“My propositions are elucidatory in this way: he who understands me finally recognizes them as nonsensical, when he has climbed out through them, on them, over them. (He must so to speak throw away the ladder, after he has climbed up on it.)
He must transcend these propositions, and then he will see the world aright.”
Ludwig Wittgenstein, Tractatus Logico-Philosophicus
Or, as I probably oversimplify it, “climb the ladder to forget the ladder.” This passage has always stuck with me. Climbing the ladder refers to the constructs we use to advance our knowledge. In essence, once you have used certain tools to conceptualize an issue, you should focus on what you’ve learned rather than fixate on what got you there.
In draft work we are constantly climbing ladders. The name of the game is summarizing a player’s entire future performance in an ever-changing sport in a single numerical number: their rank. To do this we must conceptualize, using shorthand to assign values to various aspects of a player’s game. “Strong handle” we may label to several players on a big board. We give players these labels through visual cues, box score watching, media consensus, etc., all which color in the lines of our conceptualizations of a player. When you rank someone, you imagine their game in your head, but have to shorthand it through a skill grouping (“he’s the most versatile, therefore he’s a lottery pick”), player type (“he’s the best advantage creator, therefore he’s first overall”) or player comparisons (“he’s a Jaren Jackson Jr. type, therefore he is a top five prospect”).
We all do this, it’s the only way to get through it as there is no single perfect measure to represent this complex answer. What I am proposing, however, is to be more explicit about it. This is why I made a major change to my philosophical framework this draft cycle. I took all that I had learned from prior cycles where I had used a top-down model to assign discrete values. I climbed the ladder by internalizing the lessons from this framework, but left it behind to construct an original, new model.
My top-down model in the past worked like this: for every prospect, I would assign probabilities of them being anywhere from a -8 to +8 impact per 100 possessions player at their peak. I then assigned values to each outcome, where +8 was more valuable relative to +6 than +6 was relative to +4, and so forth down the line. There is a limit to how negative a player will be, so everything -4 and worse was scored as a zero.
The goal was to better capture the extreme outcomes, but ultimately fell prey to the same biases that are involved with a pure rank. This system led me to being far too low on Brandin Podziemski, for example. While I liked everything about Podz’ game, I did not assign him any star odds given his mid major competition and seemingly mediocre lateral quickness. I let these two concerns convince me that even in an optimistic scenario he would have limitations keeping him from star upside. Now I’m not so sure.
My 2023-24 model, rather, is bottom-up. It looks at underlying components that make up a player’s relation to the game of basketball. Instead of handing out odds, I grade three factors on a scarcity scale. While I’m still coming up with numbers in a way that’s likely subject to biases, the hope is that this factor-driven approach can drastically reduce them.
The first factor is production. This is essentially how many good things the player makes happen on the court. There are many attempts (such as Box Plus-Minus) to measure this which offer helpful aids in analysis. But there are also non box-score events like screens, deflections, box outs which are technically still production even though not counted. The way I liked to think about production is “how many things does this player make happen almost by accident” to capture the moving block of skill that is a productive player.
The second factor is feel. Now, we already see issues limiting the utility of my model. How does one produce without feel? How can you gauge it separately? Well, I can try. There are some measures that give clues, like assist to turnover ratio and stocks to foul ratio, but that is far from the full picture. Like productivity, but even more so, we must rely on the tape.
Third is dynamic athleticism, i.e. how much dominance a player can assert through physical means. Once again, overlap with the other two, but other clues available like number of dunks, offensive rebounds, drives, free throw rate. But I again find tape-watching essential: how does a player move and will it hold up at the next level?
I took public notes on my process, writing the three pieces linked above to show examples that were helping me determine the definitions. A quick reaction time to swipe a ball away: that is productive but also high feel, and if employing physicality then a plus for athleticism as well.
One major source of comfort in this methodology is that, even if there are overlaps in my grading, it would likely be in fertile territory for growth. The goal is not to measure current performance but that of a player over the course of their NBA career. The traits that fit into all three categories are likely solid foundations to grow upon. These are the undisputed tools that feed into development as much as current production.
One example of how the tri-factor process plays out, obvious to anyone following my content this season, is Jonathan Mogbo. He grades very well in all three factors: he was a highly productive NCAA player (though with competition level questions leading to a very good rather than elite grade); he was a highly effective passer, nailing teammates on structured and improvised reads alike (though his occasionally poor defensive reaction time keeps him from elite territory); his athleticism is unquestionable, third in the NCAA in dunks at 6’6.25’’ and a 7’2’’ wingspan. Why wouldn’t he be a successful NBA player?
However, the issue still persists: this is not a clear measurement of basketball value, more like a fuzzy approximation of ability and developmental slope. The overlap between factors will lead to misses, as my biases inevitably will assign points for gray area traits in multiple categories for some players. A full cycle of providing these grades certainly helped make the lines between factors are clear as possible, but there are still limitations. While the model is bottom-up in dissecting a player’s characteristics, it does not map cleanly to on-court happenings.
That’s why we’re mixing it up again, babyyyy.
I am more than happy with my board outcome, with Zach Edey at the top, and other sleepers in Oso Ighodaro, Terrence Shannon Jr., Dylan Disu and Tristen Newton. But can we do better?
This time, we’re staying completely on the floor in the most literal way imaginable. Once again, we have three factors, but we’re splitting apart by dimensions of basketball impact. Expect a new series of three detailing this new process but the essence is this: how does a player move their skillful self around with pace while applying force? A mouthful to say “how good are they at basketball,” but a better definition of what we’re trying to measure.
In this way we can separate impact by stationary skillsets, movement traits and physical force. All items are indirectly observable through film and box score watching, and therefore have less overlap with each other as tied to direct observation. It will take some training to translate each factor distinctly, but that’s what Swish Theory is for.
Implementing a beta version of this model shows one clear beneficiary who my previous method may have been too low on: Gonzaga’s Anton Watson. While I was still higher than consensus before, now I wonder if he is a legitimate rotation piece. Here’s why.
In my previous model, I ranked Watson low in production, high in feel/processing and mediocre in athleticism. The low usage rate for a super senior was the production red flag in particular. But considering the new model, what exactly does Watson not accomplish on the court?
Anton Watson had a 9.2 Box Plus-Minus, second in the WCC to Mogbo, which he accomplished partially by being skilled for his size. At 6’7’’, Watson can dribble some, shoot some and pass some, all while being a high stocks player on defense. Getting steals is skillful, and Watson has some of the best hands in class to help him do so. While not much of a shooter over his college career, he came alive this season as a 67th percentile efficiency spot up shooter and 74th percentile efficiency on runners. His touch was elite on layups, at 92nd percentile efficiency. He can set strong screens. There are not many areas of the court where Watson can’t have an impact. There is our first dimension.
Watson is also a good mover, not necessarily through mobility (though that’s fine) but intentionality. Watson is always in the right spot, leading to a very good 2.8% steal rate and 2.1% block rate while only fouling three times per forty minutes. He also advanced his driving, up by 40% per game from the season before. Not only is he skilled, but he moves to the right spot to utilize that skill. There is our second dimension.
Finally, Watson is strong as f*ck. He is a menace when he has a head of steam, a perfect 30 for 30 on dunks this season, and with a strong 0.44 free throw rate over his college career. Watson is skillful and able to be in the right place and also strong enough to enforce his will. There is our third dimension.
There were other major shifts in my ranking, too.
Players with substantial rises up my board: Anton Watson, Jaylin Williams, Ulrich Chomche, Dillon Jones, Reece Beekman, Kevin McCullar, Jared McCain, DaRon Holmes II, Dylan Disu.
Players with substantial falls down my board: Baylor Scheierman, Cody Williams, Zaccharie Risacher, Dalton Knecht, Matas Buzelis, Carlton Carrington, Alexandre Sarr.
Let’s see what trends we can parse from these differences.
Summarizing all the stats for these groups shows the risers exceling in two areas in particular compared to the fallers: steals (+47% in steal rate) despite a decline in fouling (-9%) and assists (+33% assist rate, +21% assist to turnover ratio). The increase in steals while declining in fouls points to surgical physicality and movement ability, as does the increase in assists with only modest increase in turnovers. This fits nicely with our new conceptualization of “skillful self moving with pace and force.”
The increase in assists, particularly to this degree, may be surprising, but passing is a substantial factor for covering ground. In particular, this method puts a premium on passers with a variety of deliveries, like Tyler Kolek, Reece Beekman and Dillon Jones. The more ways a player can make the ball move, the more space the opponent has to cover.
While Carlton Carrington has a very high assist rate, he still tumbled down my board due to very poor applied physicality: Carrington has the lowest steal rate of the groups and a sub-1% block rate. A player of Ajay Mitchell’s mold, meanwhile, struggles to pick up stocks without fouling (highest foul rate in both groups at 4.7 per 40 minutes). But he makes up for it in applied force due to his shiftiness on offense. With his flexibility and change of direction ability, his defender feels as if he was pushed backwards: that’s force applied by the ballhandler.
It may be surprising to see some names I was already low on among the fallers, specifically Williams, Risacher, Knecht and Buzelis. All four of them have limitations with their passing and steal rates and are below average in applied force. There is a good chance one of the four proves me wrong, but right now I view them as borderline undraftable players. A major divergence from consensus, with validation depending on the eventual results.
Ulrich Chomche is another standout name, rising from UDFA territory to first round consideration. Chomche is very young at 18.5, playing with NBA Academy Africa this past season. A very unusual profile for the three games with stats available, Chomche shot 33% on few attempts from two but 38% on many attempts from three. He did this while also blocking a ton of fouls and almost never fouling. Watching him, this applied physicality jumps off the page, especially for his age. His passing ability and shooting form are both excellent for a 6’10.25” player of his age as well. In the little game tape that is available of Chomche, he can be found making hit ahead passes in transition, operating out of the top of the key and just generally trying things. While still raw, he may be quite difficult to play against in his prime. The bet is appealing because if he hits, he has access to much more of the game and with more tools to act than someone like the 2024 draft’s number one pick.
There will certainly be flaws in this methodology as well, which is what the next cycle is for. We all use a lens to watch the game, whether aware of it or not, so might as well be explicit about it. This new model does not replace tape or data-based prospect analysis; in fact, it supplements it as the exact purpose. While there remain points of clarification, still, let’s check back in a year from now and see if we have gotten any closer.
Tags:
Related Podcasts
No items foundRelated Articles
