ai stats weight

[Sweed]

Originally Posted by MizzouRah

I used 40/30/20/10 for many versions and after seeing an odd waiver wire player, the fictional default of 30/50/15/5 seems too high with regard to ratings weight.

I think next season I'm going with 25/50/20/5 or 20/50/25/5

Originally Posted by 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

So you've changed your mind on the bold in the first quote?

Not a criticism, just asking I'm still a week or two away from making changes and am a bit nervous about the 0 for ratings in the 0/50/25/25. Seeing SMJ now back at all 25's had me thinking that would be my choice but...

I use Markus' defaults for most things, IE PCM's etc. so, maybe
the fictional default comes into play?

In any case I'll be watching the thread, making a change, just not exactly sure which way I'll end up going? Now leaning all 25's but that may change

[SirMichaelJordan]

Originally Posted by Sweed

So you've changed your mind on the bold in the first quote?

Not a criticism, just asking I'm still a week or two away from making changes and am a bit nervous about the 0 for ratings in the 0/50/25/25. Seeing SMJ now back at all 25's had me thinking that would be my choice but...

I use Markus' defaults for most things, IE PCM's etc. so, maybe
the fictional default comes into play?

In any case I'll be watching the thread, making a change, just not exactly sure which way I'll end up going? Now leaning all 25's but that may change

I only changed because there is a new feature “scouting incorporates stats” that can basically change how the scout (ratings) view a player. With this setting on, the scout now uses a AI evaluation of his own instead of going strictly off of pure ratings.

In theory...

My Baseline of 25/25/25/25 is now not only for Managers & GM’s to base their personal settings off of, it also applies to the scouting director.

So a GM who’s AI evaluation setting is 5 (scout) 5 (current year) 4 (last year) 3 (2 years ago) all based on my baseline of 25/25/25/25. Instead of the scout’s weight being strictly off of ratings, the scouts input is more of 5 (current year) 4 (last year) 3 (2 years ago)


Sent from my iPhone using Tapatalk

[MizzouRah]

Originally Posted by Sweed

So you've changed your mind on the bold in the first quote?

Not a criticism, just asking I'm still a week or two away from making changes and am a bit nervous about the 0 for ratings in the 0/50/25/25. Seeing SMJ now back at all 25's had me thinking that would be my choice but...

I use Markus' defaults for most things, IE PCM's etc. so, maybe
the fictional default comes into play?

In any case I'll be watching the thread, making a change, just not exactly sure which way I'll end up going? Now leaning all 25's but that may change

Yes sir. The odd waiver wire pickup actually turned out to be a SP on his decline and put up terrible numbers for the team that picked him up. He's since retired.. so I guess the AI had insight and moved on to a younger SP.


Since that, I've had a few more seasons to evaluate it more and for now, can conclude it's where it needs to be. I'm actually seeing some good AI to AI trades as well.

[MizzouRah]

Originally Posted by SirMichaelJordan

I only changed because there is a new feature “scouting incorporates stats” that can basically change how the scout (ratings) view a player. With this setting on, the scout now uses a AI evaluation of his own instead of going strictly off of pure ratings.

I turned this back OFF for now. Like Matt had said he saw some odd reports and I started to see some wonkyness myself with it set to yes.


Please let us know your thoughts as well.

[SirMichaelJordan]

Originally Posted by MizzouRah

I turned this back OFF for now. Like Matt had said he saw some odd reports and I started to see some wonkyness myself with it set to yes.


Please let us know your thoughts as well.

I haven’t notice any odd so far. I switch between 100% accuracy (for testing) and very low scouting (i play with very low).

What are some of the odd things that you witness so I can look out for it?


Sent from my iPhone using Tapatalk

[MizzouRah]

Originally Posted by SirMichaelJordan

I haven’t notice any odd so far. I switch between 100% accuracy (for testing) and very low scouting (i play with very low).

What are some of the odd things that you witness so I can look out for it?


Sent from my iPhone using Tapatalk

Seems like low rated players (pitchers mostly) with excellent stats don't seem to see scouting improvement for some reason.

[PSUColonel]

Originally Posted by MizzouRah

Seems like low rated players (pitchers mostly) with excellent stats don't seem to see scouting improvement for some reason.

This is something Matt should be aware of.

[SirMichaelJordan]

Originally Posted by MizzouRah

Seems like low rated players (pitchers mostly) with excellent stats don't seem to see scouting improvement for some reason.

I’ll pay more attention to that.

But I did notice even before I was using the setting that there were a lack of 5 star SP. majority of the 5 star pitchers were relievers.


Sent from my iPhone using Tapatalk

[thehef]

Great thread!

I do mostly historical, with minors, and I've been fooling around - without much luck - with recalc, player development, and ratings bases in order to address what, for me, are the following issues:

1) players reaching the majors too quickly (compared to real life)
2) established major-leaguers spending too many seasons in the minors (compared to real life), including a lot of guys finishing their careers in the minors
3) one-off appearances in the majors of real-life minor-leaguers as one-or-two-year major-league sim starters. Basically, minor-league nobodies taking roster spots from established major-leaguers.

Don't get me wrong, I want some variations from real life, and overall I like what I see. But I'm trying to find a sweet spot that reduces the occurances of the above.

I've stayed away from the Player Eval AI settings so far, because I'm not sure what affect they may or may not have on the issues described above.

I'm totally willing to test and test and test. Any suggestions on what I might try, though? Thanks in advance.

[endgame]

Originally Posted by endgame

Maleus Dei used to mix his evals up during specific phases of the season. FWIW I borrowed the strategy for a bit, but never took the time to actually measure the impact of the results.

With the luxury- and tedious work to do it -of having a clean slate, history-free, no service time league, I finally plan to implement stages as he did in the above reference.

I was compelled to do the inaugural using 100 ratings, 0 else. I haven't laid out the specifics yet, but thinking aloud I realize even at 0 any category can have 'some' undefined weight. I'll give it marginal. Probably less marginal than the difference between 45 and 40, let's say. Can't quantify, so ignore.

I don't have an All-Star game, but at that point in time, I'll probably adjust the settings to look more closely at this year's stats, gradually increasing that - what intervals? (shrug) -until the trading deadline where I am toying with the idea of pumping this year way up there with a new league.

All kinds of intervals turn up naturally in the game via League Events, or manager notes, or 'other' league events, so it's likely I'll time the changes with those turning points. What would be great- I've suggested it before -is to have a table to enter those and they would occur automatically, but could be altered later. As is, League Event coming up = re-examine the eval criteria. It just makes personal sense to me, right or wrong.

Oh, and to throw another tweak at the unpredictable process, including TCR 200 and few unnamed, this year, finally, I'll be taking a personalized version of Bobble's Random Events http://www.ootpdevelopments.com/boar...ml#post1073468 into the mix.

Good exchanges folks. I'll continue to take notes. SirM, the 25's are going to definitely make the cut. Until then...

[ThePride87]

Originally Posted by SirMichaelJordan

I only changed because there is a new feature “scouting incorporates stats” that can basically change how the scout (ratings) view a player. With this setting on, the scout now uses a AI evaluation of his own instead of going strictly off of pure ratings.

In theory...

My Baseline of 25/25/25/25 is now not only for Managers & GM’s to base their personal settings off of, it also applies to the scouting director.

So a GM who’s AI evaluation setting is 5 (scout) 5 (current year) 4 (last year) 3 (2 years ago) all based on my baseline of 25/25/25/25. Instead of the scout’s weight being strictly off of ratings, the scouts input is more of 5 (current year) 4 (last year) 3 (2 years ago)


Sent from my iPhone using Tapatalk

So what you're saying is, with 25/25/25/25, what that should do is cause every GM's/Scout's/Manager's AI Eval settings to essentially be "pure", right? Since the weights are balanced at 25% apiece, it would remove any artificial bias in player eval, and those people would be making their judgement solely on their natural weights.

For instance....

On 65/20/10/5 MLB default settings, I see BIlly Gasparino of the Dodgers has his eval for Scout/Current/Last Year/2 Years ago as 13/4/2/1.

If I use your 25/25/25/25 system, he changes to 3/8/5/4 in his eval.....which could be his "pure" ratings. None of the 4 categories are getting an advantage from the settings.

Am I understanding you right? I'm actually really buying into this idea, without even doing much testing, I can understand why this would make sense....I'd actually prefer to play this way, where staff are themselves and not swayed by my settings. It would also add another dimension for finding the best scout...besides just looking for neutral and best ratings, if I want someone who relies less on actual stats and more on the ratings, that's a 3rd attribute I now have to consider when hiring someone, which I love.

Maybe I've been thinking about these weights incorrectly....a scout who prefers stats>ratings will still be that way more so than his rivals, just as if you mess with stat output in the settings, power hitters will still stand out in the HR dept over others regardless of what the HR norm is. The weights merely attempt to guide everyone's process in an artificial direction, so by balancing it out, there is no artificial direction.

[PSUColonel]

But what if it came out weighting 2 years ago more than the present year?

[ThePride87]

Looking through the 30 MLB teams on a new save, the closest I saw to that happening was a few had equal weights for current/last/2 year weights....so thankfully none of them are doing THAT, but the concern's valid. Any scout worth their weight would never have 2-year stats>current year, hope that doesn't happen on 25/25/25/25 settings.

[SirMichaelJordan]

Yea that never happens furthest it goes is the same weight but it’s never more than than last year let alone current year.


Sent from my iPhone using Tapatalk

[endgame]

Originally Posted by ThePride87

So what you're saying is, with 25/25/25/25, what that should do is cause every GM's/Scout's/Manager's AI Eval settings to essentially be "pure", right? Since the weights are balanced at 25% apiece, it would remove any artificial bias in player eval, and those people would be making their judgement solely on their natural weights.

For instance....

On 65/20/10/5 MLB default settings, I see BIlly Gasparino of the Dodgers has his eval for Scout/Current/Last Year/2 Years ago as 13/4/2/1.

If I use your 25/25/25/25 system, he changes to 3/8/5/4 in his eval.....which could be his "pure" ratings. None of the 4 categories are getting an advantage from the settings.

Am I understanding you right? I'm actually really buying into this idea, without even doing much testing, I can understand why this would make sense....I'd actually prefer to play this way, where staff are themselves and not swayed by my settings. It would also add another dimension for finding the best scout...besides just looking for neutral and best ratings, if I want someone who relies less on actual stats and more on the ratings, that's a 3rd attribute I now have to consider when hiring someone, which I love.

Maybe I've been thinking about these weights incorrectly....a scout who prefers stats>ratings will still be that way more so than his rivals, just as if you mess with stat output in the settings, power hitters will still stand out in the HR dept over others regardless of what the HR norm is. The weights merely attempt to guide everyone's process in an artificial direction, so by balancing it out, there is no artificial direction.

Your way of perceiving this interests me. I'll be re-reading it for additional gleaning. I might add that the 'pure' idea is more abstract than the term's reputation, I think. All of these schemes are, at core, driven by the idea of prediction. That's where what's weighted becomes preference, and in the preference it's either risk or history-learned practice to garner confidence in that preference.

I'm not saying anything new, just saying it in a different way, maybe. One adds weighted elements or lenses for a reason, and that reason often will betray the most weight, whether we see it in the numbers or it occurs on the side of the monitor. Again, cool consideration, pride.

[jimmysthebestcop]

Originally Posted by ThePride87

Looking through the 30 MLB teams on a new save, the closest I saw to that happening was a few had equal weights for current/last/2 year weights....so thankfully none of them are doing THAT, but the concern's valid. Any scout worth their weight would never have 2-year stats>current year, hope that doesn't happen on 25/25/25/25 settings.

where can you see this info at? do you have to commish and edit the scouts?

[MizzouRah]

Originally Posted by jimmysthebestcop

where can you see this info at? do you have to commish and edit the scouts?

Game settings -> Global settings -> under Scouting settings -> Incorporate stats in scouting reports (change to YES)

Once you do this, look at your scouting director and you should see his evaluation settings.

[MizzouRah]

Originally Posted by PSUColonel

This is something Matt should be aware of.

I need to look into it a bit more and screenshot some good examples. It just felt "off" to me with it on after awhile when I initially liked it.

They didn't seem to go up in ratings, even a little bit with top tier current year stats.

Need more testing....

[jimmysthebestcop]

Originally Posted by MizzouRah

Game settings -> Global settings -> under Scouting settings -> Incorporate stats in scouting reports (change to YES)

Once you do this, look at your scouting director and you should see his evaluation settings.

Oh yes now there is a bar graph for current, last, 2 year ago stats.

Is the consensus to have Incorporate stats in scouting reports turned ON or OFF? Or are we still unsure.


interesting

[SirMichaelJordan]

Originally Posted by jimmysthebestcop

Oh yes now there is a bar graph for current, last, 2 year ago stats.

Is the consensus to have Incorporate stats in scouting reports turned ON or OFF? Or are we still unsure.


interesting

Unsure. It’s a feature worth exploring though. So far no noticeable problems.


Sent from my iPhone using Tapatalk