Markus, why would the AI release this player?

[CBLCardinals]

Hello Markus,

My question is why on earth would the AI release this calibre of player?

In real life he is one of the top prospects in the game. In OOTP, he is dominating AA ball, was selected to the AA All-Star game, AND was the AA pitcher of the month in June. On July 11th, the Dodgers RELEASED him.

How many times in real life would a team give up on a dominate 20 y/o LHP ?????

[kckid4u]

He looks gay?

(Not that theres anything wrong with that.)

[bababui]

The real answer is that he is likely to take a big ratings hit soon; but for the sake of a dynasty, he could have been arrested for drugs or been involved in some other off the field nonsense.

[Markus Heinsohn]

Originally Posted by CBLCardinals

Hello Markus,

My question is why on earth would the AI release this calibre of player?

In real life he is one of the top prospects in the game. In OOTP, he is dominating AA ball, was selected to the AA All-Star game, AND was the AA pitcher of the month in June. On July 11th, the Dodgers RELEASED him.

How many times in real life would a team give up on a dominate 20 y/o LHP ?????

I have no clue, and can't tell by not knowing anything about the league at all...

[Bonedwarf]

Financial problems... Crap scouts... Drunk GM etc...

[Markus Heinsohn]

Probably drunk GM

[Faaip De Oiad]

Terrible scouts.......he 6'4" and about 220.

They cant even get that right.....

[Andy]

Originally Posted by Faaip De Oiad

Terrible scouts.......he 6'4" and about 220.

They cant even get that right.....

Blame CBL for that one!

Well, I'm assuming he is using his own rosters...

[CHOWDERHEAD]

Joking aside, this is the type of issue that has always concerned me. While I know that there are real life examples that could be pointed to, I sill hope that this is looked at, and hopefully further refined/tweaked, etc...

[zekester91]

I think the bug fix that was anounced over at FOF will help solve some of this problem, if you have your game set up so that it looks at stats from this year at like 50% and ratings at a much lower number, and it works like Markus said in that posts then the AI won't be making these moves when a player takes a ratings drop that the human gm/commish can't see.

[MizzouRah]

Originally Posted by zekester91

I think the bug fix that was anounced over at FOF will help solve some of this problem, if you have your game set up so that it looks at stats from this year at like 50% and ratings at a much lower number, and it works like Markus said in that posts then the AI won't be making these moves when a player takes a ratings drop that the human gm/commish can't see.

This what Markus had posted:

It now works as advertised, the bug screwed the whole system basically... it works like in this example now:

Player A:
Expected OPS by ratings: .900
Current Year OPS: 1.100 in 150 PA
Last Year OPS: 0.800 in 550 PA
2 years ago: 0.900 in 600 PA

Now, the AI settings are like this:
Ratings 30%
Current Year 40%
Last Year 20%
2 years ago: 10%

OK, now, the resulting OPS is:
0.3 * 0.900 (30% of the ratings)
+ 0.12 * 1.100 (150 PA from the 500 minimum is 30%, so it uses 12% (40% * 0.3) of current stats)
+ 0.48 * 0.800 (the remaining 28% from the previous step, plus 20% last year)
+ 0.1 * 0.900
= 0.876 ... so, for the AI this guy is a hitter with an 0.876 OPS.

Hope that makes sense This allows you to have the AI act entirely based on stats if there are enough PA/IP for the player in the past...

[zekester91]

Originally Posted by MizzouRah

This what Markus had posted:

Yes, which I take to mean that the AI (if your setting are close to what he has listed) will see this player as an 0.876 OPS, instead of a player who just took an small ratings bump, since it won't be looking at just ratings, since there was a bug in the old way it worked.

[MizzouRah]

Originally Posted by zekester91

Yes, which I take to mean that the AI (if your setting are close to what he has listed) will see this player as an 0.876 OPS, instead of a player who just took an small ratings bump, since it won't be looking at just ratings, since there was a bug in the old way it worked.

I'm hoping this is the case as well. Will be interesting to see the results once the game is released to the masses.

[NCBeachBum]

Originally Posted by CBLCardinals

How many times in real life would a team give up on a dominate 20 y/o LHP ?????

Perhaps the Dodgers prefer submissives?

[The Wolf]

What happens with the new AI if you set it to be stats-considering-only? I would think it would dump guys with good ratings who had crummy luck and therefore bad stats. Is that the case?

[kq76]

Originally Posted by Markus Heinsohn

Probably drunk GM

LOL. I read "Probably drunk programmer" and had to do a double take.

[NomarHits400]

Originally Posted by MizzouRah

I'm hoping this is the case as well. Will be interesting to see the results once the game is released to the masses.

I think Markus' fixing of the player evaluation will be irrelevant in this case. Unless I am mistaken- there isn't any way for the evaluation (shown here)to consider potential. So,in fact, best case scenario- any evaluation will only take into account a player's current ratings and/or stats.....I don't know of any way to put a weight on the players future potential...So if a guy is currently 20/30/40 with potentials of 80/80/80- if he has a bad season- how can the AI evaluation factor in his future if the formula is based on stats and (current) ratings (of any combination)?....

I have seen many guys like Kershaw dumped- and it will continue to happen unless the AI's balance of when to give up on a player like this is better managed. In the above example- if Kershaw showed the same ratings as a 26 year old- maybe the release could be justified- but never at 20.

[jbmagic]

ootp's AI always had problems with roster management for releases and waivers.


I hope Markus tweaks the hell out of this finally.

[SittingDuck]

Originally Posted by NomarHits400

I think Markus' fixing of the player evaluation will be irrelevant in this case. Unless I am mistaken- there isn't any way for the evaluation (shown here)to consider potential. So,in fact, best case scenario- any evaluation will only take into account a player's current ratings and/or stats.....I don't know of any way to put a weight on the players future potential...So if a guy is currently 20/30/40 with potentials of 80/80/80- if he has a bad season- how can the AI evaluation factor in his future if the formula is based on stats and (current) ratings (of any combination)?....

I have seen many guys like Kershaw dumped- and it will continue to happen unless the AI's balance of when to give up on a player like this is better managed. In the above example- if Kershaw showed the same ratings as a 26 year old- maybe the release could be justified- but never at 20.

AFAIK, right now it does. Don't hold me to that. I haven't seen any golden boys in the waiver pile, and I'm running a maxed scout.

I have my AI Evaluation settings at:

Ratings: 20
This Year's Stats: 45
Last Year's Stats: 25
Two Season's Ago: 10

I believe they are working fine in the AI roster evaluation.

Perhaps others in the transactions dept will verify if they see this post. Either way, you'll find out in a little over a week.

[CBLCardinals]

Originally Posted by Andy

Blame CBL for that one!

Well, I'm assuming he is using his own rosters...

heh INDEED!

For some of the prospects I entered into the Lahman DB I just used a default of 5'10", 190lbs. I caught most of them after they were entered into the game, just not all of them (as with the case for Kershaw).