Cal's Pro Pipeline
Beating the Draft Board

Aaron Rodgers went 24th. Keenan Allen went in the third round. Jason Kidd went 2nd and still beat expectations. Is that a pattern or a few famous exceptions? I modeled how much career value every NFL and NBA draft pick since 1980 should produce based on where he was taken, then measured which colleges' players beat that expectation most often.

Judge players against their draft slot, not each other

A #1 pick is supposed to be great, so raw career totals mostly measure where a school's players got drafted. The fairer question is whether they did better than a typical player taken at the same spot. That's the same logic as judging a sales rep against quota instead of total revenue.

For football, career value is Pro Football Reference's Approximate Value, one number for a player's total contribution (a solid starting season is roughly 8 to 10). For basketball it's Win Shares, the estimated wins a player added. The grey line shows the typical value at each pick; every dot above it is a Cal player who beat his slot.

Cal ranks #4 of 57, ahead of UCLA, USC and Alabama

Putting both sports on one scale, Cal players beat their draft slot by more than players from all but three schools that send at least 40 players to the NFL and 10 to the NBA. That's ahead of UCLA, Kentucky, Stanford, USC, North Carolina and Alabama. Football alone tells the same story: #4 of 61 schools with 60+ picks, and 40% of Cal's NFL picks beat their slot versus 32% of all picks.

It isn't just Aaron Rodgers

Rodgers produced 169 career AV against 41.2 for a typical #24 pick, the single biggest gap. But a result driven by one player isn't a pattern, so I reran the football ranking without each school's best player: Cal still ranks 5th of 61. The list runs deep, and much of it comes from the middle rounds, where teams pay the least for talent.

Cal's biggest NFL outperformers (career AV)

Cal's NBA outperformers (career Win Shares)

In basketball, Jason Kidd added 63.7 Win Shares beyond a typical #2 pick, and Jaylen Brown is already +14.4 past a typical #3 pick from his draft era. With only 18 Cal picks in the first two rounds since 1980, basketball supports the story but can't rank Cal on its own.

Strong evidence, not proof

Cal's 95% interval in the combined ranking runs from -0.02 to 0.37. It just touches zero, which means I can't fully rule out that Cal is average and got lucky. With 57 schools in the race, a few will land near the top by chance. What makes it more than luck: in football, Cal ranks 4th on the average, 5th on the share of players who beat their slot, and 5th with its best player removed. Those are different ways of measuring the same question, and they agree.

What a front office could do with this

1. Treat school as a tiebreaker, not a signal. The effect is real enough to break a tie between two similar prospects in the middle rounds, not big enough to move a player up a draft board.
2. Look for the mechanism before paying for it. Coaching, conference competition and pro-style systems are all candidates. A school effect only helps if it survives a change in coaching staff, which is the next thing I'd test.
3. Reuse the method. “Actual versus expected given the price paid” works anywhere value is bought at a known price: sales hires, marketing channels, vendor contracts.

• Expected value comes from isotonic regression: the best-fitting curve that only goes down as the pick number goes up, with no other shape assumed. Each draft class is compared only with classes within 3 years, so players still mid-career (like Jaylen Brown) aren't judged against finished careers.
• Draft classes 1980 to 2020: 11,376 NFL picks and 2,402 NBA picks (first 60 each year, the size of the modern draft). Players who never played count as zero.
• Approximate Value and Win Shares are single-number summaries and undervalue some roles (offensive linemen, defensive specialists). Players are credited to the last college they attended.
• NFL data from the open nflverse project; NBA data from Basketball Reference draft pages.