|
||||
|
04-20-2017, 11:42 AM | #1 |
Minors (Single A)
Join Date: Feb 2011
Posts: 65
|
AI Evaluation Settings
What player evaluation settings are you guys rolling with this year?
The default of 65/20/10/5 seems a bit too weighted towards ratings, I'm thinking of switching to 40/30/20/10 or something else. Anyone find a good balance yet? |
04-20-2017, 06:46 PM | #3 |
All Star Reserve
Join Date: Oct 2012
Location: Canada
Posts: 623
|
|
04-20-2017, 08:12 PM | #4 |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
base it on the behaviour you see... get a decent baseline first, then change it and see what you notice.
more stats = more knee-jerk reaction to results. if you won't want the guy who's hitting .400 with pure luck through 2 months to replace a guy that is clearly better than him hin the lineup, but with less results in a small sample, then use less stats. and vice versa. settings can influence the resulting effect... so ech person may give you good anecdotal info, but it may not apply to your league. let the resulting behaviour dictate, not some hypothesis about player evaluation dictate what you choose. it's mostly arbitrary what you choose, unless you do something extreme like 90/10 or 10/90 etc... nearing those points you may see drastically different behaviour. |
04-20-2017, 08:26 PM | #5 |
Hall Of Famer
Join Date: Aug 2010
Location: Indiana
Posts: 9,798
|
Default
|
04-20-2017, 08:35 PM | #7 |
Hall Of Famer
|
|
04-20-2017, 09:21 PM | #8 | ||
Major Leagues
Join Date: Jun 2015
Posts: 352
|
Quote:
Quote:
Thank you. To Clarify; I am asking because I just got back into OOTP with OOTP 17 (now OOTP 18) due to some issues out of my control, so I was on hiatus for quite some time. That said, I am still very new to certain things like AI evaluation and am looking for the 'best' results that would mirror/mimic the MLB around the 1960 era to the present 2010's. ANY suggestions would be great!! And, of course, I would still do my own DD (Due Diligence), yet still, I'd love to hear what the Long-timers here use as well. |
||
04-20-2017, 09:23 PM | #9 |
Major Leagues
Join Date: Jun 2015
Posts: 352
|
|
04-23-2017, 12:10 PM | #10 |
All Star Reserve
Join Date: Apr 2009
Posts: 518
|
How is the practical AI application of these settings determined or affected by whether player ratings are set at pure vs AI eval?
|
04-23-2017, 12:25 PM | #11 |
All Star Reserve
Join Date: Mar 2015
Location: Australia
Posts: 651
|
I am against setting your own ratings to AI eval. These should stay on pure.
|
04-23-2017, 12:50 PM | #12 |
Hall Of Famer
Join Date: Jul 2004
Location: The big smoke
Posts: 15,628
|
If you don't use AI eval then the settings under discussion are not used. Pure means ratings only.
__________________
Cheers RichW If you’re looking for a good cause to donate money to please consider a Donation to Parkinson’s Canada. It may help me have a better future and if not me, someone else. Thanks. “Conservatism consists of exactly one proposition …There must be in-groups whom the law protects but does not bind, alongside out-groups whom the law binds but does not protect.” Frank Wilhoit |
04-23-2017, 02:59 PM | #13 | |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
Quote:
the AI is the main reason i use this option, because the game is built to use it. just like having injuries on, trading, scouting innacuracy etc... same reason most probably like things somewhere near default on ai eval., too. i think off is a vallid setting for the discussion, but if oyu have not played with it on, it may be worth checking out. i've played both ways. you probably won't notice a huge difference. Also, if you drop accuracy below normal, it becomes more important to include stats for the AI... a human can just 'do' it... the ai has to be told to specifically include this year and last or whatever (it varies the gm strategies around that setting -- they are not all robotically the same ai evaluation.) in fact, i'd guess that if you play 100% scouting, the game probably works better with 100% ratings. otherwise you'd just be hamstringing the ai with less confident info to work with. Last edited by NoOne; 04-23-2017 at 03:01 PM. |
|
04-25-2017, 03:52 AM | #14 |
Major Leagues
Join Date: Jul 2014
Posts: 405
|
I go 40/40/15/5, with the idea that I want to see teams sometimes realistically play players who are playing well, even if they've actually just been lucky, as happens in real life.
To play fairly with that, I don't adjust downwards my view of the overall ratings of players having great years or vice versa, not allowing myself to mentally adjust for the overrating of players with good stats (and underrating of those with bad ones). |
05-03-2017, 12:23 AM | #15 |
Major Leagues
Join Date: Mar 2004
Location: Saskatchewan
Posts: 363
|
I'm experimenting with a different way of using these settings.
I think this approach could be used with other settings for weights. Also I run a single fictional league so this might not work well with multiple leagues that have different offseasons ratings, current yr, prev yr, 2 yr ago Jan 1 and for first 40 games 40,10,35,15 40-80 games 40,30,20,10 80 games-Dec 31 30,50,15,5 Change on/after Jan 1 because of updated scouting. From Jan 1 until pre season the current stats and last yr stats show the same so it's essentially like a total weight of 45. My thoughts are low weight of 10 in current stats early in season helps with hot/cold starts in a small sample size. I'm trying out the manager note in the managers office to set up a recurring reminder. |
05-03-2017, 12:47 PM | #16 |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
i know of at least one other post in the past that speaks in a similar way, you may want to search and see if he gives useful info, too.
i know someone does the trading settings dynamically depending on time of year, too. if i weren't so absent minded, i'd defintieyl jump on these bandwagons. reminders would help, but it would be a chore to me eventually... i don't do well with those. |
05-05-2017, 05:44 PM | #17 | |
All Star Reserve
Join Date: Apr 2009
Posts: 518
|
Quote:
Re reminder -- just add it as a league event. |
|
05-05-2017, 06:05 PM | #18 |
Hall Of Famer
|
The more I look at this...I believe the AI needs as much help as it can get to act rationally, therefore I am switching to the default of 65/20/10/5 and think it does the best job for the AI. If someone is absolutely hell bent on having more stats in the evaluation, then 55/25/15/5.
|
05-05-2017, 07:22 PM | #19 | ||
Hall Of Famer
Join Date: Aug 2010
Location: Indiana
Posts: 9,798
|
Quote:
Quote:
I told you so. |
||
Bookmarks |
|
|