NBA Week in Review 13: Time to Panic

Across the league, teams are crumbling under pressure. As most come to the realization that they don’t play for the Spurs or Warriors, desperation will wash over the NBA like a tsunami. Injuries are starting to stratify the league further too, as the Thunder lost a key defender for a while, Andre Roberson, while Blake Griffin’s latest incident will extend his absence a few weeks more. We’re going to see more teams tanking and perhaps a few interesting trade deadline moves — with a healthy dose of delusion, you can convince yourself that you’re one move away from challenging a team that’s on pace to go far past 70 wins. Some franchises have spent years and hundreds of millions of dollars to peak right about now, and they just aren’t good enough to compete with the giants. Is it time to panic? Yes. Yes it is.

Blatt Canned in Cleveland

In one of the more memorable Woj-bombs ever, partially because the Cavaliers had just rebounded with a nice win over the Clippers, Cleveland fired David Blatt as head coach and immediately hired and retained Tyrone Lue with a three-year contract. Everyone has an opinion on this, and a consensus emerged: the Cavs did remarkably well in the playoffs and earlier this season given their injury woes, and hiring a new, inexperienced coach mid-season will likely not result in a short-term boost the franchise desperately desires. Maybe you have an issue with his offense or his rapport with his players, but this isn’t the kind of move that’ll help them defeat the Warriors or Spurs.

I imagine, however, Cleveland’s logic was like this: the team is not equipped to beat the two Goliaths from the west, and with LeBron on the team there’s a ticking clock that’ll soon expire. Thus, a major change needs to be made, and since Blatt is clashing with some players and LeBron doesn’t approve we’ll get someone who can maximize our player talent. I’m not saying I agree with them, but that’s possibly their train of thought. I’m not optimistic on Cleveland’s chances in the Finals after this move, but the silver lining here is that this will give them an opportunity to experiment with lineups and configurations on the court. That’s at least interesting basketball, even if it doesn’t result in a better team.

Lowest Assists Per Game Team Leaders

After seeing a conversation on Twitter about Gordon Hayward leading the Jazz in assists per game with a paltry 3.7, I had to answer a question: Which player led his team with the lowest APG? First, however, I have to define the terms. The rate requirements for APG are, frankly, maddening because of how inconsistent they are. Instead I’ve tried two different filtering methods: only players with at least 50 or 30 games qualify with lockout seasons adjusted[1. The adjustment is simply 50*team games/82.]. The 50 game filter is closer to the actual rate requirements, and the results from seasons 1955 to 2015 are below with a surprising name at the top: Nerlens Noel. Through trades and injuries, Philly had few players who even amassed 50 games, and their defensive big man ended up leading the game with a mere 1.71 assists. The rest of the list is mostly a mix of older teams with newer ones that feature a lot of player movement.

Tuning it down to 30 games, and the name at the top changes to Chuck Mencel of the Minnesota Lakers. It actually Mencel’s second and last season in the NBA. He had to fulfill ROTC requirements, and then stayed in Minnesota for his family when the team moved to Los Angeles — that’s not something you’d see anymore, but it wasn’t a bad career move because he ended up as a CEO and president of a corporation that was later bought out by Caterpillar. If you lower the rate requirements even more, Whitey Skoog shows up with 3.3 assists for the Lakers, as he played 23 games. But Ray Scott, second in the table below, truly was the assists per game leader for the Pistons with no qualifications needed. As for the Jazz, while 3.7 assists is definitely low for a team in the modern era, it’s not unprecedented and it’s far from the unofficial record with numbers similar to the Pistons in 2012.

Sky-High Westbrook

Since SportVU movement data is unavailable on stats.NBA.com right now, I can’t play around with the data for Westbrook’s dunk below or another event. Instead let’s just appreciate that we have a player averaging 24 points/10 assists with 2.5 steals per game with shocking athleticism, and he’s taken a backseat to another point guard this season and has to battle other players for second place in the MVP race. And if I had to estimate, I’d say that without SportVU data his vertical is roughly 5,000 inches here.

Dwight Howard’s Legacy

As the ESPN all-time rankings roll on, it’s interesting to see how some contemporary players are rated. Dwight Howard, for instance, was one of the very best players for a few years with three Defensive Player of the Year trophies and arguably had the better case for MVP over Derrick Rose. He even took an Orlando Magic team to the finals without a second star and pushed them past the infamous 66 win Cavaliers before losing to a stacked Lakers squad — and they were missing their starting point guard Jameer Nelson. Howard is 30th all-time in MVP win shares, which is a decent barometer of historic legacy. Stating that he only ranked well in voting because of weak competition is bogus, by the way — he was fourth in 2009, which coincided with a peak LeBron season, Wade’s best and one of the best ever from a shooting guard, one of Kobe’s best seasons and one year removed from an MVP, a Chris Paul season with absurd stats for a point guard, and a few other solid campaigns from many players like Brandon Roy.

ESPN ranked Howard 67th and in the same range as players like Dennis Rodman and Alex English, who were never as good as Howard was at his best. I doubt the issue is longevity either because Howard has been producing since coming out of high school and ESPN ranked Curry 4th all-time among point guards even though here’s a mere two-time all-star and this is just the beginning of his prime (presumably.) Chris Webber is an illuminating example: he was never higher than fourth in MVP voting and fell off sharply in his early 30’s just like Howard; plus he has a poor clutch/playoff reputation. From what I remember when he played, he wasn’t seen as one of the league’s truly best, but Howard was. Yet both are ranked in the same tier. The explanation here is obvious, and it has to do with Howard’s off-court antics, while Chris Webber is well-spoken and an adored member of the media now. It’s not that fairness is an issue here; accuracy is. Memories are malleable and we penalize players for unrelated issues. Remember this for the future because it’s going to happen again and players will suddenly become better or worse because of external factors, and we’ll warp and change key aspects of the history of the NBA.

Moneyball: Buying Low

Moneyball in the NBA is imagined as a system that identifies underrated players and emphasizes players who, say, play solid defense and shoot a ton of three-pointers over midrange shots. But “Moneyball” at its heart is about spending money effectively with a limited budget. One of the best methods is to simply target players in slumps or in down seasons. As long as they’re not well into their 30’s or suffering from debilitating injuries they will likely get better and could very well out-perform their contracts. This happens to shooters frequently and it’s not usually a cause for concern because even great shooters will go cold for a few months at a time and recover eventually.

For a recent example, the Clippers brought in Jared Dudley using some of their best assets, and they dumped him after a poor season. In fact, they paid something else with a first round pick to take him off their hands, but he was only making $4.25 million and was only 28 years-old. He immediately had a great year as a role player for the surprising Bucks[2. There’s a lot more to this bizarre Dudley trade, like Doc Rivers forcing him to play through an injury, which makes it more inexplicable they gave up on him after one year.]. Channing Frye is another example. I identified him before the season based on his pattern of alternating good and poor seasons, and he’s quietly been effective and valuable for the Magic.

For a future candidate, PJ Tucker is a highly desired 3/D type with enough toughness for playing a stretch 4 too and he’s just a good enough shooter that he has to be accounted for on the perimeter. In fact, he’s been on my radar for a while because he was shooting 26.8% from behind the arc at one point, causing many people to write him off, but he’s already brought his average up to his career norm. He’s been partially forgotten because the Suns have imploded, but it’s through no fault of his own and he should be more valuable around talented players where he can focus on his specific strengths. Moneyball is about aggregating the value of a superstar through multiple smart deals, and that involves players like PJ Tucker, as banal as that sounds.

Is This Basketball?

During a game that served as an example of why we need to ban hack-a-player, Houston committed a war crime by hacking Andre Drummond to a truly absurd degree. For instance, instead of waiting until they were in the penalty, Houston opened the third quarter with five straight fouls from KJ McDaniels, who was promptly taken out after having performed his basketball duties. This sticks out like a sore thumb in the play-by-play logs, as seen below, and it’s taking this tactic to a new extreme. You don’t want to be in penalty because teams are usually about five points per 100 possessions more efficient, and Houston in particular has the “worst” penalty defense, although some of that is affected by the intentional fouls. But let’s continue this silly nonsense because you can’t change a rule that only affects a few players, as that’s exactly how rules work[3. Sarcasm font.].

Jump Balls: Logistic Modeling

Nylon Calculus provides jump ball information going back to the late 90’s, breaking things down by “open” jump balls at the beginning of the game to “live” ones that occur during. But jump ball win percentage doesn’t directly tell you who’s best at winning jump balls, especially during open ones — you have to adjust for the competition. This is especially pertinent for perimeter guys who get mixed into the fray. It’s tough to know if a guy has a high win percentage because he’s facing off against Raymond Felton or he’s actually good enough to compete with the giants without digging into the data.

With data provided by jump ball aficionado Matt Femrite, I have match-ups for over 40,000 jump balls. To determine the relative strengths of every player involved, I’m using a logistic regression model with two variables[4. This is a binomial regression where the response is either a 1 or a 0. The two players are either listed as player 1 or player 2, and if player 1 won the jump ball then the response would be 1. Unfortunately, the data was not organized so that homecourt players were always player 1, so the intercept in the model is meaningless.]. For simplicity and computational ease, every player with fewer than 30 jump balls is lumped into one bin, and for accuracy all those entries use an additional variable: height[5. I dumped every match-up where both players had fewer than 30 total.].

The results are in the sortable table below. Shaquille O’Neal is the reigning king here followed closely by the 7′ 3″ Arvydas Sabonis. Both guys have quite a bit of heft, but they have high standing reaches too. But size isn’t the overriding component here: Jerome Kersey — beloved former Portland small forward — is ranked fourth[6. What’s stranger is that the data only captures the tail-end of his career when he was in his late 30’s. If anyone has a good account of how he won so many jump balls, let me know.]. He’s a giant killer with a great win-loss record in jump balls, having beaten the aforementioned Sabonis, Dale Davis, Elden Campbell, Bo Outlaw, and Jermaine O’Neal. Conversely, Yao Ming has one of the most disappointing ratings. Height usually but does’t always win in this facet of basketball.

Note: Expected win% is the calculated percentage versus an “average” (i.e. coefficient of 0) jump ball opponent. 

At the other end of the scale, don’t let Luke Ridnour take the opening jump ball. Against an average opponent. he’d win just over 8% of the time, which is as close to zero as you can get because an opposing player may just accidentally tip the ball to the wrong guy. The guys with the worst ratings are, predictably, the tiniest players like Aaron Brooks and Damon “Mighty Mouse” Stoudamire. But a few big men sneak in, like Chuck Hayes and David Lee.

The formula is simple, and I hope that because this is Nylon Calculus this part won’t be skipped: exp( PLAYER_1 – PLAYER_2 ) / ( 1 + exp( PLAYER_1 – PLAYER_2 ) ). That’s it. For those who aren’t familiar with this type of formula. it’s the widely used logistic formula that’s used frequently with odds calculations. For an example, here’s the probability of Shaq winning a jump ball against Dwight Howard: exp( 1.319 – 0.740)/(1 + exp(1.319 – 0.740)) = 0.640 or 64.0%[7. The “exp” represents e^ if you’re not familiar with that notation.]. And here’s an example of the probability of two-time MVP Steve Nash beating two-time MVP Tim Duncan: exp( -0.953 – 0.567)/(1 + exp(-0.953 – 0.567)) = 0.179 or 17.9%. While it seems unlikely Nash would ever win, I’d note that even if you strike the ball first you can’t guarantee winning the battle because it’s about how well you control the tip and how well your teammates corral the ball. In fact, Nash has actually beaten Shaq and Karl Malone; he has a pretty good win-loss record considering his size.

Finally, I’d like to emphasize that these results, even though I used ridge regression with cross-validation, are descriptive rather than predictive. In other words, it’s a good summary of what happened and who had the most success, but the results are a bit wonky in some places because jump balls are so wild and it takes a large number of events for a specific player for stability. There are better methods available, like forming a Bayesian prior — you can’t simply penalize all the results to 0 because the players with the fewest totals are often the smallest guys. But with these more descriptive results, the outliers aren’t glossed over and we can perhaps learn new things about how players win jump balls and how much we might be overrating sheer size, for instance.

If that means teams will experiment with players like Rudy Gay and Larry Nance Jr. taking the opening jump ball, then Nylon Calculus has done its job.