Today is our first post from Andreas Shepard. Andreas is a longtime sports analytics buff. A strategy consultant by trade, he lives and works in Chicago, where he (like everyone else) is excited that Derrick Rose is finally back.
It’s pretty obvious that Stephen Curry is an excellent three-point shooter.
Last season Curry set a new single season record by making 272 three-pointers, and he didn’t break the record on volume alone: among players with more than 15 attempts, Curry finished third in the NBA in three-point percentage. This rare combination of efficiency and volume was on display when he shot 11-13 on threes during a game against the Knicks. In the 34 year history of the NBA three-pointer, there have only been 40 games where a player shot 70% or better on 11+ three-point attempts, and Curry’s performance is right at the top of that list.
When you put it all together, Curry’s season sure seemed like one of the greatest long-distance shooting performances in NBA history. But does that idea hold up to further scrutiny?
There are two dimensions to shooting greatness: efficiency and volume. To start our analysis, here is every season that has ever qualified for the 3PT% leaderboards (requirements here) graphed in terms of both efficiency (3PT%) and volume (3PT attempts).
Already you can see how unique Curry’s season was. Of all the players with nearly as many as his 600 attempts, he had the highest shooting percentage by a substantial margin. And while there have been players who shot a better percentage from deep, none was able to keep it up for nearly as many shots as he was.
Still, we need to come up with a way to compare all these seasons. One way to combine efficiency and volume into a single number is by calculating how much net benefit a player’s three-point shooting added to his team. When a player takes a shot, he has used up a possession for his team and is penalized one point. If that shot goes in, it’s worth the three points that get added to the scoreboard. Thus the formula to combine efficiency and volume into one number is 3 * 3PM – 1 * 3PA. Since 3PM = 3PA * 3P%, this incorporates both factors from the chart above. Let’s call this measure three-point production.
Here are the top 20 seasons according to three-point production:
|Rank||Player season||3PT production|
|1||Stephen Curry 2012-13||216|
|2||Glen Rice 1996-97||181|
|3||Dennis Scott 1995-96||173|
|4||Peja Stojakovic 2007-08||169|
|5||Peja Stojakovic 2003-04||166|
|6||Dana Barros 1994-95||166|
|7||Joe Johnson 2004-05||161|
|8||Mitch Richmond 1995-96||160|
|9||Ray Allen 2001-02||159|
|10||Steve Nash 2007-08||156|
|11||Ray Allen 2005-06||154|
|12||Damon Jones 2004-05||154|
|13||Kyle Korver 2012-13||153|
|14||Reggie Miller 1996-97||151|
|15||Dale Ellis 1988-89||147|
|16||Raja Bell 2005-06||145|
|17||Ray Allen 2000-01||139|
|18||Tim Legler 1995-96||139|
|19||Mitch Richmond 1996-97||135|
|20||Leandro Barbosa 2006-07||132|
According to this measure, Steph Curry’s 2012-13 season was the most productive three-point season in NBA history. But was it the most valuable?
The difference between raw production and value is that value is measured relative to the available alternatives. If Curry had vanished from the Warriors last year, his 600 three point attempts would not have vanished with him. They would have been taken by somebody else [author's note: in practice, some of them would have been replaced by turnovers, free throws, or two point shots, but for simplicity’s sake we’ll pretend that all 600 shots would have remained]. The difficulty in comparing seasons over time is that the “somebody else” taking Curry’s shot’s in 2012-2013 would have been different than the “somebody else” taking, say, Dale Ellis’ shots in 1988-89. For one thing, Ellis’ shots would probably have been taken by a worse shooter. Three point shooting percentages have been trending up league-wide over time, especially compared to the early years of the three-pointer.
Note: the three seasons highlighted in yellow correspond to the years the NBA played with a shortened three-point arc.
To get to value, we need to compare each player’s production to the production their team would have had if it had given those shots to a replacement player. This accounts for the fact that there are more good shooters in the league than there used to be, so losing Curry’s shooting last year wouldn’t leave us with quite as bad a replacement as losing Ellis’s shooting 25 years ago. It will also account for the shorter arc in the mid-90s, since shooting percentages in those years (and therefore the expected production of a replacement player) are a bit higher. The formula we’ll use for the comparison is three point production (from above), minus the three point production of a replacement player taking the same number of shots.
There are many different ways to define what a “replacement player” would be in a given season, but for the sake of simplicity we’ll evaluate two cases. In the first case we’ll assume the replacement player shoots at the league average 3PT% for that year. In the second case we’ll assume the replacement player shoots at the 25th percentile 3PT% (which means that 75% of the players in the league shoot better). This aligns with the idea that a replacement player will likely be below average.
|Rank||Player season||3PT value vs avg||Player season||3PT value vs 25th percentile|
|1||Stephen Curry 2012-13||170.2||Stephen Curry 2012-13||200.3|
|2||Dale Ellis 1988-89||158.0||Peja Stojakovic 2003-04||172.1|
|3||Glen Rice 1996-97||145.7||Dale Ellis 1988-89||168.2|
|4||Peja Stojakovic 2003-04||143.0||Glen Rice 1996-97||152.2|
|5||Joe Johnson 2004-05||136.3||Ray Allen 2001-02||149.6|
|6||Dana Barros 1994-95||133.8||Joe Johnson 2004-05||145.7|
|7||Ray Allen 2001-02||126.9||Dana Barros 1994-95||144.5|
|8||Peja Stojakovic 2007-08||124.0||Peja Stojakovic 2007-08||143.8|
|9||Steve Nash 2007-08||123.3||Kyle Korver 2012-13||142.2|
|10||Kyle Korver 2012-13||121.4||Steve Nash 2007-08||137.7|
|11||Damon Jones 2004-05||119.3||Wesley Person 1997-98||135.7|
|12||Tim Legler 1995-96||114.4||Damon Jones 2004-05||132.4|
|13||Wesley Person 1997-98||112.5||Ray Allen 2005-06||130.7|
|14||Raja Bell 2005-06||111.6||Raja Bell 2005-06||129.1|
|15||Ray Allen 2000-01||110.7||Ray Allen 2000-01||121.9|
|16||Michael Redd 2002-03||109.9||Danny Ainge 1987-88||120.0|
|17||Dennis Scott 1995-96||109.8||Dennis Scott 1995-96||119.6|
|18||Mitch Richmond 1995-96||108.2||Steve Nash 2001-02||118.9|
|19||Reggie Miller* 1996-97||108.0||Tim Legler 1995-96||118.2|
|20||Danny Ainge 1987-88||105.3||Channing Frye 2009-10||118.1|
You can see the effect of accounting for value on a player like Dale Ellis, whose production in 1988-89 becomes much more impressive once we account for how much better he was than the rest of the league that season. But no matter which way you slice it, Curry’s season remains the greatest three point performance of all time, and by a pretty comfortable margin. The only real question is whether he’ll remain healthy and productive enough to establish himself as the greatest career three point shooter ever. He’s already well on his way:
|Rank||Player||3PT production||Value vs avg||Value vs 25th percentile||Games|
Already Curry has had one of the top 25 three point shooting careers ever, and he’s only played a little over three seasons worth of games. Everyone ahead of him has played two to five times as many games! If Curry keeps up his current pace, he’ll be at the top of the leaderboard in about six more seasons. At this point, the only things that seem capable of stopping him are his balky ankles. As fans of the game, we should hope they hold up; it’s not often that you get the privilege of witnessing such a transcendent talent.