Historicals: Room for Improvement

[pstrickert]

Here's what I'd like to see improved for historicals in OOTP10 (in order of importance):

1. More accurate defensive ratings
2. More refined league strategy settings*
3. Fix the initial import season problem**
4. Choice of manager "types"***
5. Option to use as-played lineups and historical transaction files
6. Option to set player "usage" percentages

*relief pitcher usage is a problem for certain periods in league history (e.g., two-inning relief stints instead of one); also see SteveP's thread on bunting for the problem replicating the use of sac bunts for certain periods of league history

**since LTMs do not "kick in" until year 2, the intial season can produce less accurate results

***so that we can match team strategies as much as possible with their historical managers' tendencies

[Spritze]

3a. (as the cause is the same) Fix minor league totals so they are in line with major league rates.

In both cases the same adjustment needs to be made, the difference is the minor leagues need adjusting every year and for each level. Major leagues only need an initial year adjustment.

[Garlon]

Regarding pstrickert's initial post:

1. Yes, this was my biggest gripe with OOTP9 historicals.

2. We can work this out in beta-testing this year. There is only so much that we ca do in terms of reliever usage. Right now OOTP produces the correct distribution of IP between the SP's and bullpen for every season in history. Getting more pitchers into the games in historicals is actually good for a couple of reasons. #1 The AI alreaady knows which pitchers will perform the best (something that real life managers did not know) so they already get spots on the pitching staff while other pitchers who pitched poorly IRL end up in the minors or on the reserve list...this means that hitters in OOTP are always going to be facing higher quality relievers than they did IRL. #2 If we go a step further with this, then the best o the best will be pitching even more...and as a result this will drive up SP ERA's, which will of course lead to more losses or no decisions by real life SP's.

Having said that, we can still make some minor improvements with RPs.

3. We can make some Quickstarts for some select seasons to deal with the initial season results. That's about he best we can do.

4. Who atually knows anything about real manager tendencies? There's no real data on this. And it is also hard to differentiate between a manager's tendencies and the personnel he has available to him. Was Whitey Herzog crazy about SB's or was it the fact that he had guys like Vince Coleman, Ozzie Smith, and Wllie McGee on his teams?

5. There are no historical transaction files in Lahman. Attempting to put something like this into the game could cause significant issues. IRL sometimes teams will make a trade because one of their SP's gets injured and need to rteplace him...well what if in your sim season that's not what happened? Or what if the player that is supposed to be traded is actualy on the DL in the simulation? Or what if that player is a significant part of a 1st place team but gets dealt right at the trade deadline just because he "has to". This could cause all sorts of weird situations.

As for as-played lineups, the best we could do is have OOTP look at the PA/G of each player in the lineup and order the batting order in descending order. Leadoff men should have the highest PA/G, and the 8th hitter should have the least PA/G. That's about as close as you can get. Even that is an imperfect solution.

[Spritze]

Originally Posted by Garlon

3. We can make some Quickstarts for some select seasons to deal with the initial season results. That's about the best we can do.

No, that is not the best that can be done. If it were, game over.

Every consistantly reproducable incorrectancy can be programmatically fixed if there is sufficient programmer intention and consumer interest. As this bug affects 100% of all historical leagues created by 100% of all historical league players I would hope intention and interest might coincide.

[Garlon]

Spritze - Yes, you are right. There is another solution. I will propose two solutions for getting first year sim results more accurate.

[pstrickert]

Originally Posted by Garlon

Regarding pstrickert's initial post:

1. Yes, this was my biggest gripe with OOTP9 historicals.

2. We can work this out in beta-testing this year. There is only so much that we ca do in terms of reliever usage. Right now OOTP produces the correct distribution of IP between the SP's and bullpen for every season in history. Getting more pitchers into the games in historicals is actually good for a couple of reasons. #1 The AI alreaady knows which pitchers will perform the best (something that real life managers did not know) so they already get spots on the pitching staff while other pitchers who pitched poorly IRL end up in the minors or on the reserve list...this means that hitters in OOTP are always going to be facing higher quality relievers than they did IRL. #2 If we go a step further with this, then the best o the best will be pitching even more...and as a result this will drive up SP ERA's, which will of course lead to more losses or no decisions by real life SP's.

Having said that, we can still make some minor improvements with RPs.

3. We can make some Quickstarts for some select seasons to deal with the initial season results. That's about he best we can do.

4. Who atually knows anything about real manager tendencies? There's no real data on this. And it is also hard to differentiate between a manager's tendencies and the personnel he has available to him. Was Whitey Herzog crazy about SB's or was it the fact that he had guys like Vince Coleman, Ozzie Smith, and Wllie McGee on his teams?

5. There are no historical transaction files in Lahman. Attempting to put something like this into the game could cause significant issues. IRL sometimes teams will make a trade because one of their SP's gets injured and need to rteplace him...well what if in your sim season that's not what happened? Or what if the player that is supposed to be traded is actualy on the DL in the simulation? Or what if that player is a significant part of a 1st place team but gets dealt right at the trade deadline just because he "has to". This could cause all sorts of weird situations.

As for as-played lineups, the best we could do is have OOTP look at the PA/G of each player in the lineup and order the batting order in descending order. Leadoff men should have the highest PA/G, and the 8th hitter should have the least PA/G. That's about as close as you can get. Even that is an imperfect solution.

2. I'm not sure I follow your logic. Doesn't the same thing apply to batters? The AI knows which batters will perform better and, thus, will play them more.

3. Near the end of beta testing for OOTP9, Markus said he had a solution for this problem. Unfortunately, he did not have time to implement it.

4. Then we must be sure that the AI tailors its strategy to maximize the talent on its roster.

5. Numerous other baseball text sims utilize as-played files and transaction files. I hardly think it would be difficult for Markus to enable their use. The user would have to know, however, that using them would compromise the results of any long-term sims. But this would give the user another option for playing historicals. It just happens to be one of the most popular ways to play baseball text sims, too!

[Spritze]

Originally Posted by Garlon

Spritze - Yes, you are right. There is another solution. I will propose two solutions for getting first year sim results more accurate.

I got a solution myself to put forward when beta-testing begins. It uses the same process you have outlined for the regular OOTP modificators that work so well. I have extensively tested this system and it works for any initial game/league setup and for any year (1871 - 2008).

I simply create the OOTP game setup modifiers based on the spring training stats for the initial year of my game. It requires that spring training is activated of course but it works beautificently. It is a sort of spring training for OOTP as it were. After spring training but before the season starts I adjust the default 1.000 OOTP modificators in game setup and all is well. Since the modifiers are based on games played by a mix of the actual players for that particular year they fit right in with the standard modifiers and produce excellent results.

[pstrickert]

Do you use your spreadsheet to determine the LTM adjustments? (I'm referring to the spreadsheet you made available last year.)

[Spritze]

I altered it a bit but basically yes.

I take the spring rates and compare that to the OOTP rates in game setup and use the modifiers to make them match. I leave SP Endurance at 1 as Spring Training SP usage is different. RP Endurance is always 1 as it isn't used. The rest of the numbers are fair game.

[pstrickert]

I wouldn't mind taking a look at it. Could you PM me a copy? Or post it here? Thanks!

[Spritze]

Here you go.

I like to modificate in the ..............morning, I was gonna say morning, not nude.

[pstrickert]

Of course.

[Garlon]

Last year I suggested using the ST stats to calculate the modifiers for year 1. We can suggest it again.

[PhillieFever]

In case you guys are interested,I have a 1972 Phillies dynasty in the,... you guessed it dynasty section that is in the first year of the sim if you wanted to check your findings against it.All of the modifiers are default for 1972

[pstrickert]

Originally Posted by Spritze

I got a solution myself to put forward when beta-testing begins. It uses the same process you have outlined for the regular OOTP modificators that work so well. I have extensively tested this system and it works for any initial game/league setup and for any year (1871 - 2008).

I simply create the OOTP game setup modifiers based on the spring training stats for the initial year of my game. It requires that spring training is activated of course but it works beautificently. It is a sort of spring training for OOTP as it were. After spring training but before the season starts I adjust the default 1.000 OOTP modificators in game setup and all is well. Since the modifiers are based on games played by a mix of the actual players for that particular year they fit right in with the standard modifiers and produce excellent results.

For the life of me I cannot find the end-of-spring-training league stats. Where, pray tell, are they?

[Spritze]

You gotta add them up.

That is why having the program do it is imperative as the Markus feller does not have ST league totals anywhere I could find either. Just lists of detail per player/team.