As a refresher, we just introduced a new player rating metric, RAPTOR — the Robust Algorithm (using) Player Tracking (and) On/Off Ratings. As part of that, we needed to create historical RAPTOR estimates for players who would show up as comparisons for current stars. These estimates were built by figuring out how the limited data kept in earlier eras (box score plus team data and RPM for 2001-2013, and just box score/team data from 1977-2000) relates to modern RAPTOR ratings.
These ratings are not perfect. And as FiveThirtyEight editor-in-chief Nate Silver noted in his RAPTOR explainer, they raise an interesting philosophical conundrum: Does the act of retrofitting modern ratings to older players actually measure those players’ true, contemporary value … or does it simply find the historical players whose games would best fit into the modern NBA? (I don’t have a great answer for that; the truth is probably somewhere in between.)
Whoâs better, Kyrie or Kemba?
But they do give us a brand-new set of historical stats for every player in every season going back to the NBA-ABA merger in 1976. So let’s take our historical RAPTOR ratings at face value and assume they function as solid contemporary measures of performance. Who, then, is the GOAT? LeBron? Jordan? Kobe? Someone else?
Instead of simply adding up WAR, like we did for the six-year period we wrote about last week, we should make an adjustment to better handle the scope of entire careers. To that end, I’m borrowing a page from baseball’s book and adopting the JAWS Hall of Fame metric, which averages together a player’s career WAR with his best seven (nonconsecutive) seasons to balance between total and peak value. If we do this with RAPTOR, here are the leaders since 1973-74:
|RAPTOR +/- per 100||WAR|
|Player||OFF||DEF||TOT||CAREER||PEAK 7 YRS||JAWS|
This is not a definitive ranking, or even necessarily the best way to balance between peak and career WAR. But it’s a starting point for conversation, and it gives us a general idea of the kinds of players RAPTOR likes. (Feel free to download our RAPTOR data and make your own ranking!) Let’s break down some of this list’s most eye-catching results, one by one:
MJ over LeBron? I want to make one thing clear: I have long been a LeBron James backer, even in The King’s darkest days. And RAPTOR plus/minus thinks James was slightly better on a per-possession rate in 2008-09, his best career season, than Jordan was in 1990-91. (LeBron’s playoff performance that year was legendary even in defeat.) But of the top eight RAPTOR WAR seasons by either Michael Jordan or James, seven belong to MJ. Jordan routinely played many more minutes in his best seasons than James did, and he was usually more effective per possession as well — particularly on defense, where LeBron was never as good as Peak MJ and has fallen off quite a bit in recent seasons.
John Stockton, No. 3????? This is probably the most shocking item on the list. Many observers don’t even consider Stockton a top-three point guard, much less third on an overall list of players. I’m not saying I fully buy this, either, and maybe the JAWS approach is still giving a little too much credit to Stockton’s extremely long career (19 seasons and 47,764 minutes). But there’s also reason to believe that traditional analyses don’t give Stockton anywhere near as much credit as he deserves. First, he was ridiculously durable, playing in 1,504 out of a possible 1,526 games in his career (98.6 percent). He also excelled at many of the aspects of NBA offense that RAPTOR adores, including passing (he’s the all-time leader in assist rate), shooting efficiency (his career true shooting percentage ranks 14th) and floor-spacing. To the latter point, Stockton devoted more than 20 percent of his field-goal attempts to 3-pointers in the 1990s, a rate that (while low) wouldn’t be totally out of place for a point guard in 2019.1 And Stockton was a very good defender, garnering All-Defense honors five times and finishing seventh all-time in steal rate. Maybe Stockton is just a player whose style would stand out more in the modern NBA, but it also seems likely that he was far more valuable than observers realized during his career.
Chris Paul, point god. Similar to Stockton (and Jason Kidd slightly lower down the list), Chris Paul shows up as a legend in RAPTOR for his mix of hyper-efficient shooting and passing and his standout defense. I don’t think people realize that Paul has been named to the All-Defensive Team nine times in his career, and at the other end, he’s seldom had a season where his team didn’t average at least 1.2 points per possession on plays he personally was responsible for. We’ve often wondered why CP3-led teams can’t seem to get over the hump in the playoffs, and that is a fair criticism (although we are including playoff value in RAPTOR). Unlike with Stockton, Paul’s durability has also been an issue. But few players in history have ever been as statistically impressive.
Is Kobe … properly rated? In our JAWS list, Kobe Bryant ranks eighth … which happens to be exactly where Bill Simmons’ pyramid rankings had him! (Granted, there are a few pre-1976 players ahead of Kobe on Simmons’s list.) All-time rankings that rely on advanced metrics have usually struggled with Bryant, since his status as perhaps the league’s final star midrange gunslinger makes him one of the most polarizing statistical players ever. But RAPTOR actually kind of likes Kobe. Although his last few years were ugly and his defense was usually below-average in the back half of his career, Bryant was genuinely one of the best offensive threats of all-time, plus he gets extra credit for his playoff performance (the sixth-most postseason WAR since 1976).
Robinson > Olajuwon? This one is a classic when it comes to stat-based rankings. The specter of one series — Olajuwon’s epic 1995 Western Conference Finals performance against Robinson and the Spurs — will always hang over any mainstream debate between the two players. But advanced metrics have consistently rated Robinson higher, and RAPTOR is no different. Although RAPTOR thinks the majority of both players’ value came on defense, with Olajuwon’s D grading out slightly higher in their best seasons, Robinson gets the career advantage mainly for his skills on offense. There, RAPTOR says Robinson was routinely worth at least 2 points above average per 100 possessions, a threshold Olajuwon crossed only once in his entire career. In fact, RAPTOR considers Hakeem a below-average offensive contributor for most of the ’90s, aside from 1993 and his championship years of 1994 and ’95. But at least Olajuwon can say he had more career playoff WAR, beating The Admiral 27.1 to 19.6 in that regard.
Shaq too low? Shaquille O’Neal was one of the most dominant forces in NBA history, proving to be particularly unstoppable while winning Finals MVP with the Lakers in 2000, 2001 and 2002. So why doesn’t RAPTOR like Shaq more? Some of it comes down to durability and staying power: Shaq played fewer minutes than some of his rivals on the list (such as Tim Duncan, Karl Malone and Kevin Garnett), and he didn’t age very gracefully in the latter part of his career.2 But this could be another case of RAPTOR’s modern sensibilities derailing a ’90s-era center, particularly on offense. With a total lack of floor spacing and a somewhat modest assist rate even by big-man standards, Shaq’s career offensive RAPTOR might actually be somewhat generous at +2.5 if you compare him with contemporaries such as Patrick Ewing (-1.4), Alonzo Mourning (-1.0), Olajuwon (+0.2), Duncan (+1.5) or even Robinson (+1.9).
Who’s missing? If we compare our JAWS rankings with the top of a list like Simmons’s, some notable absences include Kareem Abdul-Jabbar (111.7 JAWS, though that’s missing the first seven seasons of his career), Moses Malone (99.8), Julius Erving (113.5), Isiah Thomas (108.6) and Kevin McHale (77.8). Examples of recently retired players who just missed the top 20 were Dwyane Wade (No. 22 with a JAWS of 133.0), Ray Allen (No. 24, 129.3), Manu Ginobili (No. 26, 125.1) and Paul Pierce (No. 27, 124.2).
Now that we’ve broken down the historical RAPTOR careers, let’s dive into some great RAPTOR seasons. Here are the most valuable WAR seasons since 1976:
|RAPTOR +/- per 100||WAR|
There’s a whole lot of Jordan, James and Curry atop the list — in fact, no other player invades the ranking until Paul’s 2008-09 masterpiece season lands at No. 13. And in the battle for GOAT seasons, Jordan’s 1990-91 and James’s 2008-09 are separated by fewer than a half-win at the top of the heap. The players were equally good on defense per 100 possessions, and James was slightly better offensively on a rate basis, though MJ logged more minutes. RAPTOR thinks either season is basically the gold standard for individual performances over the past four-and-a-half decades.
Finally, let’s end by looking at history’s worst seasons. We already know about Sexton, Kobi Simmons and Kevin Knox, so we’ll spare them any further ignominy. On a per-possession basis, the worst RAPTOR plus/minus seasons (minimum 500 minutes) since 1976 belong to the immortal Uwe Blab (-10.6 per 100) in 1989-90, Sharone Wright (-10.0) in 1996-97 and John Amaechi (-10.0) in 2001-02. Blab played for both the Warriors and Spurs that year, averaging 2.1 points per game and producing nearly as many turnovers (35) as made baskets (39). He’s just lucky he didn’t see enough court time to be worth less than -2.2 WAR. The “leaders” in that category are Sexton (-7.4) and Knox (-6.0) from 2018-19 and Zach LaVine (-5.4) from 2014-15, but among pre-tracking data seasons, the worst WAR seasons belong to Michael Olowokandi (-5.4) in 1999-2000, Amaechi (-5.1)3 and Marcus Fizer (-4.6) in 2000-01, Bryant “Big Country” Reeves (-4.6) in 1996-97 and Adam Morrison (-4.3) in 2006-07. (Remember when Morrison was getting hyped up as the next Larry Bird? Yikes.)
As I mentioned above, you can download all of the data I looked at for this story at FiveThirtyEight’s GitHub repository for our RAPTOR metric. Please play around with it yourself, and let me know on Twitter what your favorite great — and terrible — NBA seasons were. (And why RAPTOR might be underrating Uwe Blab’s 1989-90 campaign.) One of the great things about RAPTOR is that it’s based on publicly available data, and we want to keep that spirit alive by releasing the results for others to use for research. And who knows? Maybe the upcoming 2019-20 season will provide new entries for the all-time best and worst RAPTOR seasons — and help some of the GOATs above solidify their places on the all-time list.
Check out our latest NBA predictions.