balance between feeders/roster limits/draft rounds

[NomarHits400]

Originally Posted by RonCo

I've run through a 20+ season study and I see (at least partially) what's happening. I'll be feeding Markus some detailed data here later today. The basic issue is that the average talent in feeder leagues rises quickly each year for about 8 seasons, then hits an equilibrium--so feeder leagues player creation algorithms are not well-tuned to their levels. After about 6-8 seasons, everything comes to an equilibruim and things seem to work out okay, at least.

One interesting chart that indicates what I'm saying is the average age of major league players when viewed over the seasons. It takes a dip for a bit, then recovers....

Age by Season:
28.1682
27.9994
28.1858
28.3087
28.1618
27.7311
27.789
27.6101
27.6443
27.776
27.5768
27.6394
27.8492
28.1211
28.4033
28.4232
28.5763
28.8665
29.2188
29.1241
29.1075

There's more to it--everything is more complex than it first appears. But the good news here is that the algorithm is not fundamentally broken--but it does appear to need some fine-tuning. I'm guessing Markus can improve this fairly easily if presented with the right data. I will keep working on this today and post more as is appropriate.

Thanks for the data, but the average age of the players in a league only really reflects the time it takes a prospect to develop- and I really don't have as big an issue with that. Most of the 5 star potential players I am seeing have relatively low actual ratings to begin with.

The issue that concerns me is, that of those who DO develop into MLB players- a vastly disproportionate amount are studs. It's one thing for a player to take 6 years to develop and enter the league as Rod Barajas- and another thing entirely to take 6 years to develop and become Johnny Bench.....Way too many guys developing into Johnny Bench types- too many good player overall. I am finishing up some test leagues later today- maybe it will help clear some of this up.....

Let me know what you find out.....

[RonCo]

Yes, average age is only one metric. I've got a few more, too.

Average age does provide a sense of how the overall process is working, though, and the fact that the league gets young, then gets older is consistent with a wave of good players (relative to their peers) working through the years. This, I'm fairly sure is the root of the issue. I've posted a bunch of material in the beta area so Markus can review and work on. At the end of the day, my guess is that he'll want to tune the player creation or development algoritms a little.

[NomarHits400]

OK, sounds good....Have you had any luck with any current setting that can reduce the talent base of a feeder league- feeder leagues talent levels are just so vastly different than if you create a league witohut them.

I started a default 30 MLB league with fullminors- 5980 players in the system. The draft itself was comprised of an 800 player pool (I left the draft settings at default)- and NO feeder leagues. This is what I found- all results below are talent levels- scouts off:

9 5 star players: (1.13%)
4 4.5 star players: (0.5%)
9 4 star players: (1.13%)

2.76% 4 star and above...And thsi seems reasonable to me. I am going to try the same league as above- but add a default college and HS feeder and see the difference's.

[NomarHits400]

Originally Posted by RonCo

A few more bits of information. I set up 80 HS teams and 80 college teams as my feeders -- and had a total of 4116 players in my initial HS/College level file. Of the 50 most talented HS and College pitchers in that initial file, 13 were eventually drafted (4 from the HS ranks, 9 from college). 11 of those were in the majors after ten seasons. Five of those were very valuable pitchers. Three were at least a little above average. Three were average or a little below.

Overall, the flow and yield rate itself doesn't seem particularly off--but I am certainly going to suggest some player creation or dev changes to smooth the feeder leagues out a little. Still trying to decide exactly what those changes need to look like...but it's coming clearer.

So are you saying the other 37 top pitchers in that "class"- and by that I presume you mean before you began simming- were not drafted at all??? Do you know what happened to them? Did their talent levels fall off? I know it's hard to keep track of that many players 10 years down the road- but I am very curious....

[RonCo]

Originally Posted by NomarHits400

So are you saying the other 37 top pitchers in that "class"- and by that I presume you mean before you began simming- were not drafted at all??? Do you know what happened to them? Did their talent levels fall off? I know it's hard to keep track of that many players 10 years down the road- but I am very curious....

They're all just in big 'ol spreadsheets.

The talent development model is harsh on young pitchers. A majority had talent hits. Some appear to have had had injuries. A couple just appear to have never grown into their talents--possibly due to being passed over. Dunno. One guy just appears to have been missed. If I get some time I'll dig deeper into the game to see if can find more about his story.

Just as an aside, there were six players in the initial college/HS ranks who signed FA contracts without being drafted and were still in organizations 10 years later. I thought that was kind of cool.

I've also confirmed on the hitter side that most of the hitters who are making the majors are not receiving big talent boosts--of 140 who made the majors something shy of 15 received talent boosts along the way of greter than 40 points (equivalent of an old-timey v5 bump). Only one received multiple big bumps. Most are starting with moderately large talents and just growing into them--the lucky ones who don't dev down or just not grow, anyway. I've posted greater detail elsewhere for Markus to review.

The more I look into it, the more I think it's a matter of player creation rather than talent development.

As a short-term mitigation tactic I do suggest you drop player development speed down to about .7. It won't completely "fix" it, but I think it should make it better.

[NomarHits400]

Originally Posted by RonCo

They're all just in big 'ol spreadsheets.

The talent development model is harsh on young pitchers. A majority had talent hits. Some appear to have had had injuries. A couple just appear to have never grown into their talents--possibly due to being passed over. Dunno. One guy just appears to have been missed. If I get some time I'll dig deeper into the game to see if can find more about his story.

Just as an aside, there were six players in the initial college/HS ranks who signed FA contracts without being drafted and were still in organizations 10 years later. I thought that was kind of cool.

I've also confirmed on the hitter side that most of the hitters who are making the majors are not receiving big talent boosts--of 140 who made the majors something shy of 15 received talent boosts along the way of greter than 40 points (equivalent of an old-timey v5 bump). Only one received multiple big bumps. Most are starting with moderately large talents and just growing into them--the lucky ones who don't dev down or just not grow, anyway. I've posted greater detail elsewhere for Markus to review.

The more I look into it, the more I think it's a matter of player creation rather than talent development.

As a short-term mitigation tactic I do suggest you drop player development speed down to about .7. It won't completely "fix" it, but I think it should make it better.

Yeah, I think that too- the development portion seems ok to me- in fact I rather think the initial ratings of some of the higher rated college players are a little low- I am not seeing any Mark Teixeira types who could probably step straight into an MLB lineup and survive- though admitedly that is pretty rare.

The issue seems to be to me that if you add a default feeder league (one each for HS and college)- and then ADD teams- the number of 5 star prospects as a total % of your feeder league is equal to a smaller feeder league. Problem is, the number of top rated prospects should generally be static regardless of a league's size (with some randomness, obviously)- and the rest of the talent ot fill out the college and HS rosters is just that- filler.

For example, In a non feeder type league- as I mentioned earlier- only 2.76% of the 800 players in the draft pool were 4 star and above- for a total of 22 "top" prospects. However, if my feeder league has an 8000 man draft pool- trouble is brewin' if 2.76% of that class is 4 star and up- it should stay a relatively static 22 "top" prospects- or 0.276%. The front end is always going to be picked first- and I am convinced that, even though the prospects develop at an acceptable rate- at some point in the future- the league is going to be dealing with 220 top players, rather than 22 as in my last example...And this is EVERY year! Let me know if your work is confirming this as well- or if I am still off base.

[spleen1015]

When you guys are doing your testing, are you testing everything with varying league sizes?

[NomarHits400]

Originally Posted by spleen1015

When you guys are doing your testing, are you testing everything with varying league sizes?

Not for me- standard 30 team MLB and full minors -but I don't think this problem has anything to do with that- at least regarding feeder leagues...It just seems to me that the feederleagues are coded for X % of all players to be of a certain talent level- and this really starts to become a problem for larger feeder leagues. So to the extent that you need a larger feeder league to accomadate either A) A larger draft pool (say 50 rounds instead of 25) or B)a larger set of teams that are drafting (say 100 instead of 30)- this will become a problem....And it seems to become a bigger problem (obviously) the larger the pool of players you create.

Interestingly, though- I would think it would be a lot larger problem for scenario A than B, because at least with B, you have more teams with which to disperse the extra talent- whereas with A you may not neccesarily want a more talented draft- just a larger draft. Ideally this would be customizable- say, if the AI could suggest a % of our draft pool- either the default or feeder leagues- as say 4 star and up- and then we could customize up or down to our preferred playing style- and then the game could create the draft pool to reflect this.....Easy for me to say I guess.....Does any of this make sense- this is sort of hard for me to articulate....????

[PSUColonel]

Originally Posted by NomarHits400

Not for me- standard 30 team MLB and full minors -but I don't think this problem has anything to do with that- at least regarding feeder leagues...It just seems to me that the feederleagues are coded for X % of all players to be of a certain talent level- and this really starts to become a problem for larger feeder leagues. So to the extent that you need a larger feeder league to accomadate either A) A larger draft pool (say 50 rounds instead of 25) or B)a larger set of teams that are drafting (say 100 instead of 30)- this will become a problem....And it seems to become a bigger problem (obviously) the larger the pool of players you create.

Interestingly, though- I would think it would be a lot larger problem for scenario A than B, because at least with B, you have more teams with which to disperse the extra talent- whereas with A you may not neccesarily want a more talented draft- just a larger draft. Ideally this would be customizable- say, if the AI could suggest a % of our draft pool- either the default or feeder leagues- as say 4 star and up- and then we could customize up or down to our preferred playing style- and then the game could create the draft pool to reflect this.....Easy for me to say I guess.....Does any of this make sense- this is sort of hard for me to articulate....????

What are your reccomendations for a 30 team MLB style league with full minors? number of draft rounds?

[RonCo]

Originally Posted by NomarHits400

Not for me- standard 30 team MLB and full minors -but I don't think this problem has anything to do with that- at least regarding feeder leagues...It just seems to me that the feederleagues are coded for X % of all players to be of a certain talent level- and this really starts to become a problem for larger feeder leagues. So to the extent that you need a larger feeder league to accomadate either A) A larger draft pool (say 50 rounds instead of 25) or B)a larger set of teams that are drafting (say 100 instead of 30)- this will become a problem....And it seems to become a bigger problem (obviously) the larger the pool of players you create.

Interestingly, though- I would think it would be a lot larger problem for scenario A than B, because at least with B, you have more teams with which to disperse the extra talent- whereas with A you may not neccesarily want a more talented draft- just a larger draft. Ideally this would be customizable- say, if the AI could suggest a % of our draft pool- either the default or feeder leagues- as say 4 star and up- and then we could customize up or down to our preferred playing style- and then the game could create the draft pool to reflect this.....Easy for me to say I guess.....Does any of this make sense- this is sort of hard for me to articulate....????

I need to do a lot more testing to say for sure, but I think it has more to do with the pure distribution of talent in the feeder leagues (which is normal/gaussian/bell-shaped as far as I can tell), regardless of number of teams in a league or in the feeder system. What you say has some element of logic, though. The bell-shaped distribution "feels" right to us humans, but it causes all sorts of problems when we try to assess talent.

I've already posted more detailed information for Markus to look at regarding talent distribution. I'll also do some testing on scaling leagues tonight to try to ascertain the influence those have on the distribution of talent. My guess is that it scales.

At the end of the day, I'm guessing this will not be hard for Markus to address.

[NomarHits400]

Ok, I have found a way to fix this....NOw we just need to come up with the right settings. What you can do:

Create a feeder league team. Then, before filling the feeder with fictional players- go into the parent league's strategy settings, and edit the "Traditional OOTP Creation Modifiers"- as RonCo suggested.....You can get any type of distribution you want- I set them all to 0.1 and actually had a feeder league HS league of all one star players....Whoops! I changed this to 0.85 and re-created the feeder- and now I am averaging about 1 5 star player per team- 1 or 2 fours- and on down the line....Probably STILL a little high- but you get the idea- and it's totally customizable!!!! SWEET!....Not sure if it's a great baseline- but to give you an idea- the default draft with no feeder is producing about 20-25 4 star and up type players...I think this is a good target- for a 30 team MLB with full minors.

The beautiful thing about this is you can then go back in after creating your players for the feeder league and change the modifiers BACK to the default 1.0 for all settings- and then create your parent league without screwing things up. Still needs some work- but I think this may be one way to approach this...I will be doing some more work tonight on a 250 team HS feeder and 128 NCAA feeder for a 3o team MLB with full minors to see if Ican get the right proportions.

[Kemp]

Originally Posted by NomarHits400

Can Markus or RonCo or someone who has a better feel for how the draft is designed comment on a way to make this work...I am a little bummed out that we now have the flexibility to utilize HS and college feeder leagues but are aparently limited to a "magic number" of teams to make the draft operate properly.....Perhaps a variable where we can set the % of draft worthy prospects for our desired feeder system would work- but I don't see how it can be done currently....Help!!!!!!

I think you need to use the Player Creation Modifiers. Let's face it, even if you take the top 125 NCAA college teams, really only the top few percent of those teams have the ability to play MLB. I believe that if you use PCM's to reduce the potentials of the players, you'd be setting it up so that only the true outliers would go on to become very good ML players.

[NDSox12]

Originally Posted by Kemp

I think you need to use the Player Creation Modifiers. Let's face it, even if you take the top 125 NCAA college teams, really only the top few percent of those teams have the ability to play MLB. I believe that if you use PCM's to reduce the potentials of the players, you'd be setting it up so that only the true outliers would go on to become very good ML players.

I agree that PCMs are probably the right solution. I've been running some tests of my own, though not nearly to the level of detail that others have done. The biggest problem I'm seeing while using relatively large feeder leagues is a huge surge in offense over the course of time. I haven't checked on the details of why this is happening, but I suspect it is because the overall talent level has gone way up and the surplus of great hitters is outweighing the great pitchers.

I'm using a 28 team fictional league with 5 levels of minors, a 25 round draft, and two feeder leagues... high school and college, each with 50 teams. This is pretty close to the recommended number of feeder teams. I set the amateur draft to "feeder league only". I have all PCMs and league totals set to the default.

Initially, I accidentally started with only 30 college teams and simulated the first 50 years. The stat totals were actually pretty close to what I wanted. However, when I bumped the college teams up to 50 and simmed another 50 years, things got a little out of hand. Finally, I got rid of the college and high school feeders, changed the draft to "feeder & others" and then simmed another 50 years. The league stat totals returned to the expected level.

I'm sorry I can't quantify any of these figures because I don't have them in front of me, however it seems to me that the key is to find exactly how many feeder teams it takes to create a draft pool similar to what you would get if you had no feeder leagues. Either that, or play around with the PCMs. From my estimation, the formula suggested in the manual for # of feeder teams might be a little high, but I need to do a little more research before making a definitive claim.

[disposableheros]

Originally Posted by NDSox12

In my estimation, the formula suggested in the manual for # of feeder teams might be a little high, but I need to do a little more research before making a definitive claim.

no one formula is going to work for every possible league setup, which we can already see in the other 'feeder league recommendation' threads. IMO, its better to be too high in some scenarios and have extra players, than too low and have the dreaded 'illegal # of players' show up.

[NomarHits400]

Originally Posted by disposableheros

no one formula is going to work for every possible league setup, which we can already see in the other 'feeder league recommendation' threads. IMO, its better to be too high in some scenarios and have extra players, than too low and have the dreaded 'illegal # of players' show up.

You guys are both right- but the issue really isn't about having "enough" players for the draft- that's pretty easy to fix, but rather- as mentioned above several times- having the right number of "quality players". As it stands, the way Markus has things coded (and everything else being equal)- the larger the feeder leagues- the more talent entering the league each year....Which really sucks if you want larger feeder leagues. As has been pointed out by you and Kemp (see my post #31)- there is a delicate balance between feeder leagues and the PCM's to ensure that the league isn't dereft nor flooded with talent. I have come up with a very good set of PCM's for a 128 team college feeder and a 248 team HS feeder...Well, at least I think they are good. At least- they are close to what the game produces with no feeders.

[NDSox12]

Originally Posted by NomarHits400

I have come up with a very good set of PCM's for a 128 team college feeder and a 248 team HS feeder...Well, at least I think they are good. At least- they are close to what the game produces with no feeders.

Are those the 0.85 PCMs you mentioned above? I'll definitely give that a try myself next.

[NomarHits400]

Originally Posted by NDSox12

Are those the 0.85 PCMs you mentioned above? I'll definitely give that a try myself next.

Yes, set the PCM's to this BEFORE feeder league creation- in the 0.8 range and tune to taste- and again that is for feeder leagues of the size I described above.

[NomarHits400]

RonCo,

Do you think there is any chance Markus can modify this in an upcoming patch- because although this is workable- man is it ever a pain- and I just don't think it should take this much thought.

[PSUColonel]

yes, this is a lot of anyone to have to digest. I've been reading through this post, and I am convinced I still have no clue how to set this up properly. I would need reccomendations.

[RonCo]

Originally Posted by NomarHits400

RonCo,

Do you think there is any chance Markus can modify this in an upcoming patch- because although this is workable- man is it ever a pain- and I just don't think it should take this much thought.

I don't have any more insite to when Markus can make a change to address this issue than you do right now. I agree it needs to be improved, though, and Markus is pretty good at prioritizing important things. I'll leave it for him and SI to answer, though.