Feeder leagues are broken -- what's wrong and how to fix it - Page 3

Mr. Marlin · 05-16-2018, 08:07 AM

I'm curious, does performance follow ratings or vice versa. The reason I ask is I've been following stealofhome on this and have notices some differences in my own observations of feeder players. But it occurs to me the biggest difference is he is looking at ratings. I never see ratings because of the way I play (stats only) I have them turned off. So, out of curiosity I turned them back on and looked at ta sample of changes in MLB level players that were 5+ rd picks. A number of them were performing at a pretty high level stats wise in the feeders and minors (above league average in multiple categories). At the time their ratings actual and potential were average to below average. So, does performance change ratings or does ratings change performance?

stealofhome · 05-16-2018, 09:41 AM

Quote:

Originally Posted by Lukas Berger

That being the case, my biggest concern here is if adjusting the .PCM's is not changing the dp rating or C arm/abil or is changing the ratings incorrectly. Do you have a chart that documents that (maybe I just missed it)? EDIT: Markus fixed this, so never mind!

Awesome news!! I did have a chart prepared but super glad to hear that this was looked into, I can't wait to try out some new stuff.

Quote:

Originally Posted by Mr. Marlin

So, does performance change ratings or does ratings change performance?

I wonder this also, especially on a stats-only approach (which is the end goal for me). I guess in the end the stats are what really matter so even if a player is rated 20/20 in MLB but provides good value then who cares what their ratings are. I almost don't want to look into it because then once I turn the ratings off I might be able to figure out what they are from the stats.

Mr. Marlin · 05-16-2018, 10:40 AM

Like Brad Pitt said in the Moneyball movie "If he's such a good hitter why doesn't he hit good?"

I have enjoyed stats only for a lot of reasons. Not the least of which is the "fog of war" you get by not seeing ratings. But the ratings themselves are subjective unless you have scouting set to 100% accurate. I'd never play that way though. Really it's the AI scout's opinion of a player modified by what that scout's strengths and weaknesses are. It's been debated a lot around here whether stats only is harder or more of a challenge. I'd have to say it is not BECAUSE the ratings are subjective. If anything they may blind you to a player's quality or lead you to buy a white elephant. Of course the better the scout the more reliable the ratings.

On my fictional Miami Tropics team I had a player with an eye rating of 5 (on 1-10 scale) but his slash line is .241/.420/.305. My league average OBP was .330. The following season IIRC his eye rating was 7.

The nice thing about stats is it focuses on what the player is doing right now. Plus, when a GM looks at a player report he does not see colored bars. He sees numbers.

stealofhome · 05-21-2018, 11:04 AM

I've been playing more with the modifiers and it starts getting pretty tricky to make this work. Instead of comparing to the MLB Quickstart, I compare to the first year of a custom league with no feeders. Let the game create the draft pool because that is "working as intended."

As has been seen repeatedly, the first year draft of any league has very high quality and then that decreases rapidly over the first 10 years. This happens in the quickstart, standard leagues, and fictional leagues. One thing I've been ignoring until recently is the batting potential stats and this is where I've noticed diminishing quality. 10 years down the line, hitters have very low contact, gap, and power potential ratings. Pitchers have low movement and control. However, you can't set these modifiers in year 1 because that will cause you to create incredible players. You have to wait until year 7-10ish with no feeder leagues or year 5ish with feeder leagues.

With my feeder league setup as before, I have been getting even better results now with creating the league and letting all modifiers be 1. Sim ahead 5 years and then change the modifiers.

Contact: 1.148
Gap: 1.101
Power: 1.153
Eye: .999
K: 1.023
Stuff: 0.953
Movement: 1.189
Control: 1.099
Stamina: 0.963
Speed: 1.201

Since the players are created at the league minimum age in high school/college and are forced to have 4 years, this means these players don't show up for another 4 years down the road. My year 10 draft doing this was much higher quality than ever before.

The outstanding issue is still the fielding ratings. I have an algorithm that takes the fielding ratings and determines where the player will fit best. The DH% (hitters who are not good defensively anywhere) holds steady around 8-10% for the league with no feeders. When the feeder leagues are turned on though, that number basically doubles to 16-20%.

The problem with that though is that the fielding ratings are not all consistently too low, there are only a few that are not up to par, namely the range ratings. In the 10th year of a feeder league, the average IF RNG is 90% of where it is in the non-feeder draft and the average OF RNG is a whopping 82%. Very few fielders can cover any ground and they end up not having good positional ratings and are consequently less valuable.

The issue with this is that if you crank up the fielding rating modifier, it increases all modifiers when the range stats are really the only ones that need to be modified. I still need to look at some of the other modifiers to see if it is possible to increase range without touching anything else. I also need to see if the ratio of hitters to pitchers is throwing off these fielding numbers.

stealofhome · 05-25-2018, 04:06 PM

I have absolutely convinced myself that the game does not create players correctly, particularly with the ratio of fielding ratings. I looked at the top career players by WAR in a league I simmed forward 50 years and compared it to real life MLB HOF. ~1.4% of players get in the hall of fame, so I found the top 1.4% in WAR for my league. The reason I'm comparing to WAR is because the HOF that the AI created for the league is even worse (absolutely no 2B/SS).

In real life, the HOF is split 69/31 Hitter/Pitcher and quite evenly by position - 7-10% at all position players. The top players in my OOTP league are also split right at 65/35 Hitter/Pitcher - pretty good. But by position is absolutely horrendous. I looked at the majority of innings played by position for the career which was a huge majority for almost all players.

C: 2%
1B: 17%
2B: 5%
3B: 9%
SS: 4%
LF: 14%
CF: 5%
RF: 10%

The IF/OF mix looks right on, but not enough players can play the skilled positions as I have been repeating now over and over and over. There are too many 1B/LF/DH types because the players are not created properly.

stealofhome · 06-01-2018, 09:40 AM

I've been running some more tests since the patch and feeder leagues are definitely behaving differently! I haven't run a long term simulation yet to see what the league turns into over many decades, but this is a step in the right direction.

I wonder if batting eye is treated differently in the way players are created. When I adjust the modifiers, almost all of them need to be somewhere around 1.1 or 1.2 but batting eye stays pretty much the same with no needed modifications.

Another thing that I have done is turned off two-way players which seems to help when changing modifiers as well.

stealofhome · 06-01-2018, 06:01 PM

I finished the 50 year sim and the results are extremely promising. The hall of fame is split well between multiple positions and the drafts are much stronger. (Except there seems to be a bug where no pitchers were put in the hall of fame)

I used these PCM on the day of the MLB draft in year 5 but before it happens.

Contact: 1.183
Gap: 1.16
Power: 1.135
Eye: 0.989
Avoid K: 1.13
Defense: 1.05
Speed: 1.544
Stuff: 1.107
Movement: 1.188
Control: 1.145
Stamina: 1.093

These work pretty well but by year 50, the draft quality drops again. I'm running some more tests to see where the PCM need to be past year 10 in order to hold the draft quality more steady longer into the future.

torpidbeaver · 06-01-2018, 06:46 PM

Thanks for your work on this, man. I've been following with a lot of interest.

Just to be clear, these draft class issues are happening regardless of whether you're playing in a Feeder League, Quickstart, or Fictional crazy-time 50 team-sub-leagues, correct?

stealofhome · 06-01-2018, 07:45 PM

From what I've seen, yes. I haven't tried any crazy fictional setups so I don't know to what extent, though in all settings. It does seem more pronounced in feeder leagues so the modifiers have to be tweaked a little higher down the road.

The other unknown is if it is on purpose. Does the game calculate the current major league talent level, aging curve, etc. and create players based on that? Not sure.

stealofhome · 06-02-2018, 03:13 PM

The modifiers do appear to need changing every 10 years for maybe the first 50 years. These are the PCM I landed on for the year 15 change:

Contact: 1.251
Gap: 1.275
Power: 1.299
Eye: 0.964
Avoid K: 1.247
Stuff: 1.202
Movement: 1.399
Control: 1.252
Stamina: 1.098
Speed: 1.736
Defense: 1.071

Interestingly enough, the modifiers themselves (with the exception of batter eye) are changing at a logarithmic rate (just like everything else in this world). If they continue to follow this pattern, there shouldn't be much change necessary past year 50, maybe just every 25 years instead of every 10.

smythejd · 06-02-2018, 05:28 PM

I have an existing league and I want to add feeders. Should I use your modifiers immediately (as it is a league with a lot of history) or wait 5 years to change the modifiers?

darkcloud4579 · 06-02-2018, 05:31 PM

Astounding work, thanks for sharing your findings!

stealofhome · 06-02-2018, 05:32 PM

Quote:

Originally Posted by smythejd

I have an existing league and I want to add feeders. Should I use your modifiers immediately (as it is a league with a lot of history) or wait 5 years to change the modifiers?

Hmm, not really sure and I honestly don't know how well these modifiers will transfer to other leagues. You could try making a backup of the league and then adding feeders and look at the first draft. I'd be curious as well to see if the draft quality resets to the first year quality or if it would default to a lower quality level.

smythejd · 06-02-2018, 07:43 PM

I will give it a shot. After the last draft class (7 players over a 3½ star rating, 40 team league), I really have nothing the lose.

stealofhome · 06-03-2018, 11:48 PM

PCM year 25 to 45:

Contact: 1.325
Gap: 1.375
Power: 1.31
Eye: 0.97
Avoid K: 1.278
Stuff: 1.289
Movement: 1.448
Control: 1.283
Stamina: 1.072
Speed: 1.529
Defense: 1.043

I went 20 years this time and the ratios are much smaller than before so the talent is starting to level out. The average change in PCM ratio in year 5 was 1.16, in year 15 it was 1.08 then in year 25 it was 1.01. Year 35 was 1 but year 45 will be higher since the talent did drift enough to make an impact. I haven't played through a league with this but it does seem to be a little too talented...might be good with extreme injury settings and high TCR. Lots of 45/55 players in the free agent pool in the middle of the season, etc.

mbyrd75 · 06-05-2018, 12:58 AM

awesome work stealofhome.

I'm working on creating an online stats only league with a full feeder setup. Are you leaving your Saber creation modifiers at default when you setup the league initially?

stealofhome · 06-05-2018, 01:27 AM

Quote:

Originally Posted by mbyrd75

awesome work stealofhome.

I'm working on creating an online stats only league with a full feeder setup. Are you leaving your Saber creation modifiers at default when you setup the league initially?

That's right, I've left them all the same and only change the mlb modifiers down the road. I force all feeder players to be created at the minimum age so the creation modifiers have a delayed impact which is why I begin changing them at year 5. I'm running another league with a little different setup to see how well the modifiers I've posted hold up.

stealofhome · 06-08-2018, 03:41 PM

I finally ran through the league - 70 years with feeder leagues and modifiers - and there is definitely an increase in the quality of MLB players. The number of 35+ potential players in the free agent pool at year 70 without any feeder leagues or modifiers is 17. With the feeders and modifiers, that number is at 190.

Also, the average potential for all players in the MLB at year 70 without feeders and modifiers is 44. With them it is 54. It looks like the modifiers may need to be tweaked down a tiny bit - back to the drawing board I suppose.

mbyrd75 · 06-08-2018, 05:56 PM

Quote:

Originally Posted by stealofhome

I finally ran through the league - 70 years with feeder leagues and modifiers - and there is definitely an increase in the quality of MLB players. The number of 35+ potential players in the free agent pool at year 70 without any feeder leagues or modifiers is 17. With the feeders and modifiers, that number is at 190.

Also, the average potential for all players in the MLB at year 70 without feeders and modifiers is 44. With them it is 54. It looks like the modifiers may need to be tweaked down a tiny bit - back to the drawing board I suppose.

Very positive results. Rather have too much good and dial it back vs not being able to find a way to fix it.

Are you seeing any change at all in the distribution of players able to pay the more skilled positions?

stealofhome · 06-08-2018, 06:12 PM

Quote:

Originally Posted by mbyrd75

Are you seeing any change at all in the distribution of players able to pay the more skilled positions?

Definitely, but I also turned off two-way players at the same time so I don't know what kind of an impact that had. I was previously getting about 17% of the top 300 players at C/SS/CF with 20% DH. Now I'm getting closer to 20-25% C/SS/CF with only 10% DH. Much of those DH players are now showing up as pitchers in the top players. My guess is that the pitching ratings were so low that too many hitters were getting into the top players.

05-21-2018, 11:04 AM	#44
stealofhome Hall Of Famer Join Date: Apr 2014 Posts: 2,282	I've been playing more with the modifiers and it starts getting pretty tricky to make this work. Instead of comparing to the MLB Quickstart, I compare to the first year of a custom league with no feeders. Let the game create the draft pool because that is "working as intended." As has been seen repeatedly, the first year draft of any league has very high quality and then that decreases rapidly over the first 10 years. This happens in the quickstart, standard leagues, and fictional leagues. One thing I've been ignoring until recently is the batting potential stats and this is where I've noticed diminishing quality. 10 years down the line, hitters have very low contact, gap, and power potential ratings. Pitchers have low movement and control. However, you can't set these modifiers in year 1 because that will cause you to create incredible players. You have to wait until year 7-10ish with no feeder leagues or year 5ish with feeder leagues. With my feeder league setup as before, I have been getting even better results now with creating the league and letting all modifiers be 1. Sim ahead 5 years and then change the modifiers. Contact: 1.148 Gap: 1.101 Power: 1.153 Eye: .999 K: 1.023 Stuff: 0.953 Movement: 1.189 Control: 1.099 Stamina: 0.963 Speed: 1.201 Since the players are created at the league minimum age in high school/college and are forced to have 4 years, this means these players don't show up for another 4 years down the road. My year 10 draft doing this was much higher quality than ever before. The outstanding issue is still the fielding ratings. I have an algorithm that takes the fielding ratings and determines where the player will fit best. The DH% (hitters who are not good defensively anywhere) holds steady around 8-10% for the league with no feeders. When the feeder leagues are turned on though, that number basically doubles to 16-20%. The problem with that though is that the fielding ratings are not all consistently too low, there are only a few that are not up to par, namely the range ratings. In the 10th year of a feeder league, the average IF RNG is 90% of where it is in the non-feeder draft and the average OF RNG is a whopping 82%. Very few fielders can cover any ground and they end up not having good positional ratings and are consequently less valuable. The issue with this is that if you crank up the fielding rating modifier, it increases all modifiers when the range stats are really the only ones that need to be modified. I still need to look at some of the other modifiers to see if it is possible to increase range without touching anything else. I also need to see if the ratio of hitters to pitchers is throwing off these fielding numbers. Last edited by stealofhome; 05-21-2018 at 02:09 PM.

06-01-2018, 09:40 AM	#46
stealofhome Hall Of Famer Join Date: Apr 2014 Posts: 2,282	I've been running some more tests since the patch and feeder leagues are definitely behaving differently! I haven't run a long term simulation yet to see what the league turns into over many decades, but this is a step in the right direction. I wonder if batting eye is treated differently in the way players are created. When I adjust the modifiers, almost all of them need to be somewhere around 1.1 or 1.2 but batting eye stays pretty much the same with no needed modifications. Another thing that I have done is turned off two-way players which seems to help when changing modifiers as well. Last edited by stealofhome; 06-01-2018 at 09:56 AM.

06-02-2018, 05:31 PM	#52
darkcloud4579 Hall Of Famer Join Date: Jun 2003 Posts: 8,712	Astounding work, thanks for sharing your findings! __________________ ------ My Mods OOTP Advanced Stats & DFS Suite Managerial Strategy Pack Competitive Balance Tax Calculator Major League Women's Baseball (OOTP24) quickstart Indian Premier League \| 300+ years of baseball quickstart \| Expatriate League quickstart \| Off-Field Injuries Update \| Women's Name File for OOTP \| ---- Dynasty classics: Centurion comes to OOTP5 \| DC Moneyball Dynasty (2004)

06-03-2018, 11:48 PM	#55
stealofhome Hall Of Famer Join Date: Apr 2014 Posts: 2,282	PCM year 25 to 45: Contact: 1.325 Gap: 1.375 Power: 1.31 Eye: 0.97 Avoid K: 1.278 Stuff: 1.289 Movement: 1.448 Control: 1.283 Stamina: 1.072 Speed: 1.529 Defense: 1.043 I went 20 years this time and the ratios are much smaller than before so the talent is starting to level out. The average change in PCM ratio in year 5 was 1.16, in year 15 it was 1.08 then in year 25 it was 1.01. Year 35 was 1 but year 45 will be higher since the talent did drift enough to make an impact. I haven't played through a league with this but it does seem to be a little too talented...might be good with extreme injury settings and high TCR. Lots of 45/55 players in the free agent pool in the middle of the season, etc. Last edited by stealofhome; 06-03-2018 at 11:52 PM.

05-16-2018, 08:07 AM	#41
Mr. Marlin All Star Starter Join Date: Apr 2010 Posts: 1,278	I'm curious, does performance follow ratings or vice versa. The reason I ask is I've been following stealofhome on this and have notices some differences in my own observations of feeder players. But it occurs to me the biggest difference is he is looking at ratings. I never see ratings because of the way I play (stats only) I have them turned off. So, out of curiosity I turned them back on and looked at ta sample of changes in MLB level players that were 5+ rd picks. A number of them were performing at a pretty high level stats wise in the feeders and minors (above league average in multiple categories). At the time their ratings actual and potential were average to below average. So, does performance change ratings or does ratings change performance?

05-16-2018, 10:40 AM	#43
Mr. Marlin All Star Starter Join Date: Apr 2010 Posts: 1,278	Like Brad Pitt said in the Moneyball movie "If he's such a good hitter why doesn't he hit good?" I have enjoyed stats only for a lot of reasons. Not the least of which is the "fog of war" you get by not seeing ratings. But the ratings themselves are subjective unless you have scouting set to 100% accurate. I'd never play that way though. Really it's the AI scout's opinion of a player modified by what that scout's strengths and weaknesses are. It's been debated a lot around here whether stats only is harder or more of a challenge. I'd have to say it is not BECAUSE the ratings are subjective. If anything they may blind you to a player's quality or lead you to buy a white elephant. Of course the better the scout the more reliable the ratings. On my fictional Miami Tropics team I had a player with an eye rating of 5 (on 1-10 scale) but his slash line is .241/.420/.305. My league average OBP was .330. The following season IIRC his eye rating was 7. The nice thing about stats is it focuses on what the player is doing right now. Plus, when a GM looks at a player report he does not see colored bars. He sees numbers.

05-25-2018, 04:06 PM	#45
stealofhome Hall Of Famer Join Date: Apr 2014 Posts: 2,282	I have absolutely convinced myself that the game does not create players correctly, particularly with the ratio of fielding ratings. I looked at the top career players by WAR in a league I simmed forward 50 years and compared it to real life MLB HOF. ~1.4% of players get in the hall of fame, so I found the top 1.4% in WAR for my league. The reason I'm comparing to WAR is because the HOF that the AI created for the league is even worse (absolutely no 2B/SS). In real life, the HOF is split 69/31 Hitter/Pitcher and quite evenly by position - 7-10% at all position players. The top players in my OOTP league are also split right at 65/35 Hitter/Pitcher - pretty good. But by position is absolutely horrendous. I looked at the majority of innings played by position for the career which was a huge majority for almost all players. C: 2% 1B: 17% 2B: 5% 3B: 9% SS: 4% LF: 14% CF: 5% RF: 10% The IF/OF mix looks right on, but not enough players can play the skilled positions as I have been repeating now over and over and over. There are too many 1B/LF/DH types because the players are not created properly.

06-01-2018, 06:01 PM	#47
stealofhome Hall Of Famer Join Date: Apr 2014 Posts: 2,282	I finished the 50 year sim and the results are extremely promising. The hall of fame is split well between multiple positions and the drafts are much stronger. (Except there seems to be a bug where no pitchers were put in the hall of fame) I used these PCM on the day of the MLB draft in year 5 but before it happens. Contact: 1.183 Gap: 1.16 Power: 1.135 Eye: 0.989 Avoid K: 1.13 Defense: 1.05 Speed: 1.544 Stuff: 1.107 Movement: 1.188 Control: 1.145 Stamina: 1.093 These work pretty well but by year 50, the draft quality drops again. I'm running some more tests to see where the PCM need to be past year 10 in order to hold the draft quality more steady longer into the future.

06-01-2018, 06:46 PM	#48
torpidbeaver Major Leagues Join Date: Aug 2011 Posts: 391	Thanks for your work on this, man. I've been following with a lot of interest. Just to be clear, these draft class issues are happening regardless of whether you're playing in a Feeder League, Quickstart, or Fictional crazy-time 50 team-sub-leagues, correct?

06-01-2018, 07:45 PM	#49
stealofhome Hall Of Famer Join Date: Apr 2014 Posts: 2,282	From what I've seen, yes. I haven't tried any crazy fictional setups so I don't know to what extent, though in all settings. It does seem more pronounced in feeder leagues so the modifiers have to be tweaked a little higher down the road. The other unknown is if it is on purpose. Does the game calculate the current major league talent level, aging curve, etc. and create players based on that? Not sure.

06-02-2018, 03:13 PM	#50
stealofhome Hall Of Famer Join Date: Apr 2014 Posts: 2,282	The modifiers do appear to need changing every 10 years for maybe the first 50 years. These are the PCM I landed on for the year 15 change: Contact: 1.251 Gap: 1.275 Power: 1.299 Eye: 0.964 Avoid K: 1.247 Stuff: 1.202 Movement: 1.399 Control: 1.252 Stamina: 1.098 Speed: 1.736 Defense: 1.071 Interestingly enough, the modifiers themselves (with the exception of batter eye) are changing at a logarithmic rate (just like everything else in this world). If they continue to follow this pattern, there shouldn't be much change necessary past year 50, maybe just every 25 years instead of every 10.

06-02-2018, 05:28 PM	#51
smythejd All Star Starter Join Date: Jun 2008 Posts: 1,025	I have an existing league and I want to add feeders. Should I use your modifiers immediately (as it is a league with a lot of history) or wait 5 years to change the modifiers?

06-02-2018, 07:43 PM	#54
smythejd All Star Starter Join Date: Jun 2008 Posts: 1,025	I will give it a shot. After the last draft class (7 players over a 3½ star rating, 40 team league), I really have nothing the lose.

06-05-2018, 12:58 AM	#56
mbyrd75 Minors (Rookie Ball) Join Date: Feb 2012 Posts: 40	awesome work stealofhome. I'm working on creating an online stats only league with a full feeder setup. Are you leaving your Saber creation modifiers at default when you setup the league initially?

06-08-2018, 03:41 PM	#58
stealofhome Hall Of Famer Join Date: Apr 2014 Posts: 2,282	I finally ran through the league - 70 years with feeder leagues and modifiers - and there is definitely an increase in the quality of MLB players. The number of 35+ potential players in the free agent pool at year 70 without any feeder leagues or modifiers is 17. With the feeders and modifiers, that number is at 190. Also, the average potential for all players in the MLB at year 70 without feeders and modifiers is 44. With them it is 54. It looks like the modifiers may need to be tweaked down a tiny bit - back to the drawing board I suppose.