Thanks contrarian

I broke this down a little differently, based on the order the sim does the calculation. I started with walks and hit by pitches, then hit or not, then home run or not, then single, double or triple:

	HBP or BB %	Batting Average	HR% given hit	1B% given non-HR hit	2B% given non-HR hit	3B% given non-HR hit
AFCS	0.06365637	0.250957521	0.079431148	0.769781462	0.162396383	0.067822155
Petco	0.072906404	0.226709175	0.043359375	0.709677419	0.155165374	0.135157207
.	.	.	.	.	.	.
Factor	0.873124531	1.10695794	1.831925578	1.08469206	1.046601951	0.501801988

higher walks in Petco is a bit odd, perhaps due to differences in the pitching staffs.

then we see a 10.7 percent bump in batting average for AFCS, and then almost twice as many home runs per hit.

then we get the interesting part. triples are twice as likely at Petco. and then, because P(3B) + P(2B) + P(1B) = 1 once you know there is a non-HR hit, P(2B) + P(1B) must be greater for AFCS than in Petco. we see most of that higher total is in P(1B) for AFCS.

Posted by contrarian23 on 4/19/2020 6:58:00 AM (view original):
This is fascinating to me...a hit in Petco is more likely to be for extra bases than a hit in AFCS. About 10% more likely.

As a reminder, here are the park adjustments for each location:

Park	1B	2B	3B	HR LF	HR RF	PF
AFCS	2	0	-2	3	3	1.12
Petco	-3	-3	2	-3	-3	0.82

this is mostly just due to the hitters involved being perfect fit for Petco, if you had Dave Kingman types instead you'd see way more XBHs at AFCS

The fewer the parks, the easier it is to read the contrast I would think. I would advise four at at time, but not 6. I will try to be in on the next version.

I am actually curious about some other parks - Fenway for example which continues to intrigue me (like the tooth that your tongue keeps poking to see if it still hurts maybe? :-)), the Polo Grounds, the Bakers Bowl, the Astrodome, Wrigley, Tiger Stadium.

The other problem with triples is that deadball guys get screwed on triples frequency. A guy like Lance Johnson, Willie Wilson or Curtis Granderson will always outperform guys like Owen Wilson, Roger Connor, etc. If you were to run this again for triples, try to get more modern day guys.

I once ran an 80M theme league using only players from the 1910's. I don't recall the exact numbers, but despite playing against contemporaries, triples were at around 35-40% of real life totals. The reason seems obvious to me. The deadball pitchers limit HRs allowed, but on the flip side, they should allow more triples than modern day pitchers - but they don't.

Posted by contrarian23 on 4/21/2020 11:14:00 AM (view original):
I deliberately chose players from a mix of eras...partially to get a read on exactly the point schwarze is making. We'll review each players stats at the end and see if we can draw any conclusions.

I will note that it is possible that some of the triples that the deadball era guys are losing COULD be becoming home runs, especially in AFCS. There are several Connor and Chief Wilson's on the HR leaderboards.

Is that even how the decision tree works, though?

Posted by schwarze on 4/22/2020 3:22:00 PM (view original):
The other problem with triples is that deadball guys get screwed on triples frequency. A guy like Lance Johnson, Willie Wilson or Curtis Granderson will always outperform guys like Owen Wilson, Roger Connor, etc. If you were to run this again for triples, try to get more modern day guys.

I once ran an 80M theme league using only players from the 1910's. I don't recall the exact numbers, but despite playing against contemporaries, triples were at around 35-40% of real life totals. The reason seems obvious to me. The deadball pitchers limit HRs allowed, but on the flip side, they should allow more triples than modern day pitchers - but they don't.

I think this is compounded this way because ERA is factored into non-HR XBH (i.e., doubles and triples) allowed, so these deadball pitchers have better HR rates, but instead of seeing higher XBH rates, they see lower XBH rates. All because of their ERA, which is also lower due to the era and not because they allowed fewer XBH, but because there were more errors committed on the field, and thus less earned runs.

Posted by just4me on 4/22/2020 5:08:00 PM (view original):

Posted by schwarze on 4/22/2020 3:22:00 PM (view original):
The other problem with triples is that deadball guys get screwed on triples frequency. A guy like Lance Johnson, Willie Wilson or Curtis Granderson will always outperform guys like Owen Wilson, Roger Connor, etc. If you were to run this again for triples, try to get more modern day guys.

I once ran an 80M theme league using only players from the 1910's. I don't recall the exact numbers, but despite playing against contemporaries, triples were at around 35-40% of real life totals. The reason seems obvious to me. The deadball pitchers limit HRs allowed, but on the flip side, they should allow more triples than modern day pitchers - but they don't.

I think this is compounded this way because ERA is factored into non-HR XBH (i.e., doubles and triples) allowed, so these deadball pitchers have better HR rates, but instead of seeing higher XBH rates, they see lower XBH rates. All because of their ERA, which is also lower due to the era and not because they allowed fewer XBH, but because there were more errors committed on the field, and thus less earned runs.

I know there was more data out there at one point on how they calculate 2B/3B allowed for pitchers, but they run/ran some kind of regression using ERA as a starting point with known information like H and HR allowed to act as bounds. This is all I could find at this point as a reference to that older discussion:

Posted by ozomatli on 4/22/2020 2:24:00 PM (view original):

Posted by contrarian23 on 4/21/2020 11:14:00 AM (view original):
I deliberately chose players from a mix of eras...partially to get a read on exactly the point schwarze is making. We'll review each players stats at the end and see if we can draw any conclusions.

I will note that it is possible that some of the triples that the deadball era guys are losing COULD be becoming home runs, especially in AFCS. There are several Connor and Chief Wilson's on the HR leaderboards.

Is that even how the decision tree works, though?

I think this slide would say yes:

That last slide would also indicate they don't use a ballpark multiplier for determining singles, just for hits, HR, and 2B/3B...

Posted by just4me on 4/23/2020 11:59:00 AM (view original):
That last slide would also indicate they don't use a ballpark multiplier for determining singles, just for hits, HR, and 2B/3B...

They can't use a multiplier for all three of singles, doubles and triples in that step, because the probabilities of the three have to sum to 1, and if you adjusted all three the probability wouldn't. So they adjust two of them and then the probability of a single ends up at 1 - Probability(double) - Probability(triple). They could just as easily use the multipliers on singles and triples and derive doubles, or singles and doubles and derive triples.

I think the singles rating for ballparks ends up very close to the factor in the hit or out step, but can sometimes vary in extreme cases like Dodger Stadium, where doubles and triples are suppressed so much that singles have to go way up, and end up higher than neutral even though the hit or out step factor is likely below 1.

Posted by 06gsp on 4/23/2020 2:46:00 PM (view original):

Posted by just4me on 4/23/2020 11:59:00 AM (view original):
That last slide would also indicate they don't use a ballpark multiplier for determining singles, just for hits, HR, and 2B/3B...

They can't use a multiplier for all three of singles, doubles and triples in that step, because the probabilities of the three have to sum to 1, and if you adjusted all three the probability wouldn't. So they adjust two of them and then the probability of a single ends up at 1 - Probability(double) - Probability(triple). They could just as easily use the multipliers on singles and triples and derive doubles, or singles and doubles and derive triples.

I think the singles rating for ballparks ends up very close to the factor in the hit or out step, but can sometimes vary in extreme cases like Dodger Stadium, where doubles and triples are suppressed so much that singles have to go way up, and end up higher than neutral even though the hit or out step factor is likely below 1.

I agree with you, and there’s another slide out there somewhere that shows the actual decimal multipliers that the +/- 1,2,34 actually represent (if I recall correctly a -3 is somewhere between .63-.7 and +4 is somewhere around 1.4+.

inshared that as there’s been more discussion lately around whether the +/- 1B rating on ballpark effects is on hits in general or just singles... that conversation is part of what this trial is supposed to answer be comparing ballparks at different extremes of 1B & 3B ratings and how those ratings effect hit rates and distribution.

I’ve always felt it was fairly clear in regards to the above...

you fools been here

twenny years here



way to go down swingin

I’m not sure if there’s a perfect way to test this, but I’ve always felt I hit more triples when the triples multiplier is more than the doubles multiplier, not less, or the same, as the doubles multiplier.

Like I’d hit more triples at say, Chase Field (+1, +1, +3) than at Municipal (+1, +3, +3) or Palace (+2, +4, +3)

Might be the case, but keep in mind all +3s are not created equal. A +3 could be +2.50000001 or +3.49999999.

Another way to look at it is that even in a season with a high level of triples, players still typically hit more doubles. Rollins, when he hit 20 triples had nearly 40 doubles. So, when the calculations are multiplying their 2B/H and 3B/H rates by the park factor, unless the triple rate is being significantly multiplied, the double rate will still be higher, which means more non-HR XBH will become doubles than triples even in a ++triples park. You’d need to have doubles reduced simultaneously (if the triple rates isn’t being significantly increased beyond the double rate) to have your players hit more triples than doubles, even as excellent triples hitters in a + triples park.