#SLOE: People's Basketball Sim League

Referencing to the height article: viewtopic.php?f=7&t=5799

Players by position are shorter by an inch or so and there is less variation than what we had with the historical draft classes. I stated in the article but I'll do it again, this is not a "fix" proposal because the height is not hurting the league. Height will level out. Instead, the distribution of height for college players is tightly packed compared to what we had before, whereas the historical draft classes had more variations that the league loves (randomness).

I want to formally propose that we edit the College configuration file to match that of the Pro configuration file based on several reasons:

* Let's have more randomness for player height.
* We do not play on the college index and changing the height in the game will not affect any GM.
* We are currently importing players with College height distribution into our Pro game. We are playing Fast Break Pro Basketball 3, not College. Our players should not be pro players with college player height.

The change in this proposal would not be applied to this upcoming college class, as we would need to vote on this. However, if we do wait until the new college index is updated, we would have two college classes with the college height distribution, as the game generates the recruits for the next class of prospects. Something to note.

I support this fix. I'm glad that we will be providing off the books cyborg vitamins to our players before they enter the NBA.

I support this.

I support testing this solution but not implementing until we have data that t reacts the way we think and not creating another “problem”

ballsohard wrote:I support testing this solution but not implementing until we have data that t reacts the way we think and not creating another “problem”

Can we change the config file(saving a copy of the original), create a new game, see what the top prospects look like, and repeat?

I think testing would be good.


Sent from my iPhone using Tapatalk

You can actually look at the values and come up with the distributions it will create. I will post more on this later.

So, as promised, more info. I pried open both the college heights file and the pro heights height distribution files. Here's what I found:

1. In general, the college distributions peak a couple of inches shorter than the pro distributions.

2. The college distributions actually have a little more variance than the pro distributions (they're wider and have a shorter peak).











But the thing that really struck me is if you assume the game has a tendency to create all positions at an equal frequency, the height distribution overall gets problematic. There's a really good reason for paying attention not just by position but by overall distribution - college players, like pro players, get their "listed position" changed from the start of a year to the end of a year... so while a guy may be "created" at one position, he might enter the draft at a different position? How do we know? Well, look at the 2015-2016 Top Prospects thread, taken at the start of the season: viewtopic.php?f=89&t=4724 and compare with the draft preview at the end of that season at http://pbsl.ijbl.net/2015/draft.htm - you'll notice Kenneth Henderson (whom we know as Ben Simmons) started the year listed as a C (where he was ostensibly created) and shifted to PF during the season, and that's where he was drafted. So did Henry Raber and Jonathan Linden. If I recall correctly in our first college draft, we had a couple players who started the year listed at PG but would up playing college that year - and getting drafted - as SGs.

Anyway, here's what I noticed when I combined the weights across all the positions:



1. The college file and the pro file actually both peak at the same height - 6' 8" - but the college players distribution leans more to the shorter end than the pro distribution (not a surprise).

2. The college distribution is more of a smooth curve than the pro one - note the big "dips" at 6-4, 6-6, and 6-10 on the pro curve. I don't think we're quite looking for a perfect bell curve centered on 6' 8" - instead, we're looking for a curve that peaks somewhere between 6'6" and 6'9" (and 6' 8" seems as good a spot as any) but is little lopsided toward the short end. The real NBA distribution looks like this (has a spike at 6' 9" instead of peaking at 6' 8" but since this isn't weighted by playing time, I think it's okay if we aim for 6' 8" which is the peak when weighted by playing time):



What we actually get when we compare real NBA heights to the college and pro files is that the college file is pretty decent - all we need to do is extend the right end of the college curve a little bit to pretty closely resemble the "real" curve observed above. If we do that, the "left side" of the curve is actually pretty decent. The "pro" distribution file makes for a choppy curve, not a smooth one, so it's not a good curve to use.

So after all the sturm and drang about player heights, I think where I finally settle into the debate - after staring hard at all the numbers - is "do we want to put our thumb on the scale to increase the probability of college players being generated at 6' 9" and over? If so, by how much?" And I think the answer probably looks close to "double the weight that a player will be created at each height increment 6' 10" and above."

How do we begin testing this? The new college class is already in so this won’t impact the league for another 2 drafts.


Sent from my iPhone using Tapatalk

NOLa. wrote: Wed Jan 03, 2018 12:38 pm How do we begin testing this? The new college class is already in so this won’t impact the league for another 2 drafts.


Sent from my iPhone using Tapatalk

Just scrap it. If if won't affect the league for 2 full seasons and we're fine with that, then it's not a big enough issue to pursue imo. If this is a problem then it should be addressed asap and put up for a vote. If not, then why are we talking about it still since we're fine with this "problem" tarnishing another 2 full draft classes. Seems silly to me.

Adjusting a particular weight in a config file isn't rocket science. The need for "testing" also seems silly. This isn't hard concepts to grasp, change the weight and you can predict the results based on the current files and weightings. I think if we wanted to do something here it should be put up for a specific vote and voted on and put into place. And if it's something that people think isn't a big deal that we can put it off for several seasons then I can think of many other things that are more important than this.

DarthVegito wrote: Wed Jan 03, 2018 12:50 pm

NOLa. wrote: Wed Jan 03, 2018 12:38 pm How do we begin testing this? The new college class is already in so this won’t impact the league for another 2 drafts.


Sent from my iPhone using Tapatalk

Just scrap it. If if won't affect the league for 2 full seasons and we're fine with that, then it's not a big enough issue to pursue imo.

Well the reason why is because the next batch of prospects have been generated. We could start a new college game for all I care, would be able to have a new class with the new system in place. Also, it's not a bug or a problem, just an observation some GMs made about player's heights in which we found a way to adjust. I think it should be put to a vote. It's a minor adjustment but if we are voting on other aspects of the game then a change to the file we get our rookies from should be voted on. For testing, it's more peace of mind that we don't potentially create another problem by accident.

I don't understand why taking the time to do these things are a make-or-break for making decisions. It's actually a good process to discuss ideas, test and vote. Just because it takes time and work to get it right shouldn't be chopped down because it isn't convenient, and this is not directed at you Darth because you do discuss ideas out, but this last post is reflective of the general attitude of the league sometimes.

I will be putting the following proposed height file out for vote. It gives us a smooth curve, centered at 6' 8" like the old one.



I will be putting the following proposed height file out for vote. It gives us a smooth curve, centered at 6' 8" like the old one.

img=images/DistNew.jpg/img

[C]
60=0
61=0
62=0
63=0
64=0
65=0
66=0
67=0
68=0
69=0
70=0
71=0
72=0
73=0
74=0
75=0
76=0
77=0
78=0
79=0
80=10
81=55
82=120
83=257
84=273
85=180
86=90
87=27
88=13
89=7
90=7
91=1

[PF]
60=0
61=0
62=0
63=0
64=0
65=0
66=0
67=0
68=0
69=0
70=0
71=0
72=0
73=0
74=0
75=0
76=0
77=1
78=24
79=131
80=297
81=297
82=274
83=151
84=93
85=24
86=0
87=0
88=0
89=0
90=0
91=0

[SF]
60=0
61=0
62=0
63=0
64=0
65=0
66=0
67=0
68=0
69=0
70=0
71=0
72=0
73=0
74=0
75=0
76=0
77=65
78=190
79=308
80=215
81=168
82=80
83=25
84=0
85=0
86=0
87=0
88=0
89=0
90=0
91=0

[SG]
60=0
61=0
62=0
63=0
64=0
65=0
66=0
67=0
68=0
69=0
70=0
71=1
72=17
73=34
74=50
75=75
76=225
77=300
78=203
79=69
80=50
81=1
82=0
83=0
84=0
85=0
86=0
87=0
88=0
89=0
90=0
91=0

[PG]
60=0
61=0
62=1
63=1
64=1
65=1
66=3
67=3
68=4
69=4
70=27
71=81
72=100
73=150
74=189
75=225
76=125
77=28
78=18
79=10
80=1
81=1
82=0
83=0
84=0
85=0
86=0
87=0
88=0
89=0
90=0
91=0