NBA Week in Review 17: Trade Deadline

The trade deadline was mostly a dud, as there were no major moves and the biggest trades involved troubled players being shuffled around or bench pieces. But we’ll see the effects of those moves soon, and we also had a couple of historic games. It was a short week, but a lot happened — let’s take a look at the last week in the NBA.

The Siren Song of Jeff Green

There’s a so-called eye-test about the NBA where viewers are supposed to ascertain player value by watching them their own eyes, ignoring stats. Certain players are overrated by this method because they can impress physically with a few highlights but provide little overall value or production — Jeff Green is the poster-boy for this phenomenon and certainly seems to have caused more failed eye-tests than virtually anyone. He’s 6′ 9″, athletic, and long, tantalizing franchises with his potential ability as a stretch four or a big defensive small forward. But he has nevered performed consistently well in either role. He doesn’t really pick up a lot of steals, blocks, rebounds, or assists and doesn’t score at high or efficient rates.

People also have this strange desire to trade for him during the middle of the season, and as can be seen from the table below it usually doesn’t go well. Those numbers understate some of the effects too. Jeff Green’s trade was the start of Oklahoma City’s dominance. They visited the conference finals that season, and then made the finals the next season with a win percentage that would translate to 58 wins over 82 games. Even though Kendrick Perkins, who they netted in the trade, was a slog at center for them, it allowed Serge Ibaka to break-through and Kevin Durant to ascend. Jeff Green was there for the death of the Kevin Garnett-era Celtics, and his absence ignited this resurgent Brad Stevens period.

Table: mid-season trade outcomes with Jeff Green

While the Clippers situation may be desperate enough that Jeff Green will be handy, I wouldn’t be optimistic. In Memphis he was replacing Quincy Pondexter and Tayshaun Prince, and they still didn’t improve. There are other factors here, of course, but if you want to adjust for more things you’re getting into RAPM territory — he was -2.76 overall, far below average and even the average rookie, in the multi-year 2015 ratings. But this fits well with Doc Rivers’ hiring philosophy, even though I’m confused why he’s bringing Jeff Green over to replace Paul Pierce again — maybe it’ll work this time? From all the available evidence, I would not expect that.

The Markieff Gamble

The Washington Wizards, suffering through a mediocre season and clinging to the hope that they can grab Kevin Durant in the off-season, gambled on Markieff Morris, but I’m not high on this one. The glass half-full approach to him is that he’s a mobile stretch four who can play plus defense. However, he’s only had one above average year during that anomalous 2014 Phoenix season, and he’s already played nearly 350 games in his career. He’s not old, but there are — let’s put it delicately — real concerns about his maturity and work ethic. Maybe Markieff will turn it around and play well just as a slight against his old team, but I wouldn’t be surprised if he doesn’t even outperform Kris Humphries, who will probably just be waived (non-guaranteed) this summer to clear space. He hasn’t played well so far.

The Cavaliers Add Frye

The Channing Frye experience has been limited so far in Cleveland — he had to be cleared by a physical and then played sparingly few minutes — but it’s an intriguing addition. The frontcourt was already a bit of a logjam, when healthy, and Tristan Thompson is making a huge amount of money, but Frye’s teams usually do unexpectedly well when he’s on the court and his shooting fits nicely with the slashing games of LeBron James and Kyrie Irving.

Frye’s basically a personification of the reduced role people fear Kevin Love has slipped into. That could be quite useful against teams that regularly deploy big lineups, like San Antonio, but it may not be useful against the Warriors. The Cavs did not a major issue in terms of their shortage of big, perimeter defenders. LeBron James does not seem to care about defense much these days, and while that may change in the playoffs he has other responsibilities. Frye will be valuable overall, but the real magic trick is that they reduced their tax bill and won’t have Anderson Varejao’s ridiculous contract anymore[1. By the way, he would have been a powerful addition to the Clippers. They have more of a hole in the frontcourt, and he could play next to both DeAndre Jordan and Blake Griffin. Too bad — that would have made the playoffs in the west even crazier, as long as Frye is healthy and shooting well.].

No Donuts in Detroit

In a surprise, a three-way trade with Houston, Detroit, and Philadelphia was voided due to Donatas Motiejunas’ failure of a physical. This is disappointing for the Pistons, who wanted more stretch four power, and this is awkward for the Rockets[2. This is also unfair for the 76ers, who had to dump JaKarr Sampson just to make the trade work and now get nothing out of it.], who are fighting against a nightmare season and chemistry problems, but there’s a question of how good Montiejunas really is. He came into the league at 22, barely played for a while, had one decent season last year, and has been basically injured since then. Hopefully he can find a good long stretch of good health because he should be given the chance to prove how valuable he is.

Josh McBobs: Miami’s Forgotten Hope

In the distant past, the Miami Heat were trying to appease LeBron James and had to settle with a modest addition in the form of Josh McRoberts, a trendy stretch four but with the added bonus of playmaking skills. But he’s been a wreck ever since, overtaken by injuries with terrible shooting percentages. However, he had a great game last week: 19 points and 10 assists off the bench in a win versus the Hawks. With his range and passing skills, he can’t replace Chris Bosh, either replacing him off the bench or filling in as long as he’s out, but he can help the team from collapsing. Plus, he can make nifty plays like the one below. It’d be nice to see a fully rejuvenated Josh McRoberts for the Heat because they need it and the NBA audience does too.

Lillard’s All-Star Snub Tour of Destruction

I argued that excluding Lillard from the all-star game was silly at the time, and I think Lillard agrees because right after the break he destroyed the Warriors with a 51 point game. What’s remarkable is that Lillard apparently now has the highest single game DRE ever — the key ingredients here, besides the points, are the six steals and the lack of turnovers. Considering the opponent, I think his game has a reasonable argument for greatest modern day regular season game ever. It was only Golden State’s second non-asterisk win of the season, as they lost to the Bucks one day after a grueling double-overtime win over the Celtics, one game against the Mavericks without Stephen Curry, and one against the Nuggets without Draymond.

Anthony Davis: 59

In an attempt to upstage Lillard in a historic night, Anthony Davis had 59 points and 20 rebounds in a game without an overtime period. He was near that hallowed arbitrary mark of 60 points, which is a total few big men have reached — the list is just Wilt Chamberlain, David Robinson, George Mikan, Karl Malone, Shaquille O’Neal, and Tom Chambers. In fact, as was repeated last Sunday, only Davis, Shaq, and Wilt have scored 55 points in a game with 20 rebounds as well. Even though this relying heavily on arbitrary endpoints, it’s a historic game.

For Anthony’s first points, he caught a pass a few feet from the basket, leans in, and banks in a shot in an awkward yet smooth Mr. Gumby manner we’re accustomed to from Davis. Of course, the problem with trying to defend Davis is that in conjunction with his finishing skills he’s a great shooter now and can regularly drain shots like this one. Defenses have to keep track of him at all times too, or else he’ll slip to the basket and dunk high. His last points came from drawing free throws by posting up a smaller player, which helped to seal the victory for the Pelicans. If Davis’ jump shots are falling with regularity on a well-coordinated team, 50 points is a soft barrier. He just needs a more stable Pelicans team, and hopefully in the future he can get that.

Estimating Player Turnovers: 1974 to 1977

For basketball researchers who like to study the past, there’s one stubborn problem that limits your possibilities: not every stat has been tracked since the inception of the league. This is why so many queries you see start with a disclaimer about which seasons are included — you can’t always use them all. However, there is something we can estimate reasonably well: individuals turnovers because team turnovers were tracked from 1974 to 1977. With some clever adjustments, one can obtain a decent estimate, especially for players who logged a lot of time on the court.

The first instinct for someone probing these stats is probably related to some sort of broad regression model. For instance, basketball-reference estimates turnovers for their Win Shares metric, and they use what appears to be a set of coefficients from linear regression. It works decently and all, but for the best output why ignore additional information? A large portion of the players from 1974 to 1977 played in the seasons afterwards — we have their actual turnover rates from then. Since turnover rates are expressive of player characteristics, they hold up pretty well over time — even when a player’s role changes.

For a test of the power of career turnover rates, I selected a four-year stretch of 1982 to 1985 and compared real turnover totals from expected rates. The career turnover rates were all from seasons beyond 1985, yet that had a higher correlation than the expected turnovers from basketball-reference’s regression[3. An r of 0.8686 compared to 0.8505.].

Yet one can’t just use career rates because that’s ignoring information too. Thus, I used a blend of that and my own regression model, which used data from 1978 to 1989. Weighing by minutes, that version is much more successful[4. The r increases to 0.8894.]. One advantage of this is that regression has a tendency to overfit, so it may cause one to be overly confident about one’s ability to predict turnovers, and career rates don’t have that same issue with a high in-sample correlation[5. I believe basketball-reference used seasons 1978 to 1981 to build their regression, as they outperformed my career rates only method during that segment but not during any other.]. That’s a huge advantage for the period of 1974 to 1977. Plus, it’s more reflective of reality for high minute players with long careers. Certain players might be getting over- or underrated by metrics like Win Shares or BPM because a one-size-fits-all model is being applied.

Finally, there’s one last step: adjust all the estimated turnovers so the players match the team totals[6. The team totals have to be reduced by a factor of 0.985 because some turnovers are team turnovers that are not assigned to any individual. The adjustment is the same basketball-reference uses. It’s simple, but it works.]. The results are in the interactive table below and this link. The highest season total is 434.8 from George McGinnis, who later led the NBA officially in turnovers in 1979 and actually did so three times in the ABA. Rich Kelley and Tom Boerwinkle have some of the highest TOV% numbers — there were many high assist centers in the 70’s but they usually had sky-high turnovers too and could be overrated without proper accounting.

PPR: 100*(2/3*AST – TOV)/MIN[7. This does not use the team pace adjustment normally seen, but that effect is minor because it’s only adjusting against the league average.]

TOV%: TOV/(FGA +.44*FTA +TOV)

TOV100: turnovers per 100 possessions. Possessions estimated from basketball-reference’s team pace stats.

The runaway leader in both assist-to-turnover ratio and pure point rating is Don Buse in 1977. A ratio that high is rare and that pure point rating is elite too. But it’s probably accurate because that was one of his two high assist seasons, as in 1976 in the ABA he had 689 assists to 159 turnovers. That’s very close to the totals in 1977: 685 assists to 152 estimated turnovers. Don Buse is actually a really interesting player statistically overall, leading his league in steals twice and being an early adopter of the three-point shot. He’s fourth in three’s made from 1980 to 1985 and nearly a quarter of his field goals were from behind the arc during that period. There’s another good validation of the results at the extreme end: Fatty Taylor had one of the highest TOV%’s at 27, but Denver had the most turnovers as a team and Taylor had a career TOV% of 23.7 in the ABA.

With the complete results provided, hopefully that’ll be a bit of useful data for the NBA research community. Turnovers are incredibly important events, and it separates a guy like Chris Paul from the field and pushes Kendrick Perkins further away from the field. It also allows for more thorough advanced stats, like all-in-one metrics, going back to 1974. Unfortunately, before that season the NBA didn’t track steals and blocks and didn’t split rebounds by offense and defense. And no mathematical trickery can make those appear — that borders on magic.