Draft Class Data Thread

[adembroski]

EDIT: OK SCREW ALL OF THIS, ITS MEANINGLESS.

The positional class strength is a WEIGHT. The classes are equal across the board, class strength is weighted.


If you set every position as VERY WEAK and every position as VERY STRONG, you will get the same results because you've set all the positions to the same thing.

*******


Alright, I was maybe a day away from finalizing my XP sliders but a few things were making me nervous. For instance, QBs and CBs need to be massively nerfed it seems SO I'm trying to get some good data from the draft classes to see what can be done to mitigate this.


So far I've sampled 6 "Very Weak" draft classes and these are my results. I'm happy with the confidence levels where they are, so I'm going to move on to "Weak" and so forth.

Just wondering if anyone else has done any of this that might save me some work.

(Alpha is .05)

[adembroski]

Re: Draft Class Data Thread

OK, follow up to the above edit.

Best I can figure it, every position is generated the same, but there are a slightly variable number of wild cards (around 10) that are distributed according to relative positional strength and number of players at that position. More testing is needed before I can be sure this is how it works, but that is my working hypothesis.

[Mattanite]

Re: Draft Class Data Thread

OK, follow up to the above edit.

Best I can figure it, every position is generated the same, but there are a slightly variable number of wild cards (around 10) that are distributed according to relative positional strength and number of players at that position. More testing is needed before I can be sure this is how it works, but that is my working hypothesis.

This is how it was explained early on - draft class strength being the relative frequency for high quality (80+), unique and generational talent be generated. My understanding was that it shouldn't change the "rubber banding" generating the rest of the class but either severely reduce, reduce, boost or greatly boost the top of the class. See Franchise Deep Dive extract.

[Mattanite]

Re: Draft Class Data Thread

So far I've sampled 6 "Very Weak" draft classes and these are my results. I'm happy with the confidence levels where they are, so I'm going to move on to "Weak" and so forth.

Good work with the analysis, and good idea getting base data for the very weak classes first as this will demonstrate to you the range of the draft class generator without the "Blue Chip" prospects.

You shouldn't see much of a change to the base data as you move up Draft Class Strength, but the max will move up, the frequency of players near the max will move up and therefore the mean will push upwards (and create a greater spread to the median).

In terms of XP sliders, they need to account for either higher or lower amounts of Superstar and X-Factor devs and less/more 80ovr+ players entering the league each season at young ages. So if QBs and CB need nerfing, setting perpetually to Weak won't help the bulk of the OVR distribution curve but will nerf the number of rookies getting to higher OVR ranges (I explain this in my XP tool thread).

You could always set XP sliders for Normal Draft Class strength and use a Draft Class lottery generator and skew the probabilities if you need to nerf the top end of certain positions? See links in my signature.

[frnkpw13]

Re: Draft Class Data Thread

This is what we need! Keep up the good work!

[adembroski]

Re: Draft Class Data Thread

This is how it was explained early on - draft class strength being the relative frequency for high quality (80+), unique and generational talent be generated. My understanding was that it shouldn't change the "rubber banding" generating the rest of the class but either severely reduce, reduce, boost or greatly boost the top of the class. See Franchise Deep Dive extract.

I'm not sure that's what I'm seeing though. I'm seeing no rubber banding, only a base roster with a small number of "bonus players" who get distributed based on a weighted average. At least, that's what I THINK I'm seeing.

I'll do shorter tests, see if I can get a significant difference. I will say that the numbers I posted for the Very Weak class, IDENTIICAL to the Weak class. Every position spot-on-precisely the same. When I made TEs Very Strong, however, I got those same numbers except 10 TE outliers over 73.

[bwanafulana]

Re: Draft Class Data Thread

I'm not sure that's what I'm seeing though. I'm seeing no rubber banding, only a base roster with a small number of "bonus players" who get distributed based on a weighted average. At least, that's what I THINK I'm seeing.

I'll do shorter tests, see if I can get a significant difference. I will say that the numbers I posted for the Very Weak class, IDENTIICAL to the Weak class. Every position spot-on-precisely the same. When I made TEs Very Strong, however, I got those same numbers except 10 TE outliers over 73.

Have you done more testing?