Question re: AI Eval Settings & Aging

[GoNATS]

Long time FM player who only recently discovered OOTP via the Steam sale. I was really excited to find such an in-depth baseball management game, with such a helpful online community. I've squeezed in about 5 seasons and feel I have a good handle on the basics, but I am curious about how veterans have modified the aging and AI Eval for the 2014 version. Before I seriously invest in a fantasy league I would like to leverage the wisdom and insights of others. Apologies if this should be in Noob forum. My questions below:

1. What AI Eval settings have you found most useful and why? I've seen recommendations for 35/35/20/10 (PSUColonel I think) and 25 across the board (sorry can't recall who said that). I suspect lowering the values on ratings may mean an undervaluation of prospects whereas an increase in stats, particularly two seasons past, could get the AI overpaying for vets getting close to their 30s.

2. I'm also curios if folks have modified the Talent Change. on the one hand seeing a valuable player seemingly fall off the cliff is frustrating, thus warranting a lower value. But lowering the value may adversely affect the variability of how prospects develop, which of course is the fun of drafting - the fog of development.

3. Do the recommendations re: aging for the 2013 version still apply or does a new set of values work better in 2014? I'm guessing the former is not true, and given the great discussion that thread produced there was a meaningful change in the latest version. For a reference here was the balpark consensus from 2013: Batting Ageing: .250; Batting Development: 1.000; Pitcher Ageing: .375; Pitcher Development: .900. My goal is for most players to start exiting their primes in their early 30s w/o such a steep dropoff to their mid to late 30s. Maybe no as realistic but more enjoyable? Are the defaults sufficient?

I hope I'm not committing a faux-pas by asking for closely guarded secrets! Thanks in advance and Happy Holidays.

[Orcin]

I use 40/30/20/10 and .5/1.25/.5/1.25 with 50% TCR. Trade settings are Hard/Neutral. This combination has given me a very competitive fictional league. I think it would work well for a league starting from today also.

[PSUColonel]

Originally Posted by Orcin

I use 40/30/20/10 and .5/1.25/.5/1.25 with 50% TCR. Trade settings are Hard/Neutral. This combination has given me a very competitive fictional league. I think it would work well for a league starting from today also.

I too am using 40/30/20/10 with trading on very hard and favor prospects. As far as aging modifiers and talent randomness go, I use the default, as I feel there are very realistic, and are improved with each version. I recommend starting with the 2013 MLB QuickStart, and diving right in. I prefer to sim day by day using real-time. I also control depth charts, in the majors and minors. I sign all contracts, and playing in this slow detailed and methodical way allows me to become very immersed in the league.

[GoNATS]

Originally Posted by PSUColonel

I too am using 40/30/20/10 with trading on very hard and favor prospects. As far as aging modifiers and talent randomness go, I use the default, as I feel there are very realistic, and are improved with each version. I recommend starting with the 2013 MLB QuickStart, and diving right in. I prefer to sim day by day using real-time. I also control depth charts, in the majors and minors. I sign all contracts, and playing in this slow detailed and methodical way allows me to become very immersed in the league.

I'm not sure I have the strength of will/ability to defer gratification by simming in real-time! I assume you handle the cuts and promotions in the minors too. Do you think your approach noticeably improves player development, or is it a marginal improvement over the force start, great coaching and big player dev budget hands off approach to the minors. Trying not to get too sucked into the OOTP vortex, I think kids need their dad right

[Honorable_Pawn]

I like to give ratings a "nice chunk" but as far as CY/PY/PY+1 I like to equalize those a little bit.

My thought process is the following:

Let player A be a stud.
Let Player A have a bad current year (he is in a mid-season slump).
Let CY AI eval be very high compared to PY and PY+1.

The AI doesn't like this guy as much as he is worth and will grossly undervalue him. It becomes to easy to spot a stud having a bad year around the trade deadline and fleece him from the AI. Especially if you have AI favor prospects.

The opposite is true. Let's say a scrub had a sensational CY and is negotiating an extension. The AI will overvalue him and pay too much.



Another good reason to equalize CY/PY/PY+1 is so that the AI will value proven players over unproven players. You'll see less flash-in-the-pan up-and-down movement from AAA to ML. This helps the AI to preserve its minor league options and maintain its roster strength. **Actually, I recommend turning off option years**

This what I am using:

Attachment 318323

Next season I will switch ratings to 32 and PY+1 to 19.

[Honorable_Pawn]

A lot of people around here are sold on favor prospects too. Be careful. Some of us disagree.

When you make the AI favor prospects it is waaay too easy to sell a bird in the bush for a bird in the hand. Especially low-level super prospects that are about as likely to flame out as they are to develop.

I suggest:

Low
Hard (very hard)
Neutral

IMO, neutral is the best setting to make the AI strongest.

[Honorable_Pawn]

Originally Posted by GoNATS

I'm not sure I have the strength of will/ability to defer gratification by simming in real-time! I assume you handle the cuts and promotions in the minors too. Do you think your approach noticeably improves player development, or is it a marginal improvement over the force start, great coaching and big player dev budget hands off approach to the minors. Trying not to get too sucked into the OOTP vortex, I think kids need their dad right

I used to sim seasons very quickly when I started. It was fun for awhile but then I realized that I was not attached to my league. That I didn't "know" the stories. That I didn't "know" the history.

Now, I randomly play-out one game from every series. I gain attachments to my players and learn all of the other players in my league this way. I sim the other two games of each series but I only sim from in-game. I set the line-ups in-game then click the sim to "game" button.

After each game I go through my transactions screens for my minors and move guys to 7-day DL, if necessary. I bench tired players for 2-3 days so they don't get ruined by the AI. I handle all promotions but I let the AI set line-ups. I have attachments to minor league guys too. I don't just have random names in my minors. I know who they are. I know how long they have been there. I know which ones are just organizational guys. It's so much better to have that familiarity, IMO, that you can't get from simming long stretches. It takes me 3-6 weeks to get through one season.

I found that I enjoy it this way more than just blowing through season after season.

Anyhow...It's your game. These are just my thoughts.

[Honorable_Pawn]

Lastly,

You mentioned aging. There is so much debate around here about aging. I have read all of the threads around here that have used "simple" statistical comparasions and basic "eyeballing" of results in order to come up with good aging modifiers.

I feel that Marcus has done a pretty good job getting this right. I only tweak them slightly. Keep in mind that even large tweaks (unlike all other settings in the game) have diminishing results. So there won't be a big-giant change from 1.00 to 0.25. Yes, there will be a difference but not as big as you'd expect when you compare them to other modifiers in the game. **Stat output modifiers are very touchy, aging modifiers..not so much***

Anyway this is what I use for aging:

Attachment 318324

[Lukas Berger]

Originally Posted by Honorable_Pawn

I feel that Marcus has done a pretty good job getting this right. I only tweak them slightly.

I agree. Markus put a lot of effort into getting the modifiers right and used a lot of real life data to do so.

While many folks on the board have advocated their own favorite settings, I haven't seen anyone provide enough hard data to convince me that their settings are more accurate than those Markus came up with.

[le receveur]

Originally Posted by lukasberger

I agree. Markus put a lot of effort into getting the modifiers right and used a lot of real life data to do so.

While many folks on the board have advocated their own favorite settings, I haven't seen anyone provide enough hard data to convince me that their settings are more accurate than those Markus came up with.

generally agree. only place where i feel the progression is a bit fast in normal ML setting to get from Short Season to AA because there is not as much, if any, steepness between A- and A+ compared to RL where this is generally a major step for anyone but top prospects.

[GoNATS]

Originally Posted by Honorable_Pawn

Lastly,

You mentioned aging. There is so much debate around here about aging. I have read all of the threads around here that have used "simple" statistical comparasions and basic "eyeballing" of results in order to come up with good aging modifiers.

I feel that Marcus has done a pretty good job getting this right. I only tweak them slightly. Keep in mind that even large tweaks (unlike all other settings in the game) have diminishing results. So there won't be a big-giant change from 1.00 to 0.25. Yes, there will be a difference but not as big as you'd expect when you compare them to other modifiers in the game. **Stat output modifiers are very touchy, aging modifiers..not so much***

Anyway this is what I use for aging:

Attachment 318324

Many thanks for detailed response, very helpful, especially your thoughts re AI Eval. My hunch was that the defaults were quite good especially this late in development. I don't know that I'd even touch the stat outputs (I.e. League totals modifiers/ PCMs right?). Have you messed around with those? I imagine it might be fun just for kicks, otherwise hands off if you want realistic stat outputs.

[Honorable_Pawn]

Originally Posted by GoNATS

Many thanks for detailed response, very helpful, especially your thoughts re AI Eval. My hunch was that the defaults were quite good especially this late in development. I don't know that I'd even touch the stat outputs (I.e. League totals modifiers/ PCMs right?). Have you messed around with those? I imagine it might be fun just for kicks, otherwise hands off if you want realistic stat outputs.

When I feel it is time for a change in stats output I just select a year and load whatever is there. Stat output works very well on the macro level and pretty good on the micro. Naturally, we can nit-pick but all-in-all the stats produced are very good.

Once you have comfort you'll find yourself tweaking to get what you want.

One more suggestion if I may:

Don't spend a lot of time on set-up on your first few universes. Play with things...experiment. You need to break things on a universe that you can stand to lose.

Like me...I almost quit OOTP because I had invested so much time in set-up of a Florida League. I didn't realize that deleted player ID's were recycled. One time I mass deleted a bunch of free-agents. It screwed up links to all kinds of stuff in my history section.

I ended up scapping the league. I learned to always retire. Never delete. These things you need to do before you settle in on your "Keeper". Coming from FBM you probably already know what I mean.