(re) Introducing Scorer Profiles:
Allow me to (re)Introduce Scorer Profiles. I've introduced them once before through my twitter account (@rgiss11), and feel the need to dive a little deeper into them here.
Introducing scorer profiles:
— Riley Gisseman (@rgiss11) March 23, 2019
There are FGM that are assisted and unassisted, but when comparing a player like Klay Thompson to Clint Capela, a distinction needs to be made.
Here is our initial scorer profile: the Utah Jazz pic.twitter.com/fvywQaGnwS
What exactly is a scorer profile?
A scorer profile is a contextual visualization of how a player scores their points relative to others. A distinction needs to be made on who creates points for themselves and who is created for by others. After all, Rudy Gobert has a higher percentage of unassisted made baskets than Klay Thompson, but Gobert is far from the creator that Thompson is. Gobert's creation is only when another player creates for themselves and he's put into position to rebound and score, which is far from the value that we see from players who can create as a Pick and Roll Ball Handler, in a Post Up, or in Isolation. So we use putbacks to quantify the difference between assisted and unassisted baskets.
How is a scorer profile calculated?
A scorer profile is calculating by adding the vectors that come from a radar chart. Each score for the radar chart is one's percentile rank among all players in 1. Percent of baskets assisted, 2. Percent of baskets as putbacks, and 3. Percent of baskets unassisted, minus percent of baskets as putbacks.
A stronger pull towards putbacks will pull away equally from both assisted and unassisted baskets, and a strong pull from both putbacks and assisted baskets will be a very strong pull away from unassisted baskets. This creates very distinct roles, as one who only fairs well in assisted baskets is likely just a catch and shoot player, one who has a lot of unassisted baskets and a lot of putbacks will be a skill big, like Giannis Antetokounmpo or Blake Griffin. All of these roles are shown below:
How are Scorer Profiles Used Practically?
There are a variety of practical uses for scorer profiles. We can use it to compare players directly to one another, we can use it to compare players to themselves, showing progression over one's career as a scorer.
We can use scorer profiles to visualize the NBA's best and worst lineups, to see which kinds of players work best together, and which should be left apart.
Further, by identifying scoring roles, we can compare all players in one role to see how one performs against the typical player in that given role.
So we can then take the average player in each role, and weight a players' performance to that role. After all, Donovan Mitchell is technically "inefficient" when including all roles, even those that don't create for themselves, but he's really just as efficient as the average ISO scorer.
| Scorer Profile Role | TS% |
Points Per 100 Possessions
|
| Skill Bigs | 55.4 | 22.6 |
| Balanced Big/Wing | 56.7 | 20.1 |
| Balanced Guards | 54.1 | 20.7 |
| ISO Scorer | 53.8 | 24.6 |
| Roll Bigs | 59.7 | 18.3 |
| Rebounding Bigs | 59.9 | 18.0 |
| Spot-Up Shooter | 55.9 | 16.6 |
This is what I call a "Scorer Profile Evaluation". Since a scorer profile doesn't take into account how good you are at scoring, just how you score, there has to be a way to bring up how well you score as well. So we're grading you, but we'll be putting you on the scale of others in your role. How good is a player at scoring in their given role? Putting this weight on a player's numbers gives a more true value of their output than if we had taken them at base value.
All of these different ways of viewing the data can be brought together to scout out how two teams match up. Here's the Pacers vs Celtics comparison from before their first-round series.
Concluding:
As of right now, I've got full control of all formulas and numbers for our scorer profiles. I haven't yet found a way to make them easily accessible to the public, and that's something I'm working on. I'm forced into tidying up these graphs (coloring, labeling, titling) in photoshop until I learn how to make them more easily, likely through Python or R. Until then, I'll be making and posting them on here in my various posts, whether in NBA Draft breakdowns, statistical articles, or other posts. You should also occasionally see them on the BBall Index, which I do data visualization for.