NBA Draft season is coming, and I've started to fill out my NBA Draft guide. For the guide, I thought it would be cool to add a fun graphic next to all the players to help show their statistical profile. Inspired by the charts from FIFA that display different attributes of a player, these charts show NBA drafts prospects performance in 7 different areas: scoring, shot creation, shooting, finishing, playmaking, rebounding, and defense. The charts display players percentile rank in each category compared to other NBA draft prospects. The way I calculated the numbers for each category are as follows:
Scoring: 67% PPP, 33% TS%
I mostly used points per possession (PPP) for scoring. I also included True Shooting to add in an element of efficiency, but I did not add a shot difficulty element for reasons I will get into later.
Shot Creation: 40% unassisted 3's, 30% unassisted at the rim, 15% 3pt. %, 15% FG% at the rim
Unassisted 3's receives the most weight here because I am most impressed by that level of shot creation, and I do think it is to some extent the most valuable and/or rare. Creating shots around the rim also matters a lot. I added elements of how well you are making these shots, just to make sure we don't overvalue poor shot selection.
Shooting: 35% 3pt.%, 35% FT%, 10% 3PAr, 10% 3PA/100, 7.5% FTAr, 2.5% FTA
This equally weighs 3-point and free throw percentage; however free throw percentage is generally a better indicator for NBA shooting because of sample size and variability. I also added elements to account for the sample size we are looking at to make sure we don't reward low volume shooters.
Finishing: 60% FG% at the rim, 20% Rim AR, 20% unassisted at the rim
This is mostly shooting percentage at the rim. I also wanted to include how much a player shoots around the rim and how many are unassisted to try to add an element of shot difficulty.
Playmaking: 50% Ast%, 35% Ast/100, 15% TO%
This is mostly assist rate, as I think that is the best easily available measure of playmaking skill. At the college level especially, there could be some weirdness there, so I added actual assists per possession. I also added a smaller element of turnover rate to account for if a player is making too many mistakes.
Rebounding: 70% Reb%, 30% Reb/100
This is mostly rebound rate, as that is the best easily available rebounding measure. I also included rebounds per possession as again, there could be some weirdness in the college sample.
Defense: 75% DBPM, 12.5% STL%, 12.5% BLK%
Defense is the biggest crapshoot by far to try to measure. I mostly used DBPM because that is one of the only defensive single number metrics publicly available for college basketball, but I also included elements of steal and block rate.
This is a valuable project because it can show you how a player performed at the college level compared to his peers. It visualizes what they were good at and what they weren't good at in an easily understandable format. Just looking at these charts, I've had many moments of "oh, I thought he was better in that area. Maybe I have to go back and watch more of this aspect of his game." That's what this is for. This is not saying what a prospect is or isn't good at, but rather guiding an evaluation to think about if someone actually was as good at something as we thought. Also, this could be used to confirm what we thought already based on the eye test. Another useful way to use this tool is to compare prospects. You can overlay different charts to compare how prospects played in different areas. This is very useful for looking at players in similar roles that are in similar draft ranges.
There are limitations of this. First of all, this is not a tool to make evaluations, but rather to help guide them. The biggest limitation here is that it is showing what a player did in college, not what he is going to do at the next level. The purpose of the draft is deciding what a player is going to do at the next level, so obviously that is a big limitation. Another limitation is that it does not account for role, age, or strength of competition. Lastly, many of the statistics I used to make these are unavailable for G-League Ignite and International prospects.
I've talked long enough about these charts, so let’s look at some of them:
Single Charts: The Consensus Top Guys
Cade has an interesting chart. We all know the playmaking chops are there but doesn't necessarily show in the chart. A lot of that is because of context. The spacing at Oklahoma State was awful, which didn't really allow Cade to pick up those assists by making cross court reads to shooters. There were a lot of record scratch moments at OK State. That can also be somewhat reflected in the finishing numbers. It’s tough to finish around the rim if there are three defenders collapsing on you. I actually think finishing could be a little bit of a concern here, but not to the extent that his chart shows. Positively, the shooting, scoring, and shot creation is incredible.
Again, context matters. The scoring numbers weren't as high for Suggs because of the excellent talent. That excellent talent, however, makes him look better in every single other area. I'm high on Suggs and this chart is mostly positive, but just take into account the context of Gonzaga.
Mobley is interesting. This shows the limitation of not taking into account role, because for a center Mobley is awesome at shot creation and playmaking. The scoring, finishing, and rebounding is what matters most here for Mobley's positive projection. The other stuff may not show up in these numbers, but I would recommend thinking about the context of how these are made and also think about the film of him at USC.
3 Player Comparison Charts: Similar Players
I talked about all these guys above, but this is what the comparison looks like. What caught my eye first is Cade's advantage in shot creation and shooting. Anyway, all of these guys are excellent prospects.
I think these three players are interesting for comparison. This is another one where the context of what Mann and Bouknight were asked to do vs. what moody was asked to do are different things. Moody was not relied on as heavily to create on the ball, therefore his numbers there are lower. However, the Mann numbers for creation and shooting are very impressive. The Bouknight numbers for scoring and shot creation are great. Moody may not look as good compared to those guys here, but remember he is younger than both by over a year.
Three interesting big wing prospects who all had a very different experience in college basketball. This is an interesting comparison because all three are really nice all-around players who fit the mold of what the current NBA is looking for. Johnson looks better in terms of scoring (albeit a much more limited sample), Barnes looks better at finishing and playmaking, and Franz grades out really well all around. I'm going to be very interested to see where all of these players get their names called on draft night.
I like this one because these are three Freshman that struggled in their first season. I don't want to try to make any sense of the numbers for these guys, but I felt the need to throw this one up here.
4 Player Comparison Charts: Similar Players
Four wing-ish type players who are all on somewhat different timelines. First of all, look how well Terrence Shannon Jr. looks. We can see a limitation here with the charts in Kessler Edwards defensive percentile, as I actually think of him as a really good wing defender. Anyways, I also threw up the two Tennessee guys to see how they compared to slightly older wing prospects who I really like.
Look how good the older guys look! This is why it can be important to compare players of similar roles and ages, otherwise you might think Chris Duarte should be a top 5 pick. However, look at how well Davion Mitchell grades out in terms of shot creation, even compared to his teammate Jared Butler...
I wanted to throw in four deep sleepers. Yeah, these players play different roles, so I did break my own rule here. However, I do think these off the wall prospects are being a little undervalued right now. I haven't seen any of them on a top 100 board, which is a little head scratching to me.
The code for the project can be found here:
Comments