Park Effects ARE Broken

[mrbill]

Even though this topic got beaten up last time and closed, I'm reposting because I'd like confirmation that this bug is recorded. I'd love to see a patch to make 6.51 and 6.12 work (because its been a bug since 6.12 and not everyone dropped more money on 6.5), I think we're owed that.

It's disappointing this feature no longer works, its one of the most basic things OOTP lets you do, and somehow it slipped by for 2 years.

[mad0die]

Well, somebody had to close that last thread so unless they totally skipped the 1st page they should be aware of the issue now.

[mrbill]

So many comments come to mind that would make this thread worthy of closing too

Closing the thread isn't acknowledging the bug is there and needs to be fixed.

But, here's a chance for customer service to succeed, so I'll see what happens.

[A.J. Holz]

Yes, since this is a bug that affects every OOTP league, online or offline, please fix it.

[ihatenames]

Any chance this is a sample size issue. There is a fair bit of randomness in ootp?

[andymac]

I ran a quick test, but I'm either blind or can't find anywhere that lists the home/road splits for a team as a whole. Anybody know where I can find that instead of having to add up all the players numbers individually?

[Eckstein 4 Prez]

I'm sure there's not going to be another patch for 6.5, but this is a pretty shocking issue. I hope whatever causes it can at least be avoided for V2006.

[andymac]

Originally Posted by Eckstein 4 Prez

I'm sure there's not going to be another patch for 6.5, but this is a pretty shocking issue. I hope whatever causes it can at least be avoided for V2006.

I actually don't think the park effects are broken...my test stats will be up in a few minutes.

[gubaru]

I ran a few seasons as part of a test a while ago to look at park factors. I used the Total Baseball formula for batter and pitcher park factors and the component factors in the game seemed to be eye candy. I had planned to run 25 seasons but the results were so random I gave up before I was finished. I can see if I can dig up the spreadsheet I had everything in if anyone's interested in seeing it.

[andymac]

Here we go...I made a 2 team leage. Team 1 had all park factors set to 125, Team 2 had all Park factors set to 75. I took the top 8 players in at-bats for each team and listed their stats below.

Team 1 (All Park Factors: 125)

		Home			Road
Batter 1	.266-.343-.477		.222-.322-.304
Batter 2	.265-.293-.435		.204-.255-.265
Batter 3	.239-.314-.379		.241-.312-.368
Batter 4	.316-.370-.504		.256-.334-.391
Batter 5	.286-.396-.451		.312-.405-.436
Batter 6	.229-.276-.427		.227-.266-.326
Batter 7	.248-.333-.440		.222-.273-.335
Batter 8	.234-.336-.453		.216-.294-.334




Team 2 (All Park Factors: 75)

		Home			Road
Batter 1	.294-.337-.401		.305-.354-.511
Batter 2	.303-.343-.382		.272-.313-.490
Batter 3	.307-.353-.436		.319-.371-.507
Batter 4	.317-.405-.366		.255-.342-.386
Batter 5	.246-.355-.470		.346-.437-.692
Batter 6	.267-.320-.342		.352-.407-.566
Batter 7	.255-.306-.431		.287-.351-.498
Batter 8	.246-.321-.402		.272-.342-.346

[andymac]

It seems to me that the park effects may have a much stronger impact on power numbers than batting average numbers, but that is the way it should be IMO. From the above test, though, it sure seems like they work to me.

[disposableheros]

i wonder if there is an invisible cutoff, ie. 200 and 50 are too far out so the game freaks out and doesnt recognize them, but 125 and 75 are not.

[Eckstein 4 Prez]

IMO the best way to test would be to create a test league, set it to "replay," and then have all the teams play in the same park. Then run a series of seasons with the factors set to 50, 75, 100, 125, etc., and see if the total stats of the league change, and by how much.

I'd do this myself, but I'm going to spend the evening importing and editing the 1881 rookies for my solo league. So... uh, yeah. I'll let someone else get all over that.

[andymac]

Originally Posted by Eckstein 4 Prez

IMO the best way to test would be to create a test league, set it to "replay," and then have all the teams play in the same park. Then run a series of seasons with the factors set to 50, 75, 100, 125, etc., and see if the total stats of the league change, and by how much.

I'd do this myself, but I'm going to spend the evening importing and editing the 1881 rookies for my solo league. So... uh, yeah. I'll let someone else get all over that.

OK, I'm working on this now.

[gubaru]

Originally Posted by Eckstein 4 Prez

IMO the best way to test would be to create a test league, set it to "replay," and then have all the teams play in the same park. Then run a series of seasons with the factors set to 50, 75, 100, 125, etc., and see if the total stats of the league change, and by how much.

I'd do this myself, but I'm going to spend the evening importing and editing the 1881 rookies for my solo league. So... uh, yeah. I'll let someone else get all over that.

Wouldn't it just follow whatever the league totals are? I assume that OotP's stats engine would look there first and then use the component park factors to adjust up or down for the teams/players. Then again, I could be missing something.

[andymac]

Originally Posted by gubaru

Wouldn't it just follow whatever the league totals are? I assume that OotP's stats engine would look there first and then use the component park factors to adjust up or down for the teams/players. Then again, I could be missing something.

I am pretty sure that the league totals are just the "base". The park factors are applied 2nd (so to speak), so they should definitely show up in the numbers. This is why it is important for all of your park factors to balance out when looking for specific league numbers.

[mrbill]

From my first tests on this, HR, 2B and 3B all had their effects, although I saw 2B and 3B having over-amplified effects, but still probably within the range of acceptable with high assumed variance.

Average was the one I found no compelling evidence for.

With a 200 park rating in average, I expected a significant increase in hits, and instead saw none at all over 162 home games for the two sample teams (thats just shy of 6000 AB).

6000 AB with about .001 difference is sample size enough, when I expected about a .100 point BA difference with a 200 park rating.

I'd be a little concerned about changing all the park factors at once, because an increase in HR, 2B and 3B will inflate average, so isolating whether the "Avg" park factor is broken will be harder.

[mrbill]

Originally Posted by disposableheros

i wonder if there is an invisible cutoff, ie. 200 and 50 are too far out so the game freaks out and doesnt recognize them, but 125 and 75 are not.

The text box where you set park effect ratings specifically says 1-200 is the acceptable range, so I made sure to stay within specified boundaries.

I wouldnt try and go above 200, but that should work.

[gmo]

I assume everyone is using the new DIPS engine for their tests. Is that correct?

I would be interested in seeing what happens if these experiments were repeated using the old engine since one hypothesis could be that AVG is not being influenced because in the new engine the path to the result is running through BABIP which may not being get influenced by the park RAvg & LAvg settings.

[andymac]

Originally Posted by gmo

I assume everyone is using the new DIPS engine for their tests. Is that correct?

I would be interested in seeing what happens if these experiments were repeated using the old engine since one hypothesis could be that AVG is not being influenced because in the new engine the path to the result is running through BABIP which may not being get influenced by the park RAvg & LAvg settings.

I will do both.