Simply bad math.
HS players are close to 50% in the first two rounds, then 30% from rd 3-5.
73 in the first ten, is about 2.5 per team, then about 1 more in rds 11-12, at most an average of 3.5-4 HS players per team, not 4-6.
Then teams draft a lot of HS fliers after the 20th rd, the vast majority of whom will go to college.
It's not really tools, it's "projected" tools, the top college pitchers throw harder than most of the top HS pitchers, but the latter have the potential to add velocity.
But the more you have to project how a 18 or 19 year old will physically mature (and character is even harder), the more risk you take on.
Which is why LA players are even higher risk than HS players, projecting a 16 year old for the most part is a fool's errand.
Once you get past the top 2 rounds, the HS players are very high risk since the most talented (the ones that expect to go in the first round or so and made it clear they wouldn't sign for 2nd rd money) won't sign unless you go grossly above slot and the rest are flawed or they'd go higher.
The other issue is you're drafting potential performance, not potential skills. So college players may on average have lesser raw skills, but provide more information about their ability to utilize those skills. So in the first two rounds, the success rate is pretty even, the HS player may end up throwing harder, but the college player who gets drafted has proven he knows how to pitch. A HS player may have the potential to fill out and hit for more power, but you know if the college kid can hit a curve. Arbitrage ensures that these two aspects of drafting even out in terms of expected performance.