ai stats weight

[PSUColonel]

I am partial to either 55/25/15/5 or 50/30/15/5


Not much difference between the two obviously. Another I may consider trying:

45/30/20/5 or 45/35/15/5


I like the idea of keeping the ratings weight with in 5 +/- of 50


I have always like the way 40/30/20/10 works for MLB players...but that's not the best formula for minor leaguers I don't think. that's why I feel you need something in between. To me, 55/25/15/5 is likely the best of both. It's close to the default, but it definitely takes stats a bit more into consideration.

[SirMichaelJordan]

Originally Posted by PSUColonel

I am partial to either 55/25/15/5 or 50/30/15/5


Not much difference between the two obviously. Another I may consider trying:

45/30/20/5 or 45/35/15/5


I like the idea of keeping the ratings weight with in 5 +/- of 50


I have always like the way 40/30/20/10 works for MLB players...but that's not the best formula for minor leaguers I don't think. that's why I feel you need something in between. To me, 55/25/15/5 is likely the best of both. It's close to the default, but it definitely takes stats a bit more into consideration.

Fictional default is way different than the MLB default. It’s all a matter of taste.


Sent from my iPhone using Tapatalk

[PSUColonel]

Originally Posted by SirMichaelJordan

Fictional default is way different than the MLB default. It’s all a matter of taste.


Sent from my iPhone using Tapatalk

true...but I am talking more about the way the AI handles minor league players.

[SirMichaelJordan]

Originally Posted by PSUColonel

true...but I am talking more about the way the AI handles minor league players.

I haven’t had any problems with minor leagues using a stats only evaluation. I don’t use that anymore but It’s all about what you want to achieve.


Sent from my iPhone using Tapatalk

[Sweed]

Originally Posted by SirMichaelJordan

25/25/25/25

Or

0/50/25/25

Originally Posted by SirMichaelJordan

Better AI decisions, players OVR are rated better based on their performance.

I’m now using 0/50/25/25 as it get me better results because things have changed in the last few years like GM’s having their own evaluation and defense getting more love in the rating formula.

I keep it even because the %’s are prorated. Ratings will always be used despite having 0% for ratings especially if their is t much stats to go off of.

Originally Posted by SirMichaelJordan

Both. But I prefer to not have anything for ratings as of now. It gives me better results than 25 across the board with the new updates to AI evaluation.


Sent from my iPhone using Tapatalk

Originally Posted by SirMichaelJordan

I’m also seeing better roster moves by the AI which includes Contracts, trades, promotion/demotion and waivers.

I never see any players that put up great stats in the current year get undervalued but at the same time not be over valued because of their stats from the past few seasons will normalize them. Ratings still play a role but it’s not over powering in decision making for the AI. A lot of roster moves just make sense.


Sent from my iPhone using Tapatalk

I continue to come back to this thread and take in SMJ's thoughts on evaluation. I normally stay away from extreme settings (zeros in any direction) figuring Markus' and Matt's defaults were set for a reason. IE they know the game better than anybody

However I am still intrigued by SMJ's thoughts on the subject and am looking forward to giving these a try. Currently in my playoffs so a week or two away from making the change. Right now I am leaning towards the 0/50/25/25.

I guess the advantage(?) I have is I play out every game so any changes will come slowly over the ~4 months it takes me to complete a season. I intend to stick with it for a season and see what the results are.

So when to make the change and I assume a league wide re-scouting should be done at the same time? My thoughts are the day after the world series is complete? Let the new system go into effect before any of the off-season events are started.

Thoughts?

[SirMichaelJordan]

Originally Posted by Sweed

I continue to come back to this thread and take in SMJ's thoughts on evaluation. I normally stay away from extreme settings (zeros in any direction) figuring Markus' and Matt's defaults were set for a reason. IE they know the game better than anybody

However I am still intrigued by SMJ's thoughts on the subject and am looking forward to giving these a try. Currently in my playoffs so a week or two away from making the change. Right now I am leaning towards the 0/50/25/25.

I guess the advantage(?) I have is I play out every game so any changes will come slowly over the ~4 months it takes me to complete a season. I intend to stick with it for a season and see what the results are.

So when to make the change and I assume a league wide re-scouting should be done at the same time? My thoughts are the day after the world series is complete? Let the new system go into effect before any of the off-season events are started.

Thoughts?

I usually re-scout and hit the re adjust manager evaluation button.

I’m back at using 25/25/25/25 along with stats used in scouting checked in settings. It’s been very balanced so far.


Sent from my iPhone using Tapatalk

[AJFO]

Originally Posted by SirMichaelJordan

I usually re-scout and hit the re adjust manager evaluation button.

I’m back at using 25/25/25/25 along with stats used in scouting checked in settings. It’s been very balanced so far.


Sent from my iPhone using Tapatalk

Originally Posted by Sweed

I continue to come back to this thread and take in SMJ's thoughts on evaluation. I normally stay away from extreme settings (zeros in any direction) figuring Markus' and Matt's defaults were set for a reason. IE they know the game better than anybody

However I am still intrigued by SMJ's thoughts on the subject and am looking forward to giving these a try. Currently in my playoffs so a week or two away from making the change. Right now I am leaning towards the 0/50/25/25.

I guess the advantage(?) I have is I play out every game so any changes will come slowly over the ~4 months it takes me to complete a season. I intend to stick with it for a season and see what the results are.

So when to make the change and I assume a league wide re-scouting should be done at the same time? My thoughts are the day after the world series is complete? Let the new system go into effect before any of the off-season events are started.

Thoughts?

Been thinking quite a bit about this as well. A good barometer as to what is might be realistic AI player eval settings versus not - all in a MLB quickstart - might be checking and seeing how tradeable Chris Davis' contract is using some the different configurations you mention above.

For instance, I started a new save and tried "Hard", "Favor Prospects" on 25/40/27/8 with TCR at 135.

Simmed the season and come October 1st tried to trade Davis on his own, which only worked if the Orioles retained 95% of his contract (roughly ~3-4 mil. total in savings).

Davis' 2020 season had been worth -0.7 WAR over 115 PA, while batting to a .160/261/280 line.
Tried to attach a top prospect in Grayson Rodriguez for a lesser pricetag on money being retained on my end (acting as the O's GM) and voila....

.....a bunch of other 'not so great' contracts came up (see Gonzalez, Gio; Miley, Wade; Pomeranz, Drew), all of whom had accrued negative WAR during the 2020 season.

The exception was Arizona offering Jake Lamb (on an arbitration eligible contract; -0.2 WAR in 2020) and Kole Calhoun (lengthier contract, but 1.4 WAR in 2020).

Seems to work reasonably well, but I'm curious as to how a more skewed configuration might change this picture.

[SirMichaelJordan]

Originally Posted by AJFO

Been thinking quite a bit about this as well. A good barometer as to what is might be realistic AI player eval settings versus not - all in a MLB quickstart - might be checking and seeing how tradeable Chris Davis' contract is using some the different configurations you mention above.

For instance, I started a new save and tried "Hard", "Favor Prospects" on 25/40/27/8 with TCR at 135.

Simmed the season and come October 1st tried to trade Davis on his own, which only worked if the Orioles retained 95% of his contract (roughly ~3-4 mil. total in savings).

Davis' 2020 season had been worth -0.7 WAR over 115 PA, while batting to a .160/261/280 line.
Tried to attach a top prospect in Grayson Rodriguez for a lesser pricetag on money being retained on my end (acting as the O's GM) and voila....

.....a bunch of other 'not so great' contracts came up (see Gonzalez, Gio; Miley, Wade; Pomeranz, Drew), all of whom had accrued negative WAR during the 2020 season.

The exception was Arizona offering Jake Lamb (on an arbitration eligible contract; -0.2 WAR in 2020) and Kole Calhoun (lengthier contract, but 1.4 WAR in 2020).

Seems to work reasonably well, but I'm curious as to how a more skewed configuration might change this picture.

Yea I did similar test throughout the years. Trading, contracts, waivers, FA, promotion/demotion and OVR relation to other players and how they are performing.

[Poonox24]

Originally Posted by AJFO

Been thinking quite a bit about this as well. A good barometer as to what is might be realistic AI player eval settings versus not - all in a MLB quickstart - might be checking and seeing how tradeable Chris Davis' contract is using some the different configurations you mention above.

For instance, I started a new save and tried "Hard", "Favor Prospects" on 25/40/27/8 with TCR at 135.

Simmed the season and come October 1st tried to trade Davis on his own, which only worked if the Orioles retained 95% of his contract (roughly ~3-4 mil. total in savings).

Davis' 2020 season had been worth -0.7 WAR over 115 PA, while batting to a .160/261/280 line.
Tried to attach a top prospect in Grayson Rodriguez for a lesser pricetag on money being retained on my end (acting as the O's GM) and voila....

.....a bunch of other 'not so great' contracts came up (see Gonzalez, Gio; Miley, Wade; Pomeranz, Drew), all of whom had accrued negative WAR during the 2020 season.

The exception was Arizona offering Jake Lamb (on an arbitration eligible contract; -0.2 WAR in 2020) and Kole Calhoun (lengthier contract, but 1.4 WAR in 2020).

Seems to work reasonably well, but I'm curious as to how a more skewed configuration might change this picture.

So what's your recommendation after testing it? I am about 20 seasons into my league, so now it's all fictional. Wondering what gets the best results.

[Syd Thrift]

I've done 0/67/22/11 in the past but have been cool with the defaults so far...

[AJFO]

Originally Posted by Poonox24

So what's your recommendation after testing it? I am about 20 seasons into my league, so now it's all fictional. Wondering what gets the best results.

I'll give you my honest take and it likely isn't going to be the answer you're looking for (let alone much of an answer):

I've yet to sim a fictional league, so by default I'd defer to someone else's judgement. That said, so far so good in my sim with a rough 20/45/25/10 split.

That said, if it's fictional, perhaps you might be better served following some of the recommendations to try to a lesser weighting for "ratings" and more emphasis on pure stats. Just my two cents, though someone else can likely give you an entire dollar.

[PSUColonel]

Ratings are still ratings no matter whether its fictional or not. I view ratings as “perception”... which is obviously a big deal in the real world up to a point.

Perhaps 40/30/20/10 really is the best of both worlds. It’s what we all thought here years and years ago.

[jimmysthebestcop]

The problem with ratings especially on the 20-80 scale and stars is the majority of players are 2.5-3 stars. That is the bulk of the MLB. And all of those players have almost the same individual ratings with most being +/-5 and some with +/-10. I'm assuming the AI is using those ratings and not the true behind the scenes ratings.

With the majority around the same stars/ratings you can have a 2.5 star guy who is strictly a bench warmer and then another one who is an all star. So stats should be weighted higher just like it is in the fictional defaults.

Because unless something unexplained happened most of the time those dozen or so 4.5/5 star players will always perform better then the 2.5/3 mass. That is a huge difference in talent. That is why there are so few 4/5/5 star players.

I don't know what the best is or if it can ever be answered but the default MLB or default challenge mode seems to completely off base. The fictional default is much better.

Then people can argue about fine tuning it. But that should be the basis I think. And I think a lot has to do with aging. IRL the mlb is as young as its been in 50-70 years. 32+ is now ancient. And in OOTP it's exactly equal to MLB aging but players do age in OOTP especially over 32 while if they were known all stars that still might be 4+ stars but on the decline. Ratings haven't caught up to them but their stats suffer. And they still ask for crazy 7+ year deals at 15+ million.

And this aging problem influence the problem with Ratings getting heaviest weight. Those contracts kill the AI. They can't handle it.

I really think the devs need to do something about 30+ year olds and the contracts they are asking for. Even if it gets to unrealistic territory just to give the AI a better handle on finances. It could even be some kind of check box "Advanced aging player contracts". That way people could run their universe the way they want.

[MizzouRah]

I'm playing fictional this year and although I've thought about deviating from defaults.. there is really nothing in 5 seasons that has me wanting to do so.

30/50/15/5 seems to working great in my fictional baseball world

[PSUColonel]

Originally Posted by jimmysthebestcop

The problem with ratings especially on the 20-80 scale and stars is the majority of players are 2.5-3 stars. That is the bulk of the MLB. And all of those players have almost the same individual ratings with most being +/-5 and some with +/-10. I'm assuming the AI is using those ratings and not the true behind the scenes ratings.

With the majority around the same stars/ratings you can have a 2.5 star guy who is strictly a bench warmer and then another one who is an all star. So stats should be weighted higher just like it is in the fictional defaults.

Because unless something unexplained happened most of the time those dozen or so 4.5/5 star players will always perform better then the 2.5/3 mass. That is a huge difference in talent. That is why there are so few 4/5/5 star players.

I don't know what the best is or if it can ever be answered but the default MLB or default challenge mode seems to completely off base. The fictional default is much better.

Then people can argue about fine tuning it. But that should be the basis I think. And I think a lot has to do with aging. IRL the mlb is as young as its been in 50-70 years. 32+ is now ancient. And in OOTP it's exactly equal to MLB aging but players do age in OOTP especially over 32 while if they were known all stars that still might be 4+ stars but on the decline. Ratings haven't caught up to them but their stats suffer. And they still ask for crazy 7+ year deals at 15+ million.

And this aging problem influence the problem with Ratings getting heaviest weight. Those contracts kill the AI. They can't handle it.

I really think the devs need to do something about 30+ year olds and the contracts they are asking for. Even if it gets to unrealistic territory just to give the AI a better handle on finances. It could even be some kind of check box "Advanced aging player contracts". That way people could run their universe the way they want.

i believe this is the fog of war we want.

[endgame]

Originally Posted by jimmysthebestcop

I think I might try out XX/XX/XX/XX cause hey why not

At its core, the essence of things, I think.

[Mat]

Originally Posted by MizzouRah

30/50/15/5 seems to working great in my fictional baseball world

Me too!

[itsmb8]

Originally Posted by jimmysthebestcop

The problem with ratings especially on the 20-80 scale and stars is the majority of players are 2.5-3 stars. That is the bulk of the MLB. And all of those players have almost the same individual ratings with most being +/-5 and some with +/-10. I'm assuming the AI is using those ratings and not the true behind the scenes ratings.

With the majority around the same stars/ratings you can have a 2.5 star guy who is strictly a bench warmer and then another one who is an all star. So stats should be weighted higher just like it is in the fictional defaults.

Because unless something unexplained happened most of the time those dozen or so 4.5/5 star players will always perform better then the 2.5/3 mass. That is a huge difference in talent. That is why there are so few 4/5/5 star players.

I don't know what the best is or if it can ever be answered but the default MLB or default challenge mode seems to completely off base. The fictional default is much better.

Then people can argue about fine tuning it. But that should be the basis I think. And I think a lot has to do with aging. IRL the mlb is as young as its been in 50-70 years. 32+ is now ancient. And in OOTP it's exactly equal to MLB aging but players do age in OOTP especially over 32 while if they were known all stars that still might be 4+ stars but on the decline. Ratings haven't caught up to them but their stats suffer. And they still ask for crazy 7+ year deals at 15+ million.

And this aging problem influence the problem with Ratings getting heaviest weight. Those contracts kill the AI. They can't handle it.

I really think the devs need to do something about 30+ year olds and the contracts they are asking for. Even if it gets to unrealistic territory just to give the AI a better handle on finances. It could even be some kind of check box "Advanced aging player contracts". That way people could run their universe the way they want.

Totally agree. I feel like the demands are fine, but teams willing to give that 30+ year old a 5+ year, high salary contract should be turned WAY down. The vast majority of overpaid players aren't overpaid immediately after signing that contract, they're overpaid because the team gave them a huge extension when that player was in his prime. And that's okay, you sacrifice the huge risk at the end for keeping the player happy during his peak. But once that play hits 30-32, the kind of contracts teams are willing to give him goes down the tubes.

[jimmysthebestcop]

Originally Posted by PSUColonel

i believe this is the fog of war we want.

I think the only big problem is the MLB default settings. Its for like 2010-2015 MLB not present day. Too much reliance on ratings. 65 is crazy high.

I think it makes the AI much more eager to sign older free agents right away then to call up their prospects who will put up the same numbers in many cases.

And as we all know it's not just 1 setting but they all play off of each other. AI eval, aging, development, roster limits, service limits, other limits etc.

Its just the default ai eval settings I think is completely off the mark.

[AJFO]

Originally Posted by jimmysthebestcop

I think the only big problem is the MLB default settings. Its for like 2010-2015 MLB not present day. Too much reliance on ratings. 65 is crazy high.

I think it makes the AI much more eager to sign older free agents right away then to call up their prospects who will put up the same numbers in many cases.

And as we all know it's not just 1 setting but they all play off of each other. AI eval, aging, development, roster limits, service limits, other limits etc.

Its just the default ai eval settings I think is completely off the mark.

Appreciate the comprehensive rationale. I, too, am a bit miffed as to why the default AI eval settings are so high. I've yet to sim multiple games at length on '21, but was inspecting the trade log on my trial run and thought many (if not most) were quite reasonable.

In addition, the 35-40 OVR reliever carousel that has occurred for the past several IRL seasons at the MLB level also took place.

My experiences aside, I do agree that there is a fine line as to how player dev and aging speeds, TCR and player eval all tie together. Perhaps on some settings you simply have more major league-ready players that just don't get a chance to stick.

Which leads to me to wondering...if you expand your MLB set-up to 32 or 34 teams will the overall stat totals (or modifiers) also increase or stay steady and simply give everyone a smaller portion of the stat pie?

(Sorry for the tangent - was imagining a way to counteract surplus 35-40+ OVR if you set lower aging speeds to say .700.)