So these are the issues as I understand them from the BMI haterz.

So lets dive in and see if we can ease the resistance to these issues. https://twitter.com/DFBeanCounter/status/1349190128997076992

We are going to look at historical hit rates while trying to even the playing field on some of this stuff.

So first up lets deal with draft round and production concurrently.

I am going to use a 26.2 cut-off for BMI to make my argument look better. If you want to use a different one, you do a thread on it...

But tbh, isnt that how thresholds work?

Then I am going to use a 30% 'best season' college dominator as the production threshold.

Period - 2003-2018
Filters - >30% best CD, > 26.2 BMI, Round 1

So this is high bmi, high production, round 1.

HitTotalRate
Top 121839 46%
Top 242439 62%

Filters - >30% best CD, < 26.2 BMI, Round 1

This is low bmi, high CD.

HitTotalRate
Top 1228 25%
Top 2428 25%

This is obviously a small sample size but its all we have.

Pretty easily taking the high bmi guys over the low bmi guys.

Here are the players that fit in the low BMI/high production bucket.

FWIW this is where Devonta will fit in.

Next lets look at low production, high bmi.

Filters - <30% best CD, > 26.2 BMI, Round 1

HitTotalRate
Top 1229 22%
Top 2449 44%

Small sample, but not far off the high production/low bmi rates above.

Filters - <30% best CD, < 26.2 BMI, Round 1

HitTotalRate
Top 1202 0%
Top 2402 0%

This is just silly low and doesnt even count, but with it being such a small sample it *could have gone the other way too and be 100%...

These are the two low bmi, low production guys.

So basically the last three are all terribly small samples, but all also trend in the same direction.

Next lets flip to the second round and see what this looks like.

Filters - >30% best CD, > 26.2 BMI, Round 2

High CD, High BMI

HitTotalRate
Top 121338 34%
Top 241738 45%

Filters - >30% best CD, < 26.2 BMI, Round 2

High CD, Low BMI

HitTotalRate
Top 12110 10%
Top 24310 30%

Pretty clearly favors the high BMI group under the high producers umbrella.

Here are the guys in the high CD / low BMI bucket.

Filters - <30% best CD, > 26.2 BMI, Round 2
Low CD, High BMI

HitTotalRate
Top 12112 8%
Top 24212 17%

Filters - <30% best CD, < 26.2 BMI, Round 2

Low CD, Low BMI

HitTotalRate
Top 1205 0%
Top 2415 20%

Here are the low CD, low BMI players from round 2.

@FF_TravisM and @KentWeyrauch (two supersmart dudes fwiw) pointed out that in a regression model BMI is not predictive. I will defer to their expertise as I do not use a regression model.

Travis and I talked about it for quite some time tonight and I think where our hang up is is that I dont want to use BMI in place of production or draft capital or breakout age, etc. and I think people thought I was.

I want to use it to decide between two equal prospects.

It is context that helps with decision making. BMI does not factor into my process as an input. It is context to help me when ranking player x vs player y within the coinflip tier for example

All that being said, I was told that Breakout Age IS predictive within a regression model. So lets look at how it shakes out if we replace BMI with BOA

Filters - >30% best CD, < 19.9 BOA, Round 1

High CD, Good BOA.

HitTotalRate
Top 121732 53%
Top 242232 69%

Filters - >30% best CD, > 19.9 BOA, Round 1

High CD, Bad BOA.

HitTotalRate
Top 12315 20%
Top 24415 27%

Filters - <30% best CD, < 19.9 BOA, Round 1

Low CD, Good BOA

HitTotalRate
Top 1213 33%
Top 2423 67%

Filters - <30% best CD, > 19.9 BOA, Round 1

Low CD, Bad BOA

HitTotalRate
Top 1218 13%
Top 2428 25%

I ran out of tweets on this thread and its late, so lets wrap this up quick.

I had originally asked the question as to what is better between breakout age and BMI for a reason. So lets take a quick look

When I apply filters to round production, and either bmi or breakout age I get remarkably similar historical hit rates.

This is showing good production and good BMI on one and good production and good BOA on the other.

This is showing good production combined with bad BMI on one and bad BOA on the other.

Again remarkably similar.

This is what I didnt get, why are people so comically dismissive of BMI while heralding BOA as the greatest metric of all time?

Turns out its the regression thing

As for the original grievances....

I was and am in no way using BMI as a barometer of player health. It didnt even cross my mind that people would take it that way. My bad.

And the fluctuation one... I get it, BMI can change with a big meal. But if it was a real concern that fluctation in weight would change things, it doesnt bear out in the hit rates. so...

The bottom line is, I would never NOT draft a player because they are low BMI.

So dont think I have just taken Devonta Smith off my draft board. I just probably wouldnt draft him over someone that was the right size.