Understanding Analytics: Why finishing metrics can be misleading
Half-court metrics tell a different story than those that include transition looks
“If you’re digging a hole in the wrong place, making it deeper doesn’t help anything.” — Seymour Chwast
As frequent listeners of the Game Theory Podcast with Sam Vecenie and Matt Pennie, we take a lot away in regards to process. That is, how to go about making sure our process in scouting players is the right one. During their recent episode’s mailbag portion, a listener posed an important thought about using analytics to verify what the eye test might reveal.
“What analytics do you rely on when you are looking for numbers that back up what you see when scouting?”
A fantastic question that produced an equally solid answer from Matt and Sam. What stood out most about Vecenie’s response was, candidly, how staunchly he utilizes half-court finishing metrics to evaluate:
“Yeah, I think it’s pretty obvious that both of us do tend to rely more on the tape, and scouting than just purely numbers. I certainly look at the numbers, I do use Synergy a lot, I use Synergy for half-court scoring at the basket. I think it’s a really effective one in terms of [evaluating] guards. It removes transition play, I think that it really helps in terms of determining how good of a half-court finisher someone is whenever they have to finish around a bigger dude.”
From a process perspective, we agree wholeheartedly with Sam. A few years ago, we read an interview from then-Philadelphia 76ers head coach Brett Brown talking about how to coach and structure an offense in the NBA. He talked about organization, but breaking the game down into three distinct phases:
In relation to transition, the way I break this down is that 67% of the game is played against a set defense. Either due to stoppage by referee whistle or preparing after a made basket, open-court and free-flowing play for 94 feet is really only responsible for, at most, one-third of the game.
But we know that a third of the game isn’t fast break layups and dunks. No, Brown’s point on the third that is “transition after misses” relates to how to organize a team’s offense and run to spots. According to Synergy Sports Tech, at the NBA level, the average team generates about 15% of their offense in transition.
From an evaluation standpoint, finishing metrics in the half-court are much more important to find than finishes that include transition. Synergy, thankfully, posts those half-court metrics for all individual players at the college level. The numbers are there, accessible, and include different ways to gain them: out of post-ups and all non-post-up attempts.
The college game is, however, fundamentally different. Transition is more frequent, with full-court pressing, higher turnover rates and programs who can force opponents to play at their pace. In the NBA, the league-leading transition group (the Charlotte Hornets) gets 19.2% of their offense under the play type label. At the Division I level, 82 programs (or 23% of the entire country) spend more time in transition.
What this comes down to is a danger in taking non-transition numbers from prospects, especially those who play on teams that press or make a living in transition. Their finishing metrics are inflated due to the uncontested nature of open-floor finishes. Transition, with advantage situations and 1-on-1 breaks, isn’t a realistic apples-to-apples indicator of how a prospect will fare in the half-court at the NBA level.
Yet we still see countless examples of finishing metrics referencing overall numbers and not half-court. It’s a trap that too many fall into, either because they don’t know where to look for the numbers or the overall ones fit nicely into the narrative they’re trying to construct.
Here’s an example. Projected top pick Jabari Smith’s transition-inclusive metrics list his overall finishing as 23-38 (60%). Those numbers are drastically inflated by transition, though: in the half-court, he is 14-25 (56%), and he is 9-13 (69%) in transition.
More than one-third of his rim attempts come in the open floor. Auburn, who generates 21.5% of their offense from transition, is top-30 in the nation with their open-floor rate. As a result of their team defense and high-octane go-go emphasis, Smith’s transition attempts look like this:
Are there takeaways from a scouting perspective you can find, about his long-stride finger roll, highlights of finishing through some contact and misses with his left hand? There are certainly lessons learned, but they don’t do enough to deal with how Smith gets himself to the basket, which is just as important as what he does there.
In trying to project whether a prospect is going to be a good finisher at the NBA level, there are two components to figure out: can they convert on their attempts and how do they get to the rim?
Smith’s issues, as we detailed in our recent eval on him, are related more to the second question, as he’s jumper-heavy in the half-court and could stand to fix some of his driving mechanics to create separation. We still believe Jabari is a top-three prospect in this class, but he has his flaws. Dismissing them by simply stating “Smith is shooting 60% at the rim, so once he gets there more frequently, he’ll be fine” is far too broad of a leap to take.
You can’t turn a mid-range jumper at the end of the shot clock into a transition layup. So we need to stop using stats that are influenced and inflated by transition metrics to rationalize behavior in the half-court. The NBA game is mainly played against a set defense, so the stats we use to project impact at the next level are going to be most meaningful when they have those same parameters.
Be weary of all-encompassing metrics for what they include that they shouldn’t.