NBA Week in Review 18: Buyout Season

Now that the trade deadline has passed, playoff rosters are nearly set. Players had to have been cut by Tuesday if they were to be eligible to join a playoff roster of a new team. And after a few months of experimentation, we should have a good idea of what these NBA playoff teams will look like. Meanwhile, the lottery-bound teams are dreaming of Ben Simmons and wondering what the other great prospects are. But the season is still going strong and a lot of great basketball has happened the past week.

Moving the Three-Point Line

Last week, Mark Cuban suggested that the NBA should move back the 3PT line to induce more midrange shots and open up the floor with better spacing. There were already a lot of reactions to this suggestion, and it’s certainly not a new one. But I’d like to frame this argument with a different point of view. Every fan knows that the three-point rate is skyrocketing, but what’s interesting is that the league experienced a large spike when the line was shortened to 22 feet and it took a few years before the league reached those levels again.

The change was only one foot and nine inches at the most, and it was unchanged at the corners. Thus, if a similar change were to occur, I’d expect three-point volume, and accuracy too, would dip for a while but would ultimately recover. It’s too valuable of a shot. The key is finding the ultimate balancing point where it’s too tough for most NBA players to hit consistently with an accuracy where midrange jumpers are more competitive in value — 3PTers are too valuable now and that’s why the rates will continue to rise for a while. But there will always be long two-pointers and no-man’s land will increase in size with a longer three-point distance.

Moving back the line would have many effects, and how far back it “should” be moved depends on what you want. It’d be tough to force teams back to early 1990’s levels of three-point shooting without a ridiculously far line, and even then it’s hard to gauge how players will respond and ultimately adapt. Plus, there’s not enough space along the sides —  it’s tougher to make a change like that but with a wider court players would have more attack angles. This is not a way to deal with Stephen Curry because his range is seemingly limitless, but there’s a real question right now of what we want the NBA to look like and how three-point reliant the league should be. Changes to the game are natural; basketball is constantly evolving. We should just be careful with what we actually want to do.

Greatest Game Ever?

The Oklahoma City Thunder and the Warriors played an instant classic last Saturday with one of the most amazing game-winning shots you’ll ever see not just for its rarity but because of its inevitability. We immediately had breakdowns and opinions about how it was the greatest game of the season and, perhaps, ever. I just want to make sure everyone is appreciating this season. We are witnessing something special now, and instead of squabbling about how three-pointers are ruining the game and how players of the past could do better while running uphill in the snow let’s just watch and enjoy this. Look at these shots he’s taking — he’s nearly unprecedented in NBA history. The only player I can remember being so comfortable from way beyond the three-point line is Gilbert “Agent Zero” Arenas. In fact, he had a similar game-winner: a super-long pull-up three against Minnesota. He was a fun player for a brief flash, but he peaked early and his career disintegrated. I do not imagine that’ll happen to Curry, but enjoy the great moments while they’re happening. The future is not sentimental.

Under the Hood

The Utah Jazz have been juggling the point guard position ever since Deron Williams left, but they were gifted a long-term solution at the shooting guard spot with Rodney Hood. Although he’s played a bit of small forward, he’s spent most of his time at the 2. Recently there was an article comparing him to James Harden, which is almost certainly an overstatement of his offensive potential but there are a few similarities: they’re both lefties with a knack for scoring and playmaking with porous defense. The defensive concerns are real, and they’re corroborated by just about every half-decent stat out there, like RPM where he’s a pretty awful -1.87 on defense. However, his offensive value outweighs the costs there.

Rodney Hood is more than a role player on the other end of the court. He’s adept at driving, like this play where he pulls the ball out so he can zip by someone to the rim and finish with a defender right there. He’s a good passer too, so when he drives opportunities are created for others. Since he’s a multi-faceted player, you have to be wary of his pull-up jumpers too — he draws a foul here while driving. Few players make more pull-up three-pointers. Of course, since it’s 2016 his three-point shooting is a critical skill. Nearly half his field goals are from behind the arc, and he’s hit 36.5% of them both this season and the last. He’s a good catch-and-shoot artist, which will ensure a job for him in the league for a long time. I doubt he’ll become a star, but he’s a good secondary piece and hopefully his defense can improve.

A Time to Buy: Kevin Martin

Kevin Martin didn’t fit with Minnesota’s long-term plans, and he’s not even the most experienced veteran, so it’s no surprise the team bought him out. But now that he’s near the twilight stages of his career, can we figure out who exactly he was all these years? Was this efficient sniper with a herky-jerky motion underrated or was he just a gunner with no defense? Because he was stranded in Sacramento for the first part of his career, few people noticed. He was a 20+ points per game scorer with a TS% north of 60 several times. The secret was his combination of free throws and three-pointers — he had the highest Morey Index[1. (FTA+3PA)/FGA.] in 2009, for instance, and he’s usually among the league leaders. His unique shot release form lent itself to drawing in defenders and faking them out well; he could naturally extend his arms into defenders. As a career 87% shooter from the line, this was a devastating weapon.

As a precursor, Houston picked up Kevin Martin and he had one successful full season there as a James Harden prototype, but his 2012 season was disappointing and sent off course due to a shoulder injury. Since then he’s been a mere role player and hasn’t enjoyed the same success he had before, except for a some moments in Oklahoma City where he was actually replacing James Harden as their third scorer. Yet he still was not a highly coveted piece, and in retrospect perhaps there was a reason his trade value was so low, even before injuries and age slowed him down. While RAPM is by no means perfect, he’s had a long career and in a 15-year model he was seen as a negative thanks to an awful -3.57 mark on defense — that’s points per 100 possessions, and it’s well below the average of a rookie or even low minute bench players. Plus, as Tom Haberstroh discovered, he’s secretly one of the worst “clutch” players in NBA history. I have also never seen him as a good playmaker, and his assist numbers have never been outstanding. He’s great as an end-point on a play, as he’s still able to score fairly efficiently. But there’s more to basketball than that. There are all the small actions on the court that influence and shape the rest of the play, and Kevin Martin is not a positive there — especially on defense. But he can still shoot and he’ll find suitors for that skill.

Reminder that Stephen Curry Is Magic

I don’t think everyone quite understands just how ludicrous Curry’s shooting is. We’ve all gone crazy over, say, LeBron James in Miami or Kevin Durant at his best, but Curry’s in a different league. Visually, this is more apparent. Via the graph below, looking at every season since 1974 with at least 1500 minutes played, Curry is an outlier. Players with his efficiency at the best are usually close to league average shot usage, and players with his shot usage have at best a relative TS% about half of what Curry has now. Simply put, his lead over everyone else in terms of volume and efficiency is rare not just in the NBA but in all sports.

The Memphis Malaysian Sun Bears

With Marc Gasol out for the season and a roster full of new parts, the Grizzlies do not look like the same team. They’re starting Zach Randolph next to the svelte Brandan Wright or P.J. Hairston and shooting more from outside the arc than normal — since Marc has been out, their three-pointers attempted per game have increased to 22.9 from 17.7. Naturally, because this is Memphis, they’re still winning games, grinding them out, because that’s what the Grizzlies do.

However, the Grizzlies have lost their bite, as they’ve taken advantage of a weak schedule — they’re as fearsome as Malaysian sun bears now. It’s unclear what the future of the franchise will be. They’re once again a pretty decent team that will likely make the playoffs, but they have virtually no shot at a title and older core. Losing Marc Gasol didn’t actually derail the season; this has been a long time coming. But at least we get to see a chaotic puzzle of a team with Matt Barnes, Tony Allen, Zach Randolph, Chris Andersen, and Lance Stephenson.

MVP Index

There are an innumerable amount of NBA metrics out there, but when they try to translate per unit value to a total season value they usually take a similar path: calculate the value of someone using their minutes and a theoretical replacement level value. But this can overrate high minute, mediocre players who simply play for a long time, and it doesn’t properly capture the impact of guys who play at a truly elite level. Thus, I decided to calibrate a different measure based on MVP shares.

Using minutes and my historic box score measure (HBox), I built a simple regression model to predict MVP shares from 1981 to 2015 with every player with at least 1800 minutes included[2. I used the glm function in R with a logit link.]. An MVP share is based on the proportion of MVP points you received compared to the total possible points — it’s leagues better than pure MVPs because there’s more information. Also, the most predictive version was a product of HBox and minutes; there was no replacement level trickery or any other adjustments. But this was just an intermediary step. I’m also not doing this purely for MVP prediction, so I don’t use factors like points per game (voters love scoring) and team wins like Kevin Pelton did recently. This is purely about performance.

Once I have a model for MVP shares, I can create an MVP index. This is a way to rank players by elite performance. However, there’s one important tweak: the functional form I used is non-linear and flattens out as higher values. Consequently, the very best seasons are underrated because there’s a cap on MVP shares — you can’t do better than unanimous. You can see the form below; it’s the infamous S-curve.

My solution is to essentially extend the linear portion of the model past the theoretical limit of one. In practice, once an MVP share is greater than 0.5, then it becomes a line based on the linear portion of the model. This is a better method for comparing high impact seasons with lower impact ones. One major advantage is that players do not receive negative credit, like negative win shares. Lowly ranked seasons are essentially scored as zero rather than dragging a player down. Finally, the last step is to transform the results into something more readable for most NBA fans. I simply multiplied everything by 70 so that a near perfect season would be rated as 100 — it’s a 0 to 100 scale, which should be intuitive to people[3. Here’s the final formula: 65*ifelse(1/(1+exp(-(-6.453e+00 +HBox*MP*.0003563)))>.5, -.974211+.244315*HBox*MP/3000, 1/(1+exp(-(-6.453e+00 +HBox*MP*.0003563)))). The ifelse statement makes it messy, but it’s pretty basic at its core. I also prorated the lockout seasons to 82 games.].

An example of the results are shown below. LeBron James has the greatest season ever by this box score method, and it’s not particularly close. He has several of the best seasons, in fact, and as strange as it sounds Michael Jordan “only” has a few. But none of the names are out of place with legends like Tim Duncan and Kareem Abdul-Jabbar sharing the same table. The oddest inclusion is maybe Grant Hill, but at his best he was a special player. There will never be one fool-proof method of ranking players, but we can at least improve on our previous ways.

Table: players ranked by MVP Index, 1974-2015

As a bonus, I have career values too. This is a rough measure of who’s had the best career statistically by HBox, and LeBron James already has a ridiculously huge lead. He’s the only player with a season rated at 100, and as you can see there are few players with that much in their entire careers[4. You may notice Kobe Bryant is not one of those players, and I fear mentioning his ultimate placement on this list because the Laker-Kobe cult is strong. He’s high, but he’s outside the top 25 group. Don’t kill me.]. For another surprise, Magic Johnson is ranked over Michael Jordan by a hair — HBox loves Magic’s versatile stats and, especially, high assists. John Stockton is ranked highly as well despite this measure’s emphasis on peak play rather than longevity. Stockton statistically is intriguing because he has some of the highest assist rates ever, was efficient, and collected large number of steals. Even his late career RAPM numbers are excellent. Were we all underrating him? By the way, Bill Walton’s career MVP index is 28.8, which is higher than the likes of Chris Webber and Chauncey Billups but lower than guys like Patrick Ewing and Gary Payton. Walton is one of the toughest players to rank, but that placing feels correct if one values peak seasons highly.

Table: career MVP index values, 1974 to 2016

By linking MVP performance to box score production, one can get a better sense of how the truly elite seasons separate themselves from the pack. Plus, there are other interesting conclusions from studying the numbers like this. For instance, the players with the greatest difference between their MVP shares and their MVP HBox production at both ends of the spectrum were Kevin Garnett in 2005 (i.e. he “should” have won) and Allen Iverson in 2001 (i.e. the guy who “should” have lost.) While everyone knows the case against Iverson, few people bring up the fact that Garnett had an exquisite individual season but was basically disqualified because his teammates fell apart and they missed the playoffs by a hair, thus leading to one of the most surprising MVP winners ever in Steve Nash. But because of the media’s biased subjective opinions on what MVPs look like, Garnett was ignored that season. The numbers merely serve as a guide — they should never be trusted fully, but we can make better decisions when we’re informed and consider all available evidence.

Add us as a preferred source on Google

Follow