As expected, FiveThirtyEight czar Nate Silver was hot at the scene of another topical NBA story with a long, data-heavy feature last week at his relaunched site. This time around, the story was about Kyrie Irving’s future in a Cleveland Cavaliers uniform and whether the team should offer him a max contract this summer. While Silver’s article was valuable in theory, it misfired badly with its token data.
Irving’s future with the Cavs has seemingly been a hot topic ever since the 2011 draft. Now that he has completed his third NBA season, this summer represents a key opportunity where the Cavs can offer him a maximum long-term agreement. If he agreed to the extension, that deal would keep the two-time All-Star in town through at least 2019-20. The max deal would start with the 2015-16 season and would guarantee approximately $90 million over five years.
Silver attempted to chime in with historical data to answer this very, very difficult question: When should NBA teams offer maximum contract extensions to their soon-to-be fourth-year stars? That’s not an easy research project. It requires a keen understanding of the NBA’s financial landscape, the weight of its stars and how top picks have performed in the past.
I’ll say this much immediately: The Cavs should be weighing this decision heavily. It will transform the franchise and lock in either 25% or 30% of their salary cap each season until the summer of 2020. You’d be an ignorant fool if you didn’t do your due diligence on such a long-term agreement, let alone one with an undoubtedly overrated (even last October!) scoring-dependent star.
I’ll also say this: I’m a gigantic fan of Silver’s. I wrote as much in a fan-boy letter back in 2008, just before his election writings propelled the Chicago-raised sabermetrician into the mainstream. Specifically, I’m an admirer of how he has brought “aggregation” to the limelight, as I wrote in my 2013 bracketology preview. No matter your thoughts on FiveThirtyEight’s success so far, Silver has done a lot to transform the way sports and politics buffs think about the world.
But what did Silver’s research say on this topic? His first chart showed Wins Produced by draft picks throughout their careers. The chart eloquently shows how No. 1 picks are usually much better and have a better overall career arch than No. 2-5 picks, No. 6-14 picks and No. 15-30 picks, respectively. That much shouldn’t be surprising.
Essentially, this chart below is the real meat of Silver’s analysis. Later on, he delves into top-20 and top-5 percentiles of each year’s extension class, using top-5 as a heuristic for extensions. I generally agree with his conservative spending approach. But this chart below is where I want to nit-pick. This chart shows the projected Win Share value for an average No. 1 overall pick at each point along his career path. As one can see, the value disappears under the heavy financial burden1 of the max deal starting with his fifth season.
As Silver points out: This chart does a phenomenal job of showing how NBA teams shouldn’t automatically extend every No. 1 pick. That’s a given. Some selections are busts. Some get hurt. And most importantly, over time, everyone gets worse eventually. Which leads me to the topic that Silver never really touched specifically: Age.
Yes, he wrote about “years of experience” and tried to pass that off as a model for NBA aging curves. But that’s not exactly accurate when comparing different eras. Not when Kyrie Irving was the third-youngest No. 1 overall draft pick ever, barely younger than Anthony Davis (2012) and only older than LeBron James (2003) and Dwight Howard (2004). Not when the average July 1 age of No. 1 picks from 1985-2002 (first 18 years of Silver’s data) was 21.5, but it’s 19.7 since 2003. Top picks are way younger in the latter half of this data set and that warranted nary a mention.
One can easily Google search “NBA age curves” and learn tons and tons about the models of how basketball players age. There is a ton to learn here and I’d be remiss if I didn’t acknowledge some of the really good existing research. Articles exist at Bloomberg Sports, Wages of Wins Journal, Basketball Prospectus, Basketball-Reference and so many other good sites. The forums at APBRMetrics are filled with posts (like here and here) on this topic. No doubt the research is a bit more robust in baseball, of course, but there’s still plenty out there.
More specifically, tons of articles have been written about age vis-à-vis “potential” for draft prospects. Two of my favorite articles on this topic are from ESPN Insider’s Kevin Pelton and Canis Hoops’ vjl110. For Pelton, age makes up 40 percent of his rookie Wins Above Replacement Player projections, with translated performance taking up the rest. He wrote: “History shows that age is key to understanding how college players end up developing in the NBA.” Age matters, not arbitrary starting points of NBA careers.
Silver decided to use the wrong principal data point when trying to make an argument about Kyrie Irving’s potential. Certainly, it would seem initially valuable to look at season-by-season progressions when discussing max contract extensions. That’s the easiest route. But it just isn’t relevant to compare how former top picks like Derrick Coleman, Michael Olowokandi or Kenyon Martin, all 23 years old by their NBA debuts, regressed around the time of their fourth seasons2. Irving won’t even turn 23 until next spring.
In the landscape of point guards, there’s no doubt that age, not experience, is a major factor in Irving’s widely regarded upside. He’s 21 months younger than Jrue Holiday, 20 months younger than Damian Lillard, 18 months younger than John Wall and 17 months younger than Ricky Rubio. In fact, he’s actually five months younger than Michael Carter-Williams, this season’s Rookie of the Year. He’s also younger than a couple of likely first-round picks in this year’s draft.
A) Payne: February 19, 1991. B) McDermott: January 3, 1992. C) Irving: March 23, 1992. D) Napier: July 14, 1991. E) Leslie: June 25, 1991.
— Jacob Rosen (@WFNYJacob) May 28, 2014
Of course, this is not all to say that the Cavs should necessarily extend Irving just because he’s younger and Nate Silver made an oversight with his data. There are issues with Irving. The Cavs have yet to make any type of leap in the win column. He’s hardly played defense at a replacement NBA level in three seasons. And most damningly, he’s actually regressed offensively since the 2013 All-Star Game.
The larger point I’m making is that the NBA community has only seen a few examples like Irving in its history due to his remarkable age. He’s one of only 33 players to ever have 5,000 minutes played through their age-21 season3. Among those players, he’s sixth in PER, behind only Shaq, LeBron, Chris Paul, McGrady and Durant. Irving is the first player in NBA history to have three seasons with at least 1,000 minutes and a 20 PER through age 21. We don’t know yet how good he might become. He’s still so very young.
That’s the conundrum of Kyrie Irving. We have a lot of meaningful data, as counted by three seasons, 181 games and 6,102 minutes. But even with that, there’s so much projecting and guessing to know what kind of player he’ll be for these next six seasons. Even still, he’ll only be 28 in the summer of 2020. By then, he could actually turn into a top-10 player, as #NBARank predicted before this season. Or he could flop. More likely than not, the Cavs will need to take that risk. And his favorable age will be a major reason why.
(AP Photo/Mark Duncan)
- One could point out that Silver’s article also assumes that Irving’s value, even if overpaid, is easily replaceable on the open market. That’s definitely not the case in the NBA system. Irving likely also brings in entertainment value to his basketball city, regardless of his exact value on the court. [↩]
- Coleman’s best PER/minutes seasons were his third and fourth. Olowokandi’s best season was his fourth. Martin’s best seasons were his third, fourth and fifth [↩]
- Basketball-Reference uses February 1 as the endpoint for age-seasons. [↩]