prior errors in statistical analysis and attempt to discern where the numbers went wrong. Next week, we'll breakdown the Wages of Wins (WOW) forecast, how it may be incorrect and, most importantly, why fans shouldn't fret nearly as much.

Difficulty projecting players in new surroundings.

One year ago, the NBA Geek considered Ryan Anderson to be the best contract signing of the previous off-season. However, his advanced statistics, WS/48 and PER, ended up being the worst since his rookie season. They hypothesized:

Ryan Anderson is a more troublesome case. He had by far his worst year, and it is almost entirely due to poor rebounding. This is something to keep an eye on; if his rebounding does not recover, he essentially becomes a rich man's version of Andrea Bargnani. And no, that's not a good thing. What's odd about this is that one would think that his defensive rebounding would be better since he's no longer playing next to Dwight (defensive rebounds suffer from diminishing returns).

So the explanation his game has suddenly taken a turn for the worse is he's the second coming of Andrea Bargnani?

Hold your horses, worrywarts. Rather, consider this set of more plausible circumstances:

  1. Hornets woeful defense. Ryan Anderson went from playing on one of league's better defensive units, the previous several years, to one of the worst. Simply put, there were less available defensive rebounds: a) the opposition had an easier time scoring and b) you're not standing next to Dwight Howard, one of the league's best defensive weapons. Opponents were not forced into more difficult shots and, additionally, they were in better position for offensive rebounds, due to multiple defensive breakdowns. An increased free-for-all environment will naturally encumber the less athletic players in winning the 50/50 balls.
  2. Surrounded by better rebounders. On the 2011-12 Orlando team, his main competition for rebounds were Glenn Davis and Dwight Howard. Additionally, one of those players were guaranteed to be on the bench when Anderson was on the floor. In New Orleans, Jason Smith, Anthony Davis, Lance Thomas, Al-Farouq Aminu and Robin Lopez all posted TRB% well over 10%.
  3. His advanced statistics suffered considerably from inefficient shooting. On a team with few offensive weapons, Anderson was asked to do more. Unfortunately, the increase in shot attempts came from the least efficient part of the floor, 10-23 feet. Anderson more than doubled his attempts from this area than in his final year with the Magic. Factor in less shots at the rim and fewer trips to the foul line, a 7% drop in TS% (.589 -> .548) was inevitable. Thus, a sharp drop in his offensive rating hampered the once glowing advanced numbers as much as any missing rebounding prowess.
Problems projecting overhauled rosters.

Last year, WOW predicted the Nets would win 32 games, the Warriors, 37 and the Timberwolves, 50 (yes, this included the initial injuries to Ricky Rubio and Kevin Love!) Even the Hornets were supposed to win near 42 games. Also, they didn't foresee any team reaching 60 wins, but the Heat and Thunder managed the feat.

Instead of analyzing each team in detail (and making for a painstakingly long article), we'll just center our attention on the Oklahoma City Thunder. WOW claimed that the loss of James Harden would drop the Thunder at least one level, sitting somewhere below a true championship contender.

However, throughout the entire season, the statistics certainly never supported this claim. The Thunder's win percentage last year was 73.2%, up from the 71.2% the prior season. Their offensive rating rose from 109.8 to 112.4, their defensive ranking jumped from 11th to 4th and the team managed to increase their PPG by 2.6.

So what happened?

Essentially, the statistics showed Harden's production was replaced by the rest of the roster. Kevin Durant, Russell Westbrook, Serge Ibaka, Kevin Martin, Nick Collison, Thabo Sefolosha and Reggie Jackson all improved their win shares per 48 minutes (WS/48). Some players saw noticeably improvement in their playmaking (Durant, Westbrook), while others in scoring (Ibaka, Jackson, Sefolosha).

Expanding the roles of Durant and Westbrook, Ibaka taking another step forward in his growth and players like Martin, Collison and Sefolosha filling their roles perfectly, allowed the Thunder to collectively improve. It wasn't a case of addition by subtraction, but instead, player skills and talents were better allocated.

Although WOW accounts for age improvements, it is linear in nature. Any talent(s) that might have been suppressed in the past, due to a previously more limited role, will fail to be revealed in future projections. Therefore, dramatic gains in efficiency are possible when certain players are propelled into sometimes larger, sometimes different roles.

Problems correctly anticipating strategic challenges and changes.

Last year's preseason darlings, the Los Angeles Lakers, were picked by everybody and their mother to be one of the best teams in the Western Conference. Undoubtedly, they were ravaged by injury, but did you know their record, in games Steve Nash, Kobe Bryant, Dwight Howard and Pau Gasol all stepped onto the floor, was a paltry 8-14?

The couple of coaching changes didn't help matters, but the team never appeared comfortable all season, despite Mike D'Antoni guiding them in 72 games:

Mike D'Antoni was put into an tough spot when he took the reigns of this Lakers squad. Because of the flirtation with Phil Jackson prior to his arrival, his legs were cut out from under him before he even coached a game, certainly with the fans, and quite possibly with his own players as well. He dealt with the injuries and was the exact opposite of the coach so many pegged him as. Known as a "system" guy, D'Antoni tried everything to get the Lakers going in the right direction, but nothing worked. In the end, MDA's year was a failure. His short rotations left the Lakers severely lacking in energy for winnable contests. He failed to motivate the team to play the right way for at least the first 40 games of his tenure, and to this day, we don't know what defensive scheme, if any, the Lakers were trying to accomplish.

From the Princeton to some mutated version of a typical D'Antoni offense, the team went through several identities. It wasn't until the final forty games of the season, did they enjoy any sustained success by finishing 28-12.

Yet, remember all the projections were made when Mike Brown was still the head coach. The advanced numbers failed to remotely recognize the initial offensive strategy would fail. A sample size of five games is indeed small, but when one pays attention to all the comments made by the players, did the system legitimately have a chance of resounding success? Four future hall of famers, who in their careers have enjoyed a vast amount of freedom, did not want to be constrained by a restrictive philosophy.


In no way, am I claiming advanced statistics do not have a valuable place in basketball analysis. Merely, the point is one should understand their limitations. Several years ago, David Friedman stated this succinctly:

The data being used by "stats gurus" is incomplete and often inaccurate. Statistics can be a "powerful tool" (in Roberts' words) to help to understand basketball but they do not provide definitive answers: the player rankings produce by Berri or John Hollinger do not represent some absolute, objective reality but merely reflect the biases and limitations inherent in the formulas that Berri and Hollinger invented.

Next week's piece will breakdown WOW's projection for the Pelicans in detail. We'll attempt to examine their analysis and give a more positive forecast.