Here is a non-controversial statement: Stephen Curry is having one of the greatest seasons in NBA history. But just how great depends on who—and more importantly, how—you ask.
One school of thought is best described as the time machine approach. Basically, this is an imaginative exercise, in which we stuff Curry into Doc Emmett's souped-up DeLorean, set the controls for a previous basketball era—say, the 1960s NBA—and ask a simple question: how would the Golden State guard and reigning league MVP fare? Basketball legend Oscar Robertson recently said not so well, a "back in my day" argument that seems nonsensical on its face. Great ball-handling, phantasmic quickness, and limitless shooting range have always been valuable basketball attributes, and it's hard to imagine Curry doing less with them against Robertson's Cincinnati Royals than he has against the rest of today's NBA.
On the other hand, what if Curry grew up in the same era as Robertson? What if he had 1950s nutrition, crappy shoes, and booty shorts? What if he had to battle institutionalized racism just to be a part of a semi-integrated league, and didn't have the benefit of modern sports medicine and training techniques to rebuild his injured ankles? Given this context, would Curry still have the same insane game? Would he even be allowed to showcase it? The best we can do is guess at answers, and even those guesses aren't worth much.
The second major way to assess historical greatness is the approach hinted at by Hall of Famer Isiah Thomas: judge players within the context of their own eras, against their contemporaries, and then use those judgements as a basis for cross-era comparisons. This makes a lot more sense. While we can only take a rough stab at how Curry would have fared in the pre-three-point era—or how Robertson would have done in today's high-flying, buffed-up NBA—we know how players actually performed on the floor during their careers, at least to the extent that the statistical record permits.
In other words, the fairest way to compare players is to compare their numbers, while making sure to account for how the way the game was played in different eras—rules, playstyle—affects those numbers.
Looking at players this way—as products of their own time, with corresponding on-court production—helps correct for the "funny money" stats of yesteryear. For example, in 1961-62, Robertson famously averaged a triple-double. That's also the year Wilt Chamberlain averaged 50 points and 25 rebounds per game. Both feats are incredible, and unlikely to be repeated.
But both feats—all feats, actually—should be considered in context, and the game was just different back then. NBA teams shot a lot more often—about 23 field goal attempts more per game than today—and were worse at putting the ball in the basket. The net result was an additional 25 rebound opportunities per game to go around, just from errant field goal attempts. And that's before accounting for a greater number of free throw attempts, and the fact that players in 1961-62 shot a lower percentage from the free throw line. Oh, also no one kept track of turnovers, either. Acknowledging all this doesn't take away from Robertson and Chamberlain's impressive seasons. It just makes it easier to understand in a more comprehensive way.
But back to Curry's season for the ages—where does it rank all-time, and how can we determine that, taking the context of today's NBA into account? A few years ago, I worked up an overall player rating metric I call "Player Production Average," or PPA. Without going into too much detail—you can read more here—PPA is built to account for how individual players contribute to their teams winning and losing and does so by applying weights to different statistical categories, each in proportion to how that category contributes to wins and losses. It's similar to other all-in-one rating systems such as John Hollinger's Player Efficiency Rating (PER), or David Berri's Wins Produced.
PPA includes features of each—it's pace neutral (like PER), and accounts for defense (like Wins Produced, albeit differently). Where PPA differs significantly is in its handling of offensive efficiency. In my analysis, PER sets the bar too low—shooting more can increase a player's rating with a conversion rate as low as 28 percent—while Wins Produced sets the bar too high in its assessment that a player adds nothing with his shooting from the floor unless he shoots better than 50 percent. PPA hits the Baby Bear "just right" spot with a "break even" level that floats a bit year to year depending on league efficiency in that season.
Because of the above, PPA is a useful tool for measuring how players perform versus their peers, within the context of their own seasons and unique rule sets. And that makes it great for cross-generational comparisons, too. Zone defense and hand-checking rules changed from Michael Jordan's era to Curry's? Doesn't matter. PPA measures the on-court impact of each—as compared to the players they competed against in their respective seasons.
We can still talk about how today's players are almost certainly better in objective terms than those of previous eras, because today's players get to build on the legacies of previous generations, as well as enjoy the fruits of improved nutrition, training, medical science, equipment, travel, coaching, scouting reports, and so on. And we can discuss how defensive rules changes have merely restored the league's scoring environment to what it was during Jordan's heyday; few casual basketball observers seem to realize this, but it's still slightly more difficult to score in 2015-16 than it was during Jordan's career.
We can even argue about whether or not Curry is actually a more dangerous offensive weapon than even Jordan because his freakish shooting range—and Curry has a higher effective field goal percentage on shots from 28 or more feet than a dozen of this year's All-Stars have from inside three—would make vintage Detroit Pistons-style "Jordan Rules" defense impossible. Back in the day, teams could foul Jordan—sometimes hard—because His Airness wanted to attack the rim. Curry shoots better closely guarded than most players do wide open. He's also lethal to double-teams and pressure defenses because he's an effective (and willing) passer. And oh yeah, he does attack the basket with dribble drives and sublime cuts, he's among the league's best finishers at the rim, and he shoots better than 90 percent from the free throw line. While stylistically different, Curry is as close to unguardable as Jordan was at his peak.
Sorry. Got carried away. Curry has that effect! Point is, we can still have all of these time machine-style debates. And we should, because they're fun. But through PPA, and other measures like it, we can measure Curry's dominance, then compare that dominance to Jordan's and David Robinson's and everyone else's.
So, to get to our original question: just how well is Curry doing? My PPA database starts in 1977-78. (Sorry, Big O). And since that time, he's having The Greatest Individual Season Ever. Not one of the best. The Best. Better than any season from Jordan. Better than than any from Lebron James. Better than any from Larry Bird or Magic Johnson or Charles Barkley or Isiah Thomas. No qualifiers, no hedging. Watch Curry while you can: we're witnessing history.
Here's a look at the top 25 individual seasons since 1977-78. In PPA each season, 100 is average, 150 is typically the low end for All-Stars, 200 or better is All-NBA level, and it usually takes 225 or higher to be an MVP candidate.
Look again: over the years, the game's greatest players have produced standout seasons that fall into the same general range, from David Robinson's 245 in 1993-94 to Kareem Abdul-Jabbar's 262 in 1977-78 to LeBron James' 282 in 2012-13. And now realize that even in this august group, Curry is an outlier, the season-long equivalent of a halfcourt swish. Through Golden State's 70th game of the season, his PPA is a whopping 326.
How outlandish is this? By way of comparison, Kawhi Leonard's PPA through 70 games is a hefty 266, which is the sixth best mark since 1977-78. The San Antonio forward is having a pantheon season, and he's also a footnote. His performance this year fits right in with the other basketball deities through history; it's just that the Curry is a god among gods. Also note that the second-best season since 1977-78. right behind Lebron's 282 in 2012-13, is Curry's 277 from last season. Pretty nice run.
Or, think of this: the gap between Curry this year and the current best season (44 PPA points) is greater than the gap between the current best and 25th best (37 PPA points).
But wait, you may be wondering: won't the last 16 games of the season potentially drop Curry from such lofty heights? It's possible, but improbable. The history of the NBA's elite performers is that their production changes little from this point in the season; in fact, it doesn't change much after just 20 games. The biggest movers from their team's 70th game to the end of the regular season were James, Curry and Davis. Lebron and Curry added seven points to their PPA scores; Davis dropped seven. Among the top 25 in my database, in which we have game-by-game information, the average PPA change was plus-one.
In other words, it would take a catastrophic and unprecedented change in production for Curry's PPA to dip below 300 over his final 16 games. The far more probable outcome is that Curry will keep kicking butt, and in historic fashion. So add the Golden State Warriors to your calendar, make a point of watching, and let Oscar Robinson know he should do the same. No matter the era, we've never seen anything like this.