Like most of you, I have been thinking a lot about shot selection in the NBA recently.
More and more three point shots are taken in the NBA each year as teams realize the value of that extra point and players entering the league get better and better at hitting those shots. Guards who can hit these shots are currently being called 3 and D guys, even the ones that don't particularly play D. Bigs or Wings that can hit threes are called Stretches.
However, if a player can get really really really close to the basket the percentage of shots that go in goes way up, more than compensating for the extra point on the long ball. So, other guys are called Slashers, because of their ability to 'slash' to the basket. Then there a big guys that never leave the basket area and couldn't find the 'elbow' with a map and a compass.
There aren't any really good definitions of these categories, so doing analysis of their effectiveness is difficult. Luckily, thanks to play-by-play data we have counts of where every player shoots from on the court and how good they are at making shots from each location. I grabbed the shot locations of every player in the NBA last year off of HoopData.
Then I did a couple of Cluster analyses on Bigs, Wings, and Guards based on their shot locations (without regard to actually hitting the shots). Cluster analysis basically helps you group items or people together in categories based on their statistical similarity.
The analyses gave me back some fairly pleasing results in supporting the conventional wisdom and my prior beliefs, which is always nice. Of the locations listed by HoopData, three point shots and shots at the rim were by far the most distinctive and important in grouping players in all positions. Mid-range shots are spread pretty randomly, except for Bigs where the ability to hit anything outside of arms length gets a separate category.
For Guards the shot location clusters look like this, all of the percentages of shot attempts:
| COUNT of Players | AVERAGE of Rim Shots | AVERAGE 3-9 Shots | AVERAGE 10-15 Shots | AVERAGE 16-23 Shots | AVERAGE Three Shots | ||
| 304 | 36.44% | 13.41% | 7.52% | 23.18% | 18.57% | ||
| Slashing Guard | 78 | 33.18% | 10.74% | 8.46% | 24.98% | 23.00% | |
| Three Point Specialist | 90 | 18.61% | 6.11% | 5.03% | 20.19% | 49.00% | |
| Grand Total | 459 | 32.50% | 11.58% | 7.20% | 22.90% | 25.11% |
The Three point specialists took nearly half their shots from three point land, while the slashers got to the rim for 33% of their shots.
Neat right? But wait there's more!
Next I joined the Clusters to my player projection data to see how Slashing Guards and Three Point Specialists compare as groups in actual effectiveness. I filtered out the players playing less than 150 minutes last year or projected for this year. The preliminary results were pretty interesting.
| Values | |||||||
| Row Labels | Count | Ave TS% | Avg eFG% | Avg ORB% | Avg DRB% | Avg AST% | Avg TOV% |
| Slashing Guard | 64 | 51.3% | 46.8% | 2.69 | 9.94 | 23.25 | 15.76 |
| Three Point Specialist | 74 | 53.9% | 50.6% | 2.05 | 9.54 | 17.97 | 13.49 |
| Grand Total | 134 | 52.7% | 48.8% | 2.35 | 9.72 | 20.42 | 14.54 |
The three point shooters score more efficiently, with a better effective field goal percentage and a better slightly true shooting percentage. But they're lacking in other areas like offensive rebounding because they're too far from the basket most of the time.
| Values | ||||||
| Row Labels | Count | Ave TS% | Avg Proj xRAPM | Avg Wghtd 2 Yr ASPM | Avg WS/48 | Avg WP 3 yrs |
| Slashing Guard | 64 | 51.3% | -0.51 | -0.41 | 0.081 | 0.080 |
| Three Point Specialist | 74 | 53.9% | -1.09 | -0.67 | 0.089 | 0.086 |
| Grand Total | 134 | 52.7% | -0.82 | -0.55 | 0.085 | 0.083 |
As before, I am using Arturo Galleti's projection of Wins Produced, Nathan Walker's xRAPM projections, and a two year weighted average of ASPM and Win Shares with an aging factor. The advanced metrics actually liked the Slashers and Three Point Specialists about the same, with the Plus/Minus based metrics a little in favor of the Slashers.
Since I lumped the Points Guards and Shooting Guards together, it's fair to ask whether the cluster just renamed the traditonal position. The answer is not really, though there is a bias for Point Guards to be categorized as Slashers with Shooting Guards tilting a bit less heavily to Three Point specialists. See below:
| SG | PG | |
| Slashing Guard | 33% | 67% |
| Three Point Specialist | 57% | 43% |
Some of the same patterns came out between Slashers and Three Point Shooters even when only comparing Point Guards or Shooting guards. The Three Pointer shooters had higher eFG's and slightly higher true shooting percentages while the Slashers rebounded better, particularly on offense.
| New Team | (All) | ||||||
| Pos | PG | ||||||
| AvgOfMP Projection | (Multiple Items) | ||||||
| Values | |||||||
| Row Labels | Count | Ave TS% | Avg eFG% | Avg ORB% | Avg DRB% | Avg AST% | Avg TOV% |
| Slashing Guard | 43 | 51.1% | 46.5% | 2.39 | 9.50 | 27.48 | 17.33 |
| Three Point Specialist | 32 | 53.1% | 49.6% | 1.91 | 8.53 | 24.12 | 16.05 |
| Grand Total | 75 | 52.0% | 47.8% | 2.19 | 9.09 | 26.05 | 16.79 |
|
|
|
New Team |
(All) | ||||||
| Pos | SG | ||||||
| AvgOfMP Projection | (Multiple Items) | ||||||
| Values | |||||||
| Row Labels | Count | Ave TS% | Avg eFG% | Avg ORB% | Avg DRB% | Avg AST% | Avg TOV% |
| Slashing Guard | 21 | 51.6% | 47.3% | 3.31 | 10.84 | 14.59 | 12.54 |
| Three Point Specialist | 42 | 54.5% | 51.3% | 2.15 | 10.30 | 13.28 | 11.54 |
| Grand Total | 63 | 53.5% | 49.9% | 2.54 | 10.48 | 13.72 | 11.87 |
I will follow this up next time with the wings.
Comments