Counting the Baskets

Checking in on the Euro Stashes

(All Hail Nikola Jokic)

The college season has just started. However, the European leagues have been going since mid-October, meaning I have enough games to start re-cranking up the P-AWS Draft model, at least for those players. 

For this post I am going to focus on a couple of the guys that have actually already been drafted, but were among those players 'stashed' every year in Europe. The group increasingly includes US college players who are encouraged by the drafting club not to sign their offer sheet and instead go overseas. 

I am stretching the draft model a bit by applying it to players a year after they're drafted. The main concern is the aging curve since that tells us not only about the player's expected development, but also the level of interest NBA teams have shown in them, as more sought after players will come out earlier. I believe we'll still get valid results as all of the players are still in the age range on which the model was validated and the fact that they have not been signed by a team mitigates the signal of interest being drafted a year ago for all except Dario Saric, who expressly chose to stay an extra year in Europe.

There also likely isn't enough data available to build a reliable 'Euro Stash' only model based entirely on post-draft data. There are also a couple of features of the model that help us evaluate players in terms of progress including the age curve, but also adjustments for the level of competition, a replacement reversion that rewards more minutes of good production on the court, and weights for the more developmentally indicative statistical categories.

Each player is playing outside the NBA but is one year older than last year, which the model will calculate as a negative. So in order the score as well in the model as they did prior to the draft they need to show a couple of things; most notably improved efficiency and production in line with the development curve, increased playing time showing more trust and responsibility with the team, or moving up a level of competition in terms of opponents.

 On to the players summed up in the table below:

A couple of highlights from this:

• Nilola Jokic is putting up MVP level numbers in the Adriatic League.  He will definitely be ready to move up a level in competition next year within Europe if not to the NBA.  Jokic's projection last year was significantly based on potential, which he's living up to thus far. If Jokic continues this level of play for the full year his score in the model will improve even beyond his lottery pick score last year.
• Dario Saric is holding his own at the higher Euroleague level as well as in the Turkish Basketball league (TBL).  His production is also at a level that could put him higher than last year's score.
• Walter Tavares has basically held steady in every facet, competition, playing time and production, except for age. That's modestly disappointing, but given his size and relative inexperience he may still pan out.
• Vasilije Micic hasn't produced a lot yet in Euroleague play, but the model still credits him for going up a level. However, if he doesn't produce a more efficient rate, in his case produce some steals, reduce turnovers and hit some threes, his lackluster production will drag his projection down further.
• Bogdan Bogdanovic has played Jeckyl and Hyde in Euroleague and the TBL for overall mediocre numbers, though again that is a higher level of competition than last year. Phoenix took Bogdanovic in the first round likely with the understanding that he wouldn't come over, he remains a borderline prospect by my model.
• Alessandro Gentile has had a tough season so far, as has his team EA7 Milan. His numbers in Euroleague have basically been sub-replacement level. I can't imagine him coming over next year unless he and Milan turn things around to make a deep run.
• Xavir Thames, like Micic, at this point is getting credit for moving up in competition, in case from the NCAA to the Spanish and Euroleague.  But he needs to up his production at the level to prove he can move even further up to the NBA.
• Nemanja Dangubic got drafted based mostly on his performance at Euro Camp being named the MVP.  He hasn't produced terrific numbers playing before or since, basically calling the external validity of those gatherings into question. One thing every scout or executive should remember, there is nothing intrinsically more valuable in the games/plays you happen to be looking at than the ones you don't.  If you happen to see a prospect on his best/worst day you don't throw away the rest of the sample.
• Louis Labeyrie, wasn't on the Draft Express top one hundred, which I use as a filter for the model, so I didn't project him.  His numbers look solid but not extraordinary.

It looks like the two front runners from last year, Saric and Jokic, are still the leaders to find their way to the NBA and to make an impact. 

All stats per Draft Express, except TBL stats via RealGM.

A Study in Futility

This post originally appear on Nylon Calculus on 11/5/2014

On Saturday night the Boston Celtics very narrowly avoided setting a record in three point futility when Jeff Green hit the team's first three pointer after 21 misses, avoiding the 0-22 record set by the Denver Nuggets. They ended up 1-25  from three on Saturday, a game sandwiched between 36.4% and 35.5% three-point shooting performances.

Almost immediately, after the the Saturday game, the three-point skeptics came out of the woodwork. Given the immutable arithmetic of 3 > 2, the objections were more subtle and plausible than "threes don't win ringz." OK, I saw one commenter on a blog write that, not sure if it was Byron Scott or not, but to focus on that would be just be nit-picking. There were a few, potentially, valid criticisms of the Celtics' beyond-the-arc shot selection on that futile night that we can check, in part, thanks to SportVU:

• Too many threes
• Too early in the shot clock or too late in the shot clock (three point attempts are like Goldilocks, apparently)
• Too many contested threes.

So was that all true, was the shot selection process off or was is mostly a bad luck night? We can't tell from the result and I would argue it is hard to tell simply from watching as it's often very hard mentally to separate process from results. If something goes badly our natural instinct is to look for and find reasons why. If you are confident that this doesn't apply to you, then it almost certainly does.

Thanks to SportVU data at the NBA.com site, we can now perform basic checks on a good portion of those claims. Too Many The 'too many' claim is somewhat subjective. But it is easy enough to look up the NBA average and compare to the Celtics' infamous barrage.  The Celtics took 25 attempts from three in Houston out of 98 total shot attempts, for a 25.5% three-point attempt rate. League average three-point attempt rate last year was 26.1%, with 23.6% for the eventual champion Spurs and 26.2% for Eastern Conference champions the Heat[1. Per Basketball Reference].

In fact, the 1-25 night was a pretty typical three-point attempt rate so far this season for the Celtics with the other two games at 25% and 30.4%. There are certainly legitimate questions about the number of threes taken, and more particularly whether the right personnel is taking them (such as Marcus Smart, who has struggled in Summer League, preseason and three regular season games from distance). But on the face of it, it is difficult to say the Celtics took too many three attempts.

Too Early or Too Late

Personally I don't have an issue with an open catch-and-shoot three in transition by a good shooter, a shot that has been trending up in the league the last few seasons. Transition threes are good shots, though immediate threes, most likely after offensive rebounds are less efficient, interestingly.

But, whether it's advisable or not, did the Celtics take a significant number of early threes while going 1-25?

Sort of.

According to SportVU the Celtics took 20 percent of their threes in the first six seconds of the shot clock, which would include shots after offensive rebounds, and 22 percent of their twos in that part of the shot clock.  Last year for the league as a whole ~16 percent of three point attempts were taken in the first six seconds of the shot clock. But the Celtics also took more two pointers early in the clock than league average (~20%) last year in the Houston debacle, but given the number of offensive rebounding opportunities it isn't surprising that they had a few extra shots with the clock reset.

On the "Too Late" charge, in part it depends on where we decide "Late" is, NBA.com provides two late time bins, 7 to 4 seconds left and 4 to 0.  If we use the more expansive version of Late it doesn't appear that the Celtics shot more threes than typical for NBA teams, but the Celtics took 16% of their threes in the less effective final four seconds. Below is the shot distributions for the Celtics infamous performance and the entire league last year:

The Celtics shot clock usage follows the league average, with only the Very Late shots off the league average.  Shots in the last four seconds are less efficient, whether from two or three, 38 percent eFG from two and 44 percent eFG from three in the last four seconds of shot clock[2. Per NBA.com/Stats].

Too Contested

According NBA.com's Player Tracking data, 18 of the Celtics 25 threes were Open or Wide Open, for 72 percent. However, in their two more normal shooting nights against Brooklyn and Dallas, 77 percent and 87 percent of there threes were wide open. Meaning they were getting significantly more open looks, though not enough to account for the 4 percent shooting performance.

All told there evidence suggests, I think, that the Celtics near historically bad shooting night was some combination of sub-optimal process and simple bad luck, as are so many outlier events.