A bit rusty, need a refresher

towerbooks3192 · 04-04-2018, 08:37 AM

I am a bit rusty and I am trying to get back on track. Used to play older OOTP from time to time but never really committed to playing it.

I was wondering if someone can give me any advice on using ratings and stats in evaluating players?

Also one other concern I got is when creating lets say an eight team fictional league, what is the optimal number of games that I should be letting them play?

Lastly, any good baseball books that can refresh my memory? I already ordered the book, baseball between the numbers, moneyball and bill james baseball abstract. I also got smart baseball coming in the mail tomorrow. Any book advice especially with regards to stats would be great.

NoOne · 04-04-2018, 09:47 AM

https://fivethirtyeight.com/features...eams-baseball/

this article says that there is only a 40% confidence level that the current mlb schedule's best record is actually the best team in baseball. 538 is a competent site.

more games played is better and balanced is better if you want the best teams to make it more often. certainly not a requirement and won't make it that much more certain the best team gets to the playoffs.

8 teams and use a 162ish schedule and you even get mlb-like stats over time, if you want.

24*7 168games 22*7 is 154g etc. need even # of games if you want equal h/a series. with 22 you will need 1 four game series or (2) 2-game series per team per year, too.. simply pick something out..

no matter what there will be a large amount of error either way. bottom line, make it the length you enjoy playing with 8 teams.

ratings to evaluate? as in ai evaluation settings? default is great... if you want more reaction to small-ish sample sizes, increase % to the 3 years of stats and reduce ratings weight. unless you do something drastically different form deafult you probably won't notice much different. just the fringe choices of an ai manager.

if you have a larger resolution scale, like 1-100, you can draw a clearer line in the sand as far as who is average, good or great. othwerwise, rounding errors come into play.

typically the guys near 80% of your scale for are the elite. and this pretty much coincides with the color code you see on the ratings -- blue/dark blue range etc..

guys in the next tier down won't be as consistent. might still hit the same peaks, but less often / less probable etc.. and goes down from there in a similar manner.

even ~average players can have 1 great year... so this is why you have to be wary of trusting stats without a long track record.

make sure whatever window of time you are evaluating that their scouting reports tell you they are basically the same ratings during this time, or it's even less clear from stats. (assume small undulations over time as jsut normal scouting inaccuracy or minor changes that won't create a huge blip)

this is why in ootp, at least, mil stats won't tell you much about future success. it can only show you what their current ability amounts to.. and if that is a far cry from what it can develope into just what can it tell you for sure? ratings are the cause of stats in this game.. and if they are different, you have a different player form moment to moment in the minor leagues.

in real life you can predict form mil stats much better. you can easily inflate mil stats by letting them play at a level they are too good for in ootp, just like in real life too.

also notice anyhting that is an aggregate of multiple more basic ratings..

e.g. Contact equals 3 weights applied to babip ratings, avoid k's rating and power rating... so very different players can equate to teh same Contact rating that you read in the player profile. so, that tells you not all contact ratings that are equal mean the same thing when it comes to results.

same logic why you should not pay too much attention to overall rating when comparing 2 similar players in a decision. lots of combinations of players add up to a "70" out of 80 overall...therefore, not all 70 rated palyers are the same in resultingn stats.

@normal accuracy and a good scout/budget i'd strongly recommend using stats to verify scouting accuracy and not react too much to them without a large sample of data. you bump it down to very low accuracy orlow accuracy and you might want to weigh stats a bit more (for yourself, not ai evaluation settings).

Orcin · 04-04-2018, 09:58 AM

If you play to play out all of your games, you might want a shorter schedule. The early 1900's schedule was 140 games, which works out to 20 against each team.

If numbers mean a lot to you, then you should strive for 154 games. The extra 14 games will mean the season totals will be a little closer to what current MLB is. A 162-game schedule is not ideal for 8 teams because it doesn't allow for the same number of games vs. each team.

04-04-2018, 08:37 AM	#1
towerbooks3192 Minors (Single A) Join Date: Jul 2014 Posts: 70	A bit rusty, need a refresher I am a bit rusty and I am trying to get back on track. Used to play older OOTP from time to time but never really committed to playing it. I was wondering if someone can give me any advice on using ratings and stats in evaluating players? Also one other concern I got is when creating lets say an eight team fictional league, what is the optimal number of games that I should be letting them play? Lastly, any good baseball books that can refresh my memory? I already ordered the book, baseball between the numbers, moneyball and bill james baseball abstract. I also got smart baseball coming in the mail tomorrow. Any book advice especially with regards to stats would be great.

04-04-2018, 09:47 AM	#2
NoOne Hall Of Famer Join Date: Apr 2015 Posts: 7,167	https://fivethirtyeight.com/features...eams-baseball/ this article says that there is only a 40% confidence level that the current mlb schedule's best record is actually the best team in baseball. 538 is a competent site. more games played is better and balanced is better if you want the best teams to make it more often. certainly not a requirement and won't make it that much more certain the best team gets to the playoffs. 8 teams and use a 162ish schedule and you even get mlb-like stats over time, if you want. 247 168games 227 is 154g etc. need even # of games if you want equal h/a series. with 22 you will need 1 four game series or (2) 2-game series per team per year, too.. simply pick something out.. no matter what there will be a large amount of error either way. bottom line, make it the length you enjoy playing with 8 teams. ratings to evaluate? as in ai evaluation settings? default is great... if you want more reaction to small-ish sample sizes, increase % to the 3 years of stats and reduce ratings weight. unless you do something drastically different form deafult you probably won't notice much different. just the fringe choices of an ai manager. if you have a larger resolution scale, like 1-100, you can draw a clearer line in the sand as far as who is average, good or great. othwerwise, rounding errors come into play. typically the guys near 80% of your scale for are the elite. and this pretty much coincides with the color code you see on the ratings -- blue/dark blue range etc.. guys in the next tier down won't be as consistent. might still hit the same peaks, but less often / less probable etc.. and goes down from there in a similar manner. even ~average players can have 1 great year... so this is why you have to be wary of trusting stats without a long track record. make sure whatever window of time you are evaluating that their scouting reports tell you they are basically the same ratings during this time, or it's even less clear from stats. (assume small undulations over time as jsut normal scouting inaccuracy or minor changes that won't create a huge blip) this is why in ootp, at least, mil stats won't tell you much about future success. it can only show you what their current ability amounts to.. and if that is a far cry from what it can develope into just what can it tell you for sure? ratings are the cause of stats in this game.. and if they are different, you have a different player form moment to moment in the minor leagues. in real life you can predict form mil stats much better. you can easily inflate mil stats by letting them play at a level they are too good for in ootp, just like in real life too. also notice anyhting that is an aggregate of multiple more basic ratings.. e.g. Contact equals 3 weights applied to babip ratings, avoid k's rating and power rating... so very different players can equate to teh same Contact rating that you read in the player profile. so, that tells you not all contact ratings that are equal mean the same thing when it comes to results. same logic why you should not pay too much attention to overall rating when comparing 2 similar players in a decision. lots of combinations of players add up to a "70" out of 80 overall...therefore, not all 70 rated palyers are the same in resultingn stats. @normal accuracy and a good scout/budget i'd strongly recommend using stats to verify scouting accuracy and not react too much to them without a large sample of data. you bump it down to very low accuracy orlow accuracy and you might want to weigh stats a bit more (for yourself, not ai evaluation settings). Last edited by NoOne; 04-04-2018 at 09:49 AM.

04-04-2018, 09:58 AM	#3
Orcin Hall Of Famer Join Date: Aug 2010 Location: Indiana Posts: 9,798	If you play to play out all of your games, you might want a shorter schedule. The early 1900's schedule was 140 games, which works out to 20 against each team. If numbers mean a lot to you, then you should strive for 154 games. The extra 14 games will mean the season totals will be a little closer to what current MLB is. A 162-game schedule is not ideal for 8 teams because it doesn't allow for the same number of games vs. each team. Last edited by Orcin; 04-04-2018 at 09:59 AM.