The Annual Settings Thread

[The Wolf]

Originally Posted by frank_olaf

Wolf, if I just want to play a standard MLB game starting in 2016 would you still recommend stats only or do you only recommended it if I sim a few years first?

Stats only is not recommended (and won't really work) for any league where you have advance knowledge/expectations of player abilities.

[mmarra82]

Originally Posted by The Wolf

A great post from Fyrestorm about stats-only. Turn rating and stats off, set trading to hard/neutral and follow this and you're off.

Only way I play now...was tough at first but after fumbling through a bit I absolutely love it.

I do use ratings for other ratings on a 2-8 scale.

I do not use feeder leagues.

[TwinsGuy11]

Originally Posted by The Wolf

Stats only is not recommended (and won't really work) for any league where you have advance knowledge/expectations of player abilities.

See I think it could be interesting, though dependent on how far you get.

For instance, I have knowledge of certain people's abilities but what if they actually become really good players but I just don't trust it because of my prior opinion on the player?

But like you said before, this game is so flexible and can be made to appease anyone. This may just be me that thinks it could work.

[The Wolf]

It's your game, play it your way. But stats only is not recommended when there is existing player quality knowledge. It works by requiring player evaluation via scouting reports and stats only.

That being said, do whatever you like.

[Syd Thrift]

Originally Posted by TwinsGuy11

See I think it could be interesting, though dependent on how far you get.

For instance, I have knowledge of certain people's abilities but what if they actually become really good players but I just don't trust it because of my prior opinion on the player?

But like you said before, this game is so flexible and can be made to appease anyone. This may just be me that thinks it could work.

Yeah, there's a chance you could be off when you draft Reggie Jackson in a stats-only league, but where the AI only sees his college stats and maybe has just an inkling of his potential ratings, you have the additional information that he's Reggie freaking Jackson. Sure, he might not work out, but he's got a much better chance of being a great player than Billy Joe Robidoux. And that does give you an advantage in a league.

The flip side of that - that you won't be as invested in the teams if you don't know who the players are - is a real issue but I know I tend to become rather attached to these players, David Creamer style, just by having to research them to know what kind of player they are. I also like to do things like play with nicknames to help with that (you do need the editor turned on to get those but if you can make yourself not look at the right side of the screen while you're in there you should be OK).

[TwinsGuy11]

Originally Posted by Syd Thrift

Yeah, there's a chance you could be off when you draft Reggie Jackson in a stats-only league, but where the AI only sees his college stats and maybe has just an inkling of his potential ratings, you have the additional information that he's Reggie freaking Jackson. Sure, he might not work out, but he's got a much better chance of being a great player than Billy Joe Robidoux. And that does give you an advantage in a league.

The flip side of that - that you won't be as invested in the teams if you don't know who the players are - is a real issue but I know I tend to become rather attached to these players, David Creamer style, just by having to research them to know what kind of player they are. I also like to do things like play with nicknames to help with that (you do need the editor turned on to get those but if you can make yourself not look at the right side of the screen while you're in there you should be OK).

Yeah that's a good point. I think starting in 2016 may be a bit different just because you won't be drafting players that you know already but it's still a valid point.

[mmarra82]

I run stats only, start in 2016 and enjoy. I am not aware of the draft class, so I have the unknown about how they will turn out.

[koohead]

Has anyone tinkered with setting a very low (5%?) value for Ratings evaluation and displaying only Potential Ratings? This might be the "not so hard" mode for OOTP. considering this because it seems that IRL you have pages of scouting reports and evaluations from staff to go with statistical performance when evaluating your organization, so a minimal weight on ratings might make up for the lack of that depth?

[SirMichaelJordan]

Originally Posted by koohead

Has anyone tinkered with setting a very low (5%?) value for Ratings evaluation and displaying only Potential Ratings? This might be the "not so hard" mode for OOTP. considering this because it seems that IRL you have pages of scouting reports and evaluations from staff to go with statistical performance when evaluating your organization, so a minimal weight on ratings might make up for the lack of that depth?

I have no ratings for evaluation but keep all rating info on.

No different from a scouting card.

[Matt Uk]

Originally Posted by The Wolf

A great post from Fyrestorm about stats-only. Turn rating and stats off, set trading to hard/neutral and follow this and you're off.

I'm assuming yes, but do you set the AI to sabremetric lineups? I'm wondering with the AI not considering the actual ratings whether it might get too much into platooning for example by almost micromanaging the sabremetrics to the nth degree

Also are you hitting the recalculate Gm tendencies based on these weights. Again I assume so.

[chriskelly]

What I never understood is, when you give like 50% weight to current year stats, but it's 10 games into the season, how does the computer handle it?

[PSUColonel]

To me, one way to mimic stats only, while still retaining a realistic experience is to put ratings on a 2-8 scale and then use AI evaluation at 10/40/30/20....20/40/30/10, or 30/40/20/10

[NoOne]

Originally Posted by bigsmooth

"No stars, No ratings" --- scouting set to LOW --- What do you rely on when drafting and evaluating players then?

small sample sizes against poor and varying competition. essentially you won't have very reliable data - not criticism, just facts.

turning off ovr/pot is inconsequential to a human... not sure how it affects the AI. opinion: actually this sounds awesome, lol. if it forces the AI to use individual ratings moreso than ovr/pot in decisions i would love it!

future success has an extremely weak correlation to hs/college data... and minors data, too. (unless they spent 3-5 years at one level and had no talent changes.... at that point those stats might be revealing). this is undeniable fact... but not an argument against using 100% stats in any way. just a better understanding of it.

as in real-life, a player is limited by their talent. in real life scouts only guess as to what this is, whereas the game knows what this is in an absolute manner. adding innacuracy or using 100% stats helps mask this absolute knowledge of talent. opinion: doing one or the other makes sense.

personally, i use inaccurate scouting. i've always used the normal setting, but i am strongly leaning toward the next notch down. at normal accuracy and a high scouting budget with a great scout is just too good, imo. you may think differently.

assuming normal accuracy on ai eval:
this is just going to introduce a larger % error on AI decisions the more you let stats influence decisions. a guy who should hit 300/350/500 can easily have a couple bad years, and be undervalued - not unlike real-life. a more borderline player may lose his spot to a less deserving talent.. etc etc...

some combinations of skills do create more synergy than others... if the ai is using ovr/pot for the bulk of the decisions, there's not much we can do to help them by changing settings.

with all that said, i do make use of it. i want AI GM to be human in their decisions... ie make mistakes on playing the odds etc... i think it's 40ratings 36/18/6 yearly stats.

this is one of those things that people will tell you "this will work well" but realyl have no data to back it up. these are total guesses. the amount of research to flesh out the actual implication of these AI Eval settings would be totally not worth the effort, lol.

what i base any assertion on: when i go from 100% accurate to AI eval in the paste, league-wide stats dropped significantly (not based 1year of data, more like 100+). therefore, lesser players are getting playing time. LTMs are static in my leagues and all other factors were unchanged. this was not '17 nor '16. so, things could have changed. trust that the scientific method was employed, so i don't have to write a book about the sim tests.


final note:

becareful of "Normal" (assumes agreement on what it is, which is a problem of its own, obviously)

this is an opinion thread, so don't think i am applying this here. no one was expected to bring 1000 years of data to support their playing style, lol.

1) most have no idea what it is to begin with. they will make comments about too much or too little but not actually know what the baseline is. even those that are enlightened can easily make this mistake... everyone can be fooled by randomness at any time.

2) most don't bother too compare results in game with "normal." if they do, they use one year or one week of unusual behaviour to justify a change. if the sample size isn't large enough to contain all or most possibilities and in their proper proportions, the data is mostly useless... it becomes more art than math at that point. not a problem if you are finger-painting, but a big problem when trying to make a logical decision based on reality (or the reality of the situation within a video game, lol).

e.g. in these Forums, not specifically this thread, i see a lot of arguments on both sides of the fence for many topics and often neither side has made any attempt to understand what actually happens in reality and in the game before jumping to conclusions about what it "should be."

it is just a constant back and forth of what each side "feels" about it, lol. don't bother with facts... you might reach a consensus..

[PSUColonel]

I would be nice if perhaps prospects and established players could be judged differently.

[PSUColonel]

I've been using 40/30/20/10 which I feel creates a good mix

[NoOne]

Originally Posted by chriskelly

What I never understood is, when you give like 50% weight to current year stats, but it's 10 games into the season, how does the computer handle it?

i would guess it doesn't use the stats until they exist, or possibly adds an additional weight based on how much of that season's stats are available etc... bottom line, it doesn't do anythign too %$@#ed up.

i have a little experience with this lately, in '17. i erase elague history and mlb stats when i use a new set of LTM to test (finding baseline environment is my first task with each ootp version). while i cannot tell if something small is occurring, i can attest that nothing crazy drastic is happening.

if it is a concern, start your league 3+ years prior to the year you want to take control. you can erase league history, but you have to keep the mlb stats, obviously.

[PSUColonel]

It's prorated based on how far into the season you are

[SirMichaelJordan]

Originally Posted by PSUColonel

It's prorated based on how far into the season you are

it's either PA's or AB's for batters, not sure about pitchers.

[Dem Bums]

Dumb question: Does lowering AI trading frequency impact the AI's willingness to trade with a human player or does it only impact AI to AI trades?

[koohead]

Originally Posted by SirMichaelJordan

I have no ratings for evaluation but keep all rating info on.

No different from a scouting card.

Never thought about it that way. makes sense.