Starting new game soon, ratings settings suggestions invited

[Brad K]

My current game is 1951 to the end of Barry Bonds career. Its 2008 and Bonds has a league minimum contract with Cleveland after failing the vesting options for a third year with Boston at $26 million. I expect retirement because his role with Cleveland is a PH who can draw walks.

Anyway to the new game. I've always used neutralized stats but I've recently learned that "real stats" are neutralized somehow during the rating creation process. I'm unclear on this but think its done somehow using the era_balparks file.

I'm not happy with the default era_ballparks either for creating neutralized stats or for park factors. The reason is the file gives parks the same rating for everything. I understand it was created by working backwards from runs produced, which as far as overall averages go is probably OK. However my secondary purpose in playing is to see if anyone hits over 714 home runs. So I need realistic park factors and realistic player ratings to do this. I start 1951 to include all of Mays' and Aaron's careers in the game.

I'm OK with BABIP controlling singles doubles triples but I want separate L/R for that and also HRs. I've though about creating a era_ballparks file from calculated park factors but wonder if that works.

[Garlon]

My suggestion is to use real stats and the neutral ballpark file. I will need to look into this further, but I am quite certain the game is adjusting the players incorrectly based on the era_ballparks file and the game is not applying the factors correctly in the outcomes. When I did a test in that other thread, one ballpark had 100 HR in the season and another had 1 in the same league, so something is not correct there. I wonder if the game is also applying some sort of additional factor for the distance to the walls that is messing up the outcomes. If so then I need to create another file with all of the walls set to the same distance in the file.

[Brad K]

Sounds like what you're suggesting is a lesser evils choice.

Where do I get a neutral era_ballparks file?

[Garlon]

Try this file, rename it to era_ballparks to use. This works with the "Major League Stadiums 1901-2019 by Silvam14" mod in the workshop too. Make sure you install this file after the workshop mod though because the workshop mod comes with the discrete factors.

[Brad K]

Thank you for the file. What is the stadiums mod you're talking about? I have downloaded some of the silvam ballparks.

[Garlon]

There is a mod on the OOTP21 Workshop named "Major League Stadiums 1901-2019 by Silvam14" that has the entire collection of stadiums by SIlvam14 and Adion. That collection uses naming conventions consistent with this file so that when you start a league all of the stadiums will automatically load for each team. I think Markus also updated OOTP so that each season new stadiums will also load too from the file. Basically OOTP takes the ballpark names associated with each team from the era_ballparks file as well as the factors for each season and it then looks in the ballparks folder for an exact match with the "prk" files for that stadium. With this mod the models, pictures, and diagrams will all load for each stadium each season. When you install it though it will come with the discrete ballparks file though and overwrite the original era_ballparks file that comes with the game, however it does install an extra copy of the original era_ballparks file renames era_ballparks_original so that you still have the file.

[Brad K]

You mean on Steam, right? I bought direct.

[Brad K]

Originally Posted by Garlon

My suggestion is to use real stats and the neutral ballpark file.

That is what I'm going to do. I've done a lot of reading and have found statements concerning the following which may be true or may not be true. The whole thing is a mess.

1. The park factors in the file with the game don't match what's on Baseball Reference and other sites. Reason given is the game has a formula that doesn't work with "standard" factors. Well, gee, change the formula not the data!

2. Park factors don't behave as might be intuitively supposed, that is, a factor of 1.050 doesn't mean 5% more of whatever is being rated.

3. Most consider BA to be BA, not BABIP.

4. BA or BABIP may adjust 2B and 3B factors but if so no indication there are adjustments for this in the systems that provide 2B and 3B ratings.

5. One old system uses runs scored as its basis and calculates the quantity of each type of hit needed to produce that number of runs. Question becomes is park factor intended to produce realistic runs or realistic hit types.

6. There's an on line park factor calculator based on a spread sheet by Gambo, calculated factors provided by Silvam, and a calculation of origin unknown to me in the game, none of which produce the same factors.

7. There's an argument that factors based on actual events in a large part measure the home team which may have been designed for the park, and thus are only reliable for historical replays.

8. There's an argument that calculated factors represent what should be rather than what was and are better for games where players are fictional or if real many are not on their historical teams.

---------

I think there are some things I missed but this should be enough to conclude this is a huge mess. I don't see that progress is possible until its known how the game uses park factors, both in game play and in adjusting player ratings. After that, decisions need to be made concerning the purpose. Is it to produce realistic runs, produce realistic hit distribution, and is that in the context of a historical replay or a game with fictional players or historical players many not on their historical teams.

Anyway, with great disappointment I'm starting my next game with ratings based on real stats and a 1,1,1,1 factor file.

[Garlon]

1. The values you will see on baseball reference and on other sites for these values may be for runs scored or for the BA/2B/3B/HR. For OOTP everything is based upon events so anything like runs scored need to be translated into component values, which was the purpose of the quadratic formulas as it was the best option when those were created. The baseball reference formula for those factors uses their own formula and will not always match the same values you will get when using data from retrosheet for those parks. For component ratings that you see reported they do not take into account per AB as what is needed for the game and are just some general information about total HR in this stadium against the league average per game, not per AB which is what we need for the game. Everything that I generated was from the per AB values, in fact for HR it is per AB - K so that strikeout totals are not affecting the result.

2. This might be true, but that needs to be looked into as they may not be getting applied in outcomes as they are intended. We need to take a closer look at this. They should just apply to the batter portion of the outcomes as a multiplier.

3. Yes, most people do consider the BA factor to be batting average and not batting average on balls in play, but this is how it has to be because HR are a different calculation and not have it as BABIP means that it would also be affecting HR rates which have their own factor.

4. When I created the original quadratic factors I asked Markus if this was the case and he confirmed this. So 2B and 3B are a subset of BABIP as a factor.

5. Modeling accurate ballpark scoring should turn out the same with quadratic factors or discrete factors when implemented properly. We did not have the discrete data we needed when the quadratic factors were originally made for the game so I made them from the available runs data and put them through the quadratic formula based on individual runs created formulas that account for how runs were generated each season based on singles, doubles, triples, and hr to generate one single value that could be applied to all of these simultaneously to get the desired result.

6. I talked to Silvam about his park factors and he attempted to average many seasons of data using I believe an online resource that was not in a format that OOTP needs to generate outcomes. I put no stock in a calculator based upon dimensions, but people want it for fun I guess. Playing in humid conditions at sea level will be different than in dry conditions 2000 ft above sea level and then there is wind and then there is the distribution of where balls are actually hit, whatever the effect are of all of those differences in environment actually are they will be represented in the real data for stadiums though which is what the discrete factors are based upon (though I do not currently believe that OOTP is applying the factors correctly as is intended for the game). The affect of the weather file should be disabled for historical games because of this.

7. The argument about the home team is bogus and comes from people who do not understand how the factors are calculated. The factors come from the team and their opponents. Teams do not have a home squad and an away squad, bot do their opponents. Additionally, those factors are across three seasons. The idea of building a team around your home ballpark in inherently flawed anyway because you will still play half of your games on the road and if your team is specifically designed for some home advantage then it is specifically designed for some road weakness. The discrete factors use rates, not overall totals of HR as the comparison, so it does not matter if the game is in replay more or if players are on different teams.

8. The historical factors are only designed for the context of a given season with the same players in the league, but the teams they play for make no difference. There are no absolute park factors. Using stadium models out of their original context can be fun and if someone wants to have OOTP generate the factors in that situation that is fine too, but there is no real way to determine the true factors for that.

Try not to get overly concerned about the park factors. I currently prefer the neutral factor file.

[Brad K]

Originally Posted by Garlon

Try not to get overly concerned about the park factors. I currently prefer the neutral factor file.

Neutral park factors and non neutralized stats. Meaning if he plays for the Cubs in 1966 Donn Clendenon has zero chance of hitting 50 HRs.

[Garlon]

If we were to adjust Clendenon properly for a RHB playing half of his games in Forbes Field in 1966 he goes from 28 HR in 571 AB to 33.5 HR in 571 AB. If we then place him in Wrigley Field for half of his games he goes from 33.5 HR to 40 HR.

Now the question is what are the odds that a player who can hit 40 HR per 571 will hit 50 HR given the same amount of AB in a season. For that we use binomial probability. He would have a 6.25% chance of getting to 50 HR under those conditions. The problem of course is that OOTP needs to be programmed to better adjust the player ratings to account for their home historical stadium (and I have created the import factors for every stadium for every season in a file) and then it needs to properly apply the actual ballpark factors to the probability outcomes in the game.

Now what are the odds that Clendenon will still get to 50 HR if he hits 28 per 571 AB and plays in a neutral stadium? About 1 in 14285 attempts at replaying the season.

[Garlon]

This is from an 1871-2019 game using real stats and the neutral ballpark file and some other import and strategy settings.

[Brad K]

I thought Clendenon would get more of a boost than that. While a Pirate he hit 35 home runs at Forbes and 71 on the road. The 1966 breakout is 25 road and 3 home.

At this point I'd be happy to just get L/R home runs into era_ballparks. It seems HRs would be where most of the scoring difference is.

Also I wonder if somehow park factors are affecting player creation with real splits. I've noticed some really unreal splits.

[Garlon]

There are many settings when creating a league. Make sure you turn off storylines under global settings too. For any season after 1901 you should turn on automatically expand league. I took that pic when creating an 1871 league that is why it is turned off.

[Brad K]

So I figured like last time I'd start the game in 1951 and take over the Pirates 1955. In 54 the Pirates went 85-69 and won the pennant. So exactly how am I going to fire the GM and take over?!?!?!?!?

Can't really blame anyone that the lost the World Series. Their opponent was Cleveland with a regular season record of 111-43.

[HumanRainDelay]

Originally Posted by Brad K

So I figured like last time I'd start the game in 1951 and take over the Pirates 1955. In 54 the Pirates went 85-69 and won the pennant. So exactly how am I going to fire the GM and take over?!?!?!?!?

Can't really blame anyone that the lost the World Series. Their opponent was Cleveland with a regular season record of 111-43.

Haha. Sounds like a good narrative for a creative writing project, and you could just fill in the blank below for a headline:

"How the Pirates' Administration __________ Scandal of '54 Changed the Course of History."

[Brad K]

Originally Posted by HumanRainDelay

Haha. Sounds like a good narrative for a creative writing project, and you could just fill in the blank below for a headline:

"How the Pirates' Administration __________ Scandal of '54 Changed the Course of History."

Good one. There's more.

The owner wants to appoint his 20 year old niece. He says she has plenty of experience managing pirates in the Caribbean.

Anne Bonny.

.

[Brad K]

Originally Posted by Garlon

Try not to get overly concerned about the park factors. I currently prefer the neutral factor file.

I think your suggestion is correct and I appreciate you making the file available. But the more I think about the necessity for this the more unhappy I get. Why, I had non neutralized stats and neutral park factors in my VERY FIRST baseball game!!!

[HumanRainDelay]

Originally Posted by Brad K

Good one. There's more.

The owner wants to appoint his 20 year old niece. He says she has plenty of experience managing pirates in the Caribbean.

Anne Bonny.

.

Well played!

[nebben]

Originally Posted by Garlon

Make sure you turn off storylines under global settings too.

Just curious, why turn off storylines?