SMILODON 3.5: Hello Wingspan, My Old Friend

Four score and a few years ago, minus the four score, I introduced the SMILODON draft projection system. SMILODON, which stands for Skills Model In Lieu Of Dexterity Operating Numbers, is an attempt to peer into the future by isolating specific, important basketball skills from a player’s statistical profile. Rather than running sophisticated statistical regressions, I decided to keep the model to mostly simple arithmetic formulas that create a series of weighted averages in different categories. The players are rated in those four categories on a color-coded scale from “Poor” to “Elite,” with a “+” or “-” sometimes attached to the rating to indicate if the player is at the very top or bottom of that range.

These terms are intended to compare players to an average first round prospect – “Decent” shooting means “decent by the standards of a first round pick.” This allows for determination of the top overall prospects, but also means that similar players by this method will often be stylistic matches – a cromulent consideration when evaluating a player ‘type.’

For years, I have resisted the urge to transform SMILODON into an all-in-one metric. The formulas are admittedly a series of kludges and there was no a priori reason why it should work when mushed together. It didn’t, quite, so I added a few more kludges to penalize or reward certain combinations. There is now a numerical ranking but, as with imbibing dangerous substances or operating heavy machinery, there is no substitute for good judgment. Each of the components is listed and explained below.

Finally, I have run pieces explaining the system before, but there have been so many changes since the last one that I thought it would be good to re-introduce the system almost as if it was new. For a sneak preview, here are the top 30 players from 2018 as ranked by the system, albeit with incomplete data and some small sample sizes. Never seen this before? Wondering what any of this means? Read on to find out!

(All box score stats are per 40 and pace adjusted. PBP counting stats are per 40 but not pace adjusted.)

Perimeter Players

This includes everyone who isn’t a center, and there are four categories plus age. The first of these categories is shooting, which is a blend of free throw percentage, three point makes, jumper percentage, and height. The first two categories are the most important, sometimes leading to players with low shooting percentages being rated as “Decent” or “Good” shooters. Sometimes this works well, as in the case of Donovan Mitchell (rated Decent +). Other times, the system will miss on one of these players, as with R.J. Hunter (Good -). Shooting proficiency generally takes time, so it makes sense to be excited about players who can shoot at 18 or 19, even if the rest of their game isn’t there yet. On the other hand, someone who still can’t shoot by 21 is extremely unlikely to learn.

The second category is driving, which is a weighted average of shots made at the rim (in the half court) and free throw attempts. This is often indicative of players with some combination of strength, handle, and speed. Players who can get to the basket effectively in college can often translate that offense into the NBA, like T.J. Warren, Kyle Kuzma, or Collin Sexton. However, players who can drive but don’t develop enough of a shot to keep the defense honest often find that those lanes get closed off, like our own Jarrett Culver or Elfrid Payton.

Next, we have point skills, which measures assists and turnovers*, modified by height. This is obviously important for guards, but forwards with poor point skills can disappoint despite a good overall statistical profile, as Jabari Parker, Stanley Johnson, and K.J. McDaniels have proven. Point skills synergize well with shooting, as players who can both shoot and make good decisions generally succeed in today’s NBA. That being said, “Good” point skills, but “Poor” driving throws up a red flag, as those players may not put enough pressure on the rim to open up passing lanes, as Kendall Marshall and Denzel Valentine have found out. Point skills can also be an indicator of basketball IQ, which can be a useful indicator for a player who may succeed defensively.

(*all turnovers in SMILODON are adjusted for usage)

Finally, players are evaluated on “do shit.” This category combines, in order of importance, steals, blocks, offensive rebounds, and defensive rebounds. It is not exactly a measure of defense, but is a good indirect indicator of functional athleticism, aggression, and reaction speed, three extremely important qualities. Players who “do shit” and can shoot are often great role players, like Mikal Bridges and Robert Covington. Some even develop a broader skill set around that ability, like Jayson Tatum or C.J. McCollum (in his last full NCAA season, not his injury-shortened 2013). It is also important to mentally downwardly adjust players from zones like Syracuse’s that have allowed players like Dion Waiters, Tyler Ennis, and James Southerland to gamble their way into a SMILODON rating beyond their actual capabilities.

To arrive at a final rating, these categories, including age, are converted to numbers (Poor = 0, Subpar = 1, etc; pluses and minuses are +1/3 or -1/3), then added to a physical rating based on wingspan/height differential, weight adjusted for height, and standing vertical leap (given an artificial minimum of 26 for calculation purposes). That means that players without a known standing vertical require a slight mental adjustment up or down, usually between -1 and +1, to their overall rating based on the perception of their explosive athleticism. There are also some artificial maximums and minimums for extremely short or incredibly long players. Finally, there are a number of penalties or bonuses based on some of the commentary above.

The first of these is the initiator bonus, given to players who have a combined value of more than 8 in the shooting, driving, and point skills category. This is based on the observation that proficiency in all three of these skills is multiplicative and makes a player nearly impossible to guard. There are only three players in the nine years of SMILODON data to receive this bonus: Luka Doncic, Trae Young, and Damian Lillard.

Players who do not receive the above bonus can be given one of two smaller bonuses. The first is the 3&D bonus, which can be given to players with a shooting rating of at least “Decent –” and a “do shit” of “Elite –” or players with a shooting rating of at least “Decent +” and a “do shit” of at least “Good.” The other bonus is “Skill”, given to players with a shooting of at least “Decent +” and point skills of at least “Decent –”. In the 2019 1^st round, Chuma Okeke, Matisse Thybulle, and Dylan Windler got the 3&D bonus, while Coby White and Cam Johnson got the skill bump.

As shooting is so important, especially adjusted for age, there is a “shoot” bonus for players who have a shooting rating of “Good” or better at the May 1st age of 19.9 or younger. Tyler Herro, Bol Bol (on a tiny sample size), Trae Young, and Gary Trent Jr received this bonus in the 2018 and 2019 drafts. There are corresponding “no shoot” penalties for perimeter players who have a “poor” shooting rating at 21 (a small penalty) or 22 and older (a larger penalty). There are players who didn’t shoot well in college who learned in the NBA – OG Anunoby and Jaylen Brown come to mind – but just about all of those players were 18 or 19 at the time of the draft.

As driving is so contextual, there are a couple more penalties involving that score. The first, “point”, is for players who have at least a “Decent +” point skills rating, but a “Poor” driving rating. As these players rarely live up to the hype in the NBA due to their inability to finish inside, Lonzo Ball being a recent example, I thought it fair to downgrade the effect of their passing skills. The other pair of penalties, denoted as “brick”, is for players who post “Elite” driving with “Subpar” or “Poor” shooting (the worse the shooting, the bigger the penalty). With that match of skills, Jarrett Culver and Jawun Evans have not found the same opportunities at the rim in the NBA as they did in the Big 12.

Finally, unlike for international players and bigs, both of whom have less control over their team context, there is a sliding scaled penalty, dubbed “lose”, for perimeter players whose NCAA teams won less than 70% of their games. If you can’t win games as a college offensive initiator, it will be even more difficult in the NBA. For international players, I have not instituted this penalty, as an 18 year old lead initiator will find it far more difficult to win games against adults than against Georgia Tech, largely through no fault of their own. Many international prospects also play smaller roles on their teams than NCAA players do, giving them less control over the outcomes of the games.

Centers

There are also four ratings for centers, some of which may look slightly different to those of you who have been paying close attention, as I made a few tweaks to the “awareness” and “finishing” formulas in the past week.

The first is shooting, which is extremely similar to the rating for perimeter players. Since that doesn’t need any more explanation, I’ll move to awareness, which is a big man only rating that incorporates assists, turnovers, steals, and age. This is similar to a rating that had existed for a couple of years, but I very recently made some small changes to the formula to add age as a factor and de-emphasize turnovers. Adding age is a double dip, but I thought it was justified as big men naturally see a fairly dramatic improvement in assists over a college career – more so than most other stats – and I did not think the current age adjustment was sufficient. Bigs with “Good” or “Elite” awareness includes some late round surprises such as Nikola Jokic, Larry Nance, and Dwight Powell, but also some misses, including Frank Kaminsky and T.J. Leaf.

After awareness, we have “finishing”, which includes shooting percentage at the rim in the half court, free throw rate, offensive rebounds, a penalty for turnovers, and standing vertical leap (sometimes conservatively estimated). This attempts to identify those players who have the athleticism and power to pressure the rim at the next level, while filtering out the stone-handed lummoxes who never venture from the rim.

Recent changes added offensive rebounds back into the formula while emphasizing % at shots at the rim over volume. It is very difficult to succeed as a prospect while doing poorly in this category. The complete list of first round picks with a “Poor” in this category is Dragan Bender, Noah Vonleh, Omari Spellman, Tyler Lydon, Henry Ellenson, Skal Labissiere, Chris McCullough, Adreian Payne, and Livio Jean-Charles. It’s not impossible to succeed with that rating, but Dwight Powell, Thomas Bryant, and Dewayne Dedmon are the biggest success stories from the past eight drafts, making it one of the more predictive single categories. Intriguing international prospect Aleksej Pokusevski will attempt to buck that trend this year (with small sample size caveats). This category pairs well with “awareness,” as players who have both good court vision and finishing ability tend to be very effective offensive players.

Finally, there is “rim protection,” which is just block rate adjusted by standing reach. Players who can protect the rim and do just about anything else – so, not Ike Anigbogu – tend to develop into good NBA centers. Most recent blue chip center prospects, like Karl-Anthony Towns, Joel Embiid, and Rudy Gobert, have excelled in this category, even if that defense hasn’t always translated.

There are a couple bonuses for centers, including a rim bonus for players with at least a “Good” rim protection score, filtered for age, which can be added to a “scorer” bonus for big men who have at least a “Decent –” in both awareness and finishing. The majority of players with those bonuses over the past eight years have been worth their draft spot and then some. Even beyond the lottery, this method picked out Robert Williams, Brandon Clarke, Pascal Siakam, Nikola Jokic, Clint Capela, and Jusuf Nurkic. (And a few busts, to be sure.) The only player with that bonus this year is Obi Toppin, providing a reason for optimism despite his disappointing overall profile.

Example

Now, let’s put this all together with a look at the 2016 lottery. How does this system rate that year?

Ben Simmons had a lot of blue and green in his profile. Now, the poor shooting combined with the “Elite –“ driving leads to the “brick” penalty, while LSU’s relatively unsuccessful season leads to a “lose” penalty. However, the combination of weight and an eye-popping no-step vert gets that back (and more) in the physical bonus. It is worth noting that No-Step Vert number comes from LSU’s pro day and should be taken with a heaping of salt. Even if that number were knocked down to “33” or “34”, which might be more accurate, his rating would still be right around “13”, still the highest in the class by a significant amount.

Names in red posted their numbers in a small sample size, while Wingspan and No-Step Verts in red are my conservative estimations for players for whom I thought it would make a difference for this example (under 30 provides a penalty, over 30 a bonus). Dragan Bender’s No-Step Vert was actually lower, but, as mentioned above, I instituted an artificial minimum at 26.

Murray’s No-Step Vert is also a bit suspect, coming from Kentucky’s Pro Day, but he also receives the “shoot” bonus as a young player who could really shoot. His generous vertical also helps make up for the Calipari effect, under which many UK prospects put up worse numbers than their talent should allow under his system. He and Ingram were clearly the 2^nd and 3^rd best prospects on this list, with Papagiannis playing under 300 minutes against backups in Greece (with a vertical that might have been worse than my estimation).

Jaylen Brown and Domantas Sabonis were underrated by this method. Brown’s case is simple: he couldn’t shoot at Cal and learned how in Boston. If Jaylen was merely a “Subpar” shooter in college, he would have been on the same level as Ingram and Murray; “Decent” and he would have been above both. Sabonis is more complicated. He shot 72% at the rim in the half court, which is good, but not fantastic by college big standards. His free throw rate was good, but not great. He posted some assists, but his assist to turnover rate was actually mediocre. His best skill was defensive rebounding, which hasn’t been very predictive of NBA success in other cases. His other defensive stats were unimpressive. At the end of the day, he translated interior scoring and defensive rebounding in the WCC to interior scoring and defensive rebounding in the NBA to a nearly unprecedented degree.

(It’s also worth noting that the model has underestimated a few players with NBA family members, including Sabonis, Tim Hardaway Jr, Seth Curry, and Jalen Brunson. There may be some social/cultural factors there, or perhaps only survivorship bias. I haven’t looked at the issue systematically yet.)

Denzel Valentine is an interesting example of a player who received both a penalty and a bonus. Valentine could both shoot and pass, but also couldn’t get to the rim at all. In his case, the extra flags cancelled each other out – but his physical measurements brought his overall profile down, thanks to his substandard vertical leap. The other players in this draft given at least a 9.2 rating by SMILODON were Wade Baldwin (9.5), Caris LeVert (9.5 – SSS), Malcolm Brogdon (9.5), Isaiah Whitehead (9.2), Zhou Qi (10.5 – Chinese League), and Derrick Jones Jr (9.5 – undrafted).

2020 NBA Draft

Now, let’s look at this year’s draft class:

This is the ESPN top 15 and the first thing to note is that there is a lot of red (small sample size) and brown (international) here. In addition, there are no standing verticals and I’m much less confident in the accuracy of these height, wingspan, and weight measurements. Lack of accurate physical measurements, especially for top or obscure prospects, is one of the main barriers to refining the model, and in some cases, I have been forced to estimate (numbers in red) to put together reasonably accurate projections.

Killian Hayes, for instance, is listed anywhere between 192 and 215 depending on the source – a discrepancy that could move his overall rating from an 11.7 to 12.3. For now, I’m keeping him at the low number, but I suspect that will be changed before the draft. His team also won a fewer percentage of its games than Anthony Edwards’ team, but I think it’s harder to pile up wins at the professional level than against a cupcake non-conference schedule.

The last top 20 lead ballhandler to come out of the BBL was Dennis Schroder, whose team won only 37.5% of its games. Schroder was a very different prospect in many ways, but has not underperformed his SMILODON projection, which rated him the 16th best player in his draft year.

Onyeka Okongwu is another player who could really benefit from better measurements, as a good standing vertical from him could bump his finishing rating to a “Decent +” or “Good –“ as well as juice his “physical” score, vaulting him several spots higher in the statistical rankings.

James Wiseman is technically the top prospect by this method – though the recent update tanked his awareness score – but his hold on that spot likely would not have held up as he faced better competition, as is indicated by his 8.3 in the extremely small sample size of the one game he played against a functioning college team. R.J. Hampton has a pretty good overall score, but it is held together by a good steal rate in a small sample size. Skinny guards who can’t shoot or dribble very well tend to not work out very well (see: Smith, Zhaire). Athletic testing would be very important for him to do well enough to be taken in the lottery by this method.

Tyrese Haliburton is a roller coaster of bonus and penalties unto himself. He can both shoot and pass – that’s good! He’d also be eligible for the 3&D bonus if he hadn’t received the skill bonus! But he gets the “point” penalty because he can’t get to the rim! (And the ensuing Lonzo comparisons.) He also gets the “lose” penalty because Iowa State had a losing record in games he played. That’s really bad for an NCAA starting point guard! He’s also extremely light for his height, giving him a penalty for his physical attributes.

Even after all of that, however, he still rates ahead of four of ESPN’s top ten – Achiuwa, Okoro, Avdija, and Toppin. It is certainly possible, though, that some of those players could pass him if standing verticals were measured. I now have Haliburton rated 9^th on my board, despite some tantalizing skills, mostly due to what I see as the fragility of that player type.

Some of the other players the model currently likes will probably be downgraded once the model learns they can’t jump over a phone book (Mason Jones), but Aleksej Pokusevski (11.2 – SSS), Kira Lewis (9.3), Theo Maledon (9.0), Devin Vassell (8.7), and Saddiq Bey (8.7) have the best ratings of the players ESPN has ranked 15-30. Of the next fifteen, Devon Dotson (9.3 – can’t shoot), Malachi Flynn (8.9), and Payton Pritchard (8.7) seem like the top guards, while big men Udoka Azubuike (8.8) and Vernon Carey (8.3) look pretty good, at least without a standing vertical score to drag Carey’s rating down. I do subjectively like Xavier Tillman (7.4) and Desmond Bane (7.9) a little more than their ratings suggest and am a little worried about Azubuike’s mobility, but I’ll elaborate more on these individual ratings in future articles.

I’ll be using these numbers to look at this year’s prospects in more detail in future installments, including lots of similar player comparisons, but for now I’ll leave you with the link to the spreadsheet going back to 2012 and an open comments section.