Does he have it all? A detailed look at Monta Ellis' decision making

Wednesday night, Monta Ellis was having the kind of game that makes his detractors groan. Halfway through the fourth quarter, he'd shot more than 20 times and made, well, not much. But then he did this:

It was thrilling, and the fact it came the night after this only made it that much sweeter:

Before Ellis came to Dallas, he’d developed a reputation as a chucker whose game was based on high volume, poor shot selection, and low efficiency. But as his play seemed to improve over his first year as a Maverick, a new narrative began to develop, one in which Bad Monta morphed into Good Monta, a newly efficient player redeemed by the brilliance of Rick Carlisle’s offense and the gravitational pull of Dirk Nowitzki.

But as we saw this past week, the good and the bad can come mixed together.  To really make sense of Monta Ellis this year (if that’s even possible) I wanted to dive deep. Perhaps deeper than any person should. I used NBA.com’s detailed statistical splits and a handful of Excel tables to fully profile his promising but problematic game and its evolution over time. Dive with me, if you dare.

Good Monta vs. Bad Monta

Let's start with an analysis Kirk Goldsberry offered on Grantland last year that constitutes the previous state of the art in Monta Ellis Studies. Goldsberry offered this pretty (and pretty persuasive) shot chart:

He sure looks like a more efficient player! There's not a spot of red on the shot chart from his last season with the Bucks, but the chart from his first year in Dallas is littered with red and orange He's making more of his shots and seems to be taking a higher concentration of shots where he's best at them. Where Bad Monta was chucking shots from behind the arc at will in Milwaukee, Good Monta is a focused shooter, finding his spot and drilling shots at an above-average rate.

These charts are great for drawing broad contrasts, but things can get a little tricky when it comes to the details, and in this case the details matter a lot. And comparing just two years of a veteran player's career really doesn't give you enough information to understand the trends. If we look closer, we'll see there's less evidence than one might hope of a real transformation in Monta's game. To make the case that Monta underwent any sort of shift in playing style or efficiency requires looking at more than just the single Bucks season we see above. His last season in Milwaukee was one of his worst overall and his absolute worst in terms of shooting efficiency. It's fair to criticize him for his poor performance, but that season alone isn't a good approximation of his pre-Dallas shooting.

Monta's efficiency over time

So how efficient was Ellis before he came to Dallas? About as efficient as he's been in Dallas:

The memories of fans and analysts alike can be short, so it was tempting to turn that upward sloping line between his 2013 season in Milwaukee and his 2014 season in Dallas into a story about his entire career, but neither this year nor last looks terribly different from his earlier seasons. It's true that he hasn't been a consistently efficient shooter, but his only full season as a Maverick isn't really an outlier when you look at his career as a whole.

That's not to say that Ellis hasn't had years where he's played substantially more efficiently than others. Whether you focus on shooting or look at advanced stats to evaluate his overall performance, the 2007-08 season was Monta's best. It's the only season in which he was an impressively efficient shooter with a 53.5 effective field goal percentage, so if you want to craft a narrative about what separates Good Monta from Bad Monta, that's a good year to look at.

Shot location

So what was it that Monta did differently that year? Is there some structural explanation for the changes in his efficiency?

You can start to see a difference between his more and less efficient years when you look at where his points come from. His two most efficient years (the seasons ending in 2007 and 2008) correspond to the years in which he scored the largest share of his points by getting into the paint. You can see it visualized in the chart above or numerically in the data below, but his best years follow this pretty intuitive pattern: when Monta is able to shift his scoring toward more efficient shots, he's a more efficient player.

He may be above league average at making those long twos (as we saw in his 2014 shot chart above), but that doesn't mean that those are the most useful shots for him to be taking. They're difficult shots, and unless you take the time to step back behind the three-point line, you don't get points for difficulty in the NBA. Monta's best years all see him looking to take higher-value shots. Last year (his fourth-most efficient) followed a similar pattern, but was hardly transformational.

So it looks like Ellis has frequently had a problem with inefficient shots, and it looks like that didn't really change much when he got to Dallas. But sometimes a long two is the best shot your team can get before the shot clock expires. When that's the case, a player who's good at making difficult shots can be a real asset. To truly understand, we need to dive even deeper.

Shot timing

Fortunately, NBA.com's stat page now lets us see when in the shot clock players tend to shoot. Here's Monta:

Ellis takes a decent number of his shots "very early" in the shot clock and has a ridiculously high field goal percentage on those. Most likely these are high-value lay-ups and open threes in transition. Like most players, Monta takes a plurality of his shots nine to 17 seconds into the shot clock, in the flow of the offense. But the row highlighted in blue deserves more scrutiny.

Outside of that "average" shot clock range, Ellis is most likely to take shots "early" in the shot clock (between six and nine seconds into a possession), and at first pass, he seems to be choosing them pretty efficiently. But if you break those shots down into two- and three-pointers, it looks more problematic. A small fraction of those shots (3.7 percent) are three-pointers and he's made an impressive 50 percent of those, but most of them (14.7 percent) are two-point shots that Ellis makes at a rate of just 43.6 percent.

There's not a breakdown of distance from the basket for these shots, but given the frequency with which Ellis has been getting these shots (2.6 per game) and the low rate at which he's converting them, it seems likely that these are mid-range jumpers taken before the shot clock has run down. But whatever they are, it's clear that rather than taking a few extra seconds to try to generate a more efficient shot, Ellis is taking almost 15 percent of his shots before he really needs to and without being in an optimal position.

What is to be done?

So should we worry about Ellis going forward? Despite the craziness of some of the more recent games, his numbers look ok, as you can see in the numbers above and on the chart below:

But when you look at both his timing and his distribution so far this year, you see that some worrisome old habits seem to be popping up. He's relying a lot on mid-range shots taken fairly early in the shot clock, and his usage numbers and the relative efficiency of his teammates means that could be a problem going forward. Despite being the fifth most efficient shooter on the team (and that's not counting players like Jae Crowder who are more efficient but unlikely to be on the court), Ellis has the highest usage rate and plays the most minutes, so if he's making bad decisions, that's going to impact the team significantly, especially since the schedule is going to get a lot tougher going forward.

The good news is that Monta really is a very skilled basketball player who can score reasonably effectively from different places on the floor. He doesn't have to take early-in-the-shot-clock long twos. If he gets the ball with time left on the clock, he should drive if he can and pass if he can't. That simple change to his game could make him a dramatically more effective contributor. The bad news is that so far, there's little evidence Carlisle has really turned him into a better decision-maker. For the Mavs to contend in April, that's going to have to change.