Blind Height-Pilling – A Caution from a Stats/Eye Guy
There's more to hoops than being tall (though it obviously helps!)
**INTRODUCTION**
Hi all – thanks for the support and the read always… it’s the dregs of the basketball calendar with Summer League wrapped and the college and pro seasons months away. It also happens to be the least busy time of the year at my day job – fortunately (or unfortunately if you don’t enjoy what I’ve just produced) I decided to take advantage of that and loop back to a somewhat controversial claim that I made a while back.
In my piece from a year or so ago, “Who Might Be A Guy for Next Year?” I put a pin in the claim that statistical indicators tend to favor okay players on good teams (still agree) and productive big men (what I’d like to address here). In hindsight, perhaps this claim was a bit too broad strokes... rebounding is good, scoring easy points efficiently is good, and being harder to shoot over/pass around is good. If I want to actually make that claim, it’s necessary to investigate what is being scored/valued in these formulas, and be more granular about how this can or should be applied to player evaluation. There’s a growing sect in draft-twitter that has become increasingly “height-pilled” – “he’s enormous and he produced, therefore he’s good,” – I think that’s a mistake in a specific, identifiable way.
**1) THE BPM COEFFICIENTS – WHAT IS ACTUALLY IN HERE?**
Basketball-Reference publishes Daniel Myers’ full BPM 2.0 regression; the take that I somewhat gave – “the formula rewards bigs” – oversimplifies what’s actually in it. Per 100 possessions, comparing the Position-1 (point guard) coefficient to the Position-5 (center) coefficient:
- AST: 0.580 → 1.034 (nearly double for centers)
- DRB: 0.116 → 0.181 (about 56% higher for centers)
- ORB: 0.613 → 0.181 (guards actually get credited *more* per offensive rebound than centers)
- STL: 1.369 → 1.008 (worth less for centers)
- BLK: 1.327 → 0.703 (worth about half as much, per block, for a center as a point guard)
That last two cut against my narrative – a steal or block from a center is worth meaningfully less, per event, than the same stat from a guard. However, it’s important to remember and adjust for volume, given centers rack up several times the blocks and defensive rebounds of guards. Thus, even a discounted per-event coefficient produces a bigger total contribution. The most significant “lift” for bigs lies in assists and defensive rebounds specifically – the two categories where centers get a HIGHER per-event coefficient AND accumulate a lot of them.
There’s also no explicit “center bonus” in BPM’s position adjustment. The constant is a “point guard penalty” — it sits at 0 for anyone SF or bigger and scales down to -0.818 at true point guard, justified (per Myers) by the idea that guards’ defensive value is least captured by the box score. That’s a real, stated, and potentially valuable design assumption … rim protectors are able to guard a larger proportion of the part of the floor where shots come from on their own, and on-ball defenders are really only dealing with one player or one action at a time (let alone some guards who stand and defend the corner because they’re traffic cones). However, it doesn’t really tease out who amongst each position is a good defender and mediocre center defenders are less penalized than their guard counterparts because of it.
**2) ANOTHER MECHANISM: ASSISTS AS A PROXY**
Arguably more important than the coefficient table itself is that in Myers’ own writeup, the reasoning for weighting a center’s assists so heavily isn’t really about playmaking value. Myers states that “post players that pass well are typically better defenders,” and separately, in the offense/defense split, “assists are a significant indicator of defensive skill for bigs.” BPM isn’t just crediting a center’s assist as an assist; it’s using assist rate as an instrument – or stand-in for a defensive skill (processing, discipline, being where you’re supposed to be) the box score can’t observe directly. That’s a legitimate modeling move, the same kind you’d use in a factor model when you lack a clean measurement of the ability/outcome you actually want. But an instrument is only as good as its historical correlation with the target – which may prove to be a bit misleading.
**3) BPM WAS FIT, NOT DESIGNED**
“The formula is rigged for bigs” – the incorrect interpretation of what I said a year ago, my fault for being vague – implies that the model was fit to intentionally inflate centers. That’s not what happened. BPM’s coefficients come from a linear regression of box score inputs against a Bayesian-prior-informed Regularized Adjusted Plus-Minus (RAPM) basis, built from four separate five-year windows spanning 1996-97 to 2015-16. If centers’ defensive rebounding and assist rates got a bigger regression weight, it’s because, across those two decades of actual on-court plus-minus data, box-score-productive centers really did carry predictive signal for team performance. That’s empirically true, not just an assumption.
What this is telling us is that the centers who existed and earned heavy minutes in that specific sample were good at these things. A rebounding, passing 5 who held down rotation minutes for two decades of NBA history was, on average, a good player – survivorship bias is carrying the load here. The regression can’t distinguish “rebounding and passing are causally valuable for a center” from “the kind of center who rebounds and passes well historically also tended to have the mobility and processing that actually drove winning… and we’re only picking up the correlated bundle.” A new prospect who matches the box-score profile (high AST%, high DRB%) without the underlying mobility that historically came bundled with it will still get the model’s full credit for the real thing. That’s not hypothetical – there are players who fit this mold who have failed and more who will continue to fail despite being propped up for being large and efficient.
**4) THE OBVIOUS “NON-GUYS”**
Enough on theory and coefficients – there has been a clear pattern (one that the NBA has largely learned to avoid) rather than a single anecdote; one guy busting proves nothing, several guys with the same profile failing to pan out for similar reasons should at least demonstrate some causality. Hunter Dickinson, a player who still has fans on drafttwitter (why….?), isn’t an outlier. Dickinson is one of the more recent instances of recurring type: the multi-year, back-to-the-basket, heavily-accoladed college center whose production never finds a real NBA foothold.
- **Hunter Dickinson** — five years, two blue-blood programs (Michigan, then Kansas), essentially uninterrupted double-double production (17.9/10.9 his first year at Kansas, 17.4/10.0 on 52.6% shooting his last). Measured a 7’3.35” wingspan and a 9’1” standing reach at the 2021 combine on a 7’0”-plus frame — genuinely plus length for the position. Went completely undrafted in 2025; signed a two-way deal with New Orleans as a free agent. Every scouting outlet cited the same thing: can’t defend in space, liability in a pick-and-roll, no age-based upside at age 24.
- **Kofi Cockburn** — two-time consensus All-American at Illinois, the only player in the country averaging 20 and 10 in his final college season. An even longer prospect than Dickinson: 7’4.25” wingspan and 9’2.5” standing reach on a 293-pound frame at the 2022 combine. Undrafted in 2022. Didn’t even make it through an E-10 invite with the Jazz. Of this group, Cockburn is probably the single most useful mental model to keep on hand: when a college big man’s per-game numbers (and his measurements) get you excited, the honest base-rate comparison for “productive, physically dominant, back-to-the-basket college center” is closer to Cockburn than it is to an NBA rotation player. That’s an uncomfortable but more accurate prior than the one draft-stats culture tends to default to.
This is the proxy breaking exactly the way Section 3 describes. BPM’s regression learned that a productive rebounding, high-AST% center is valuable because, historically, the centers who put up that stat line usually had the mobility and processing bundled in too. Dickinson and Cockburn are cases where the production is completely real – multiple national awards, program and conference records, genuinely elite length by any measure – and the bundled trait that’s supposed to come along with all of it, the ability to guard in space, mostly just isn’t there. In both cases, the market (30 front offices, collectively, choosing not to spend a single pick) priced that in before either guy ever logged an NBA possession.
**5) PREMIUM CAPITAL PICKS, SAME RESULTS**
The undrafted cases are the cleanest evidence, but they leave an obvious rebuttal on the table… maybe front offices are just risk-averse late in the draft, and a team willing to actually spend real capital on this archetype would get a different result. It doesn’t hold up. The same profile has failed just as often near the top of first rounds, when teams had every incentive in the world to be right and spent premium picks on it anyway.
- **Hasheem Thabeet** – two-time Big East and NABC National Defensive Player of the Year at UConn, 4.2 blocks a game and 417 for his career (2nd in program history), a 61.1% career field goal mark that STILL stands as the school record, Second Team AP All-American. He measured a 7’6.25” wingspan and a 9’5” standing reach at the 2009 combine. Memphis took him 2nd overall, directly ahead of James Harden and Steph Curry (yeah, that Harden and Curry). His NBA career lasted five seasons across four teams: 2.2 points, 2.7 rebounds, 0.8 blocks a game in 224 total appearances. As complete a translation failure as a top-3 pick has ever been… physical profile and all. He’d have some massive fans on drafttwitter in 2026…
- **Jahlil Okafor** – National Freshman of the Year and ACC Player of the Year at Duke on a national title team: 17.3 points, 8.5 rebounds on 66.4% shooting as a true freshman, an elite low-post scoring package by any era’s standard. Philadelphia took him 3rd overall in 2015. His NBA career settled at 10.3 points and 4.7 rebounds a game across four teams before he moved to international leagues – a real NBA player, unlike Thabeet, but nowhere close to what a 3rd overall pick with that kind of offensive résumé should have become, because he was essentially unplayable defensively against modern offenses. Okafor is the clearest case in this whole piece of skill alone not being enough… all the offensive polish in the world, sunk by the exact same space-defense gap that got Dickinson and Cockburn.
- **Mohamed Bamba** – led the Big 12 and ranked 2nd nationally in blocks as a Texas freshman (3.7 a game), plus 10.7 rebounds a game, on the strength of a record-setting 7’10” wingspan (longest in combine history, for what it’s worth) and a 9’7.5” standing reach, second-longest ever recorded. Orlando took him 6th overall in 2018. Seven years and six teams later (Magic, Lakers, 76ers, Clippers, Pelicans, Jazz), he’s a 364-game journeyman averaging 6.7 points, 5.4 rebounds, 1.3 blocks – never a real rotation piece anywhere, despite having the single most extreme statistical AND physical profile of anyone in this piece. Bamba had a pretty impressive sheets profile, but it was pretty clear though most stages of his career and development that he lacked natural feel and defensive positioning to use his immense physical gifts.
Wingspan and standing reach are appealing numbers for exactly the same reason BPM’s AST% and DRB% coefficients are – they’re clean, objective-looking proxies for something real – length should mean more shots altered, a bigger rebounding radius, easier finishes over smaller defenders. But a proxy is only as good as its tether to the underlying skill, and for a center, that tether runs entirely through mobility: can he actually get into the right position to use the length before it even matters. Thabeet and Bamba both measured among the longest wingspans and standing reaches in combine history, and neither one became what those measurements were supposed to promise – the length was real, the mobility and feel to cash it in just wasn’t.
**6) WHAT ACTUALLY SEPARATES THE HITS**
It’d be lazy to just stack up busts and call it a day here – plenty of centers from this same stretch turned into legitimately good, young NBA players, and the fair test is asking the same question of them: were they pure production/BPM darlings too, or is something else doing the work? Rather than compare this with players who were allocated elite draft capital and therefore a greater opportunity set, I think it’s more useful (and more honest) to look at the lower-profile guys… the ones who didn’t quite get the hype but still found real NBA roles; this is the more common and replicable path. Given that this whole piece has been about actually reading the advanced numbers instead of just eyeballing a box score, it would be hypocritical not to go pull them for this group too.
- **Day’Ron Sharpe** (29th overall, 2021, drafted by Phoenix and traded to Brooklyn on draft night) – the honest outlier of the bunch. His UNC freshman line was modest: 9.5 points, 7.6 rebounds in under 20 minutes a night, buried behind older Tar Heel bigs. Here’s the wrinkle though – his rate stats weren’t bad at all, a 25.5 PER and 51.9% eFG in limited run, but just 2.3 win shares on the season shows you how little of that ever turned into actual accumulated value, because the minutes simply weren’t there. Scouting reports lean on his motor and “elite hands,” and explicitly flag that he “might not be very mobile” and needs to become more switchable – a real, acknowledged weakness, not one anyone’s hiding. He’s carved out a genuine rotation role anyway, but as a backup center in a narrow, complementary niche (rebounding, screening, finishing), not a featured piece. Informative in its own right… even a real mobility limitation doesn’t automatically sink a center if the role stays small enough to hide it.
- **Neemias Queta** (39th overall, 2021, Sacramento, now Boston) – the profile that looks the most like the bust list on paper, and the advanced numbers back that up rather than undercut it: a 26.3 career PER, 13.1 career win shares over 86 games, and he led the nation in defensive win shares (2.9). His junior year alone he posted a 12.2 BPM and a 59.2% true shooting mark – by the numbers, an unambiguous college superstar, at a mid-major program. The difference here isn’t a trait, it’s time. He spent four NBA seasons as a fringe two-way and backup piece before a real breakout in 2025-26 – 10.2 points, 8.4 rebounds, 1.3 blocks across 75 starts, followed by a four-year, $56 million extension. Queta’s a useful reminder that “this profile is risky” isn’t the same thing as “this profile can’t work” – it’s a reason to expect a longer runway and temper your year-one expectations, not a reason to write someone off entirely.
- **Jaylin Williams** (34th overall, 2022, Oklahoma City) – the cleanest example in this whole piece of a center whose reputation was built almost entirely outside the box score, and the advanced stats confirm it rather than hide it: a middling 17.6 PER as an Arkansas sophomore, nowhere near Sharpe’s or Queta’s rate numbers, despite 6.3 win shares built mostly off heavy rebounding volume (10.9 points, 9.8 rebounds, a school single-season record of 364 boards, SEC All-Defensive Team). What actually got him drafted was the stuff no rate stat comes close to capturing – elite closing/rotation timing to draw charges, disguised post position, baiting opposing ball-handlers into bad reads. That’s precisely the kind of processing value BPM’s assist-as-proxy mechanism is trying (imperfectly) to back into, and in his case the box score genuinely doesn’t have a line item for it. Five NBA seasons and 118 games later, he’s still relatively a role player – but he’s that piece on a championship roster – a quality outcome even without a single gaudy number or impressive physical traits attached to it.
- **Dereck Lively II** (12th overall, 2023, Oklahoma City, immediately traded to Dallas) – arguably the least productive college season of anyone in this entire piece by the raw box score (lol box scores): 5.2 points, 5.4 rebounds in 20.6 minutes a game as a Duke freshman. But the rate numbers tell a completely different story than the counting stats do – a 22.6 PER and a 66.7% eFG, with EvanMiya rating him as Duke’s single most efficient player that season. Lively wasn’t statistically invisible, in other words… he was just capped on volume by a stacked, veteran Duke frontcourt, which is a different problem entirely than not being good. What actually separated him, on top of that real efficiency, was functional mobility – an easy lob target who could also survive on the perimeter in a pinch, not just anchor the restricted area. Started 42 of 55 games as an NBA rookie, averaging 8.8 points, 6.9 rebounds, 1.4 blocks in 23.5 minutes. The lesson isn’t “ignore the stats entirely” – it’s that a big man’s counting numbers and his actual per-minute quality (along with a good ole’ eye test for those of us who still do that kind of thing) can tell two totally different stories. It’s worth checking both before you write someone off as unproductive, and context always matters.
Line these up against Dickinson and Cockburn and the pattern holds regardless of draft slot or era: the hits all have a specific, identifiable trait sitting underneath the production. Sometimes that shows up as real per-minute efficiency that a crowded role was masking (Lively), sometimes as a genuine defensive limitation that’s small enough to hide in a narrow role (Sharpe), sometimes as processing and instinct that no rate stat fully captures (Williams), and sometimes it’s just patience – giving real tools years to finally show up in a bigger role (Queta). None of them hit purely because the numbers were big and the player was tall. And outside of Queta, who really was a full-blown statistical darling by any measure, none of these guys were built on gaudy counting numbers either. That’s the actual dividing line, and it has nothing to do with how high anyone got drafted.
**CONCLUSION**
None of this means discount every productive big man – it means be specific about which mechanism is doing the work before you trust the number. If a center’s BPM is being carried by AST% and DRB%, ask yourself whether that’s real processing and mobility or a proxy variable riding on a sample that doesn’t actually look like this player. The same question applies to his measurements… a great wingspan or standing reach is a proxy for length actually mattering on the floor, not a guarantee of it. And the evidence here comes in two flavors that reinforce each other: when teams pass entirely (Dickinson, Cockburn), that’s the market pricing in what the box score won’t; when teams spend real premium capital anyway (Thabeet, Okafor, Bamba), the results are just as bad. Both point to the same conclusion – production and measurements built on bullying overmatched college length, without the mobility and feel to actually use either at NBA speed, is exactly the kind of profile draft-stats culture is currently most likely to overrate, because the box score, the physical testing, and the college-competition context all conspire to make “huge and productive” look identical to “huge, productive, and translatable.” Those are not the same claim. There is always a real, identifiable trait doing the actual work underneath the stat line.