What's up with Park Factors?

[sircrayon]

I couldn't find anyone asking about this, but I feel it's a big change. The park factors in OOTP20 seem to have jumped many parks towards more HRs and hits, which I'm guessing is due to MLB trends. But the factors are also rather, well, lazy I guess I would say?

Take Miller Park in Milwaukee, for example. It's a known LHB paradise (looking at you, Yelich). Here are the factors in OOTP19:

AVG Overall: .987
AVG LHB: 1.020
AVG RHB: .970
Doubles: 1.000
Triples: 1.020
HR Overall: 1.114
HR LHB: 1.140
HR RHB: 1.100

Now let's look at the park factors in OOTP20:

AVG Overall: 1.011
AVG LHB: 1.011
AVG RHB: 1.011
Doubles: 1.000
Triples: 1.000
HR Overall: 1.011
HR LHB: 1.011
HR RHB: 1.011

Now in this case the Brewers actually had a decrease in HR overall, but look at the factors...no difference for handedness.

This is the case for a lot of stadiums, but some remain unchanged, like Yankee Stadium.

Was this intentional? If so, why do places like Miller Park and Wrigley Field get genericized while Yankee Stadium still has unique park factors for handedness?

[Sweed]

This was posted when the vXX first came out. It does look weird to my eyes too with a lot of factors hovering close to 1.000. Until someone comes up with some data showing this doesn't work I'll take their word that it was tested in development and produces "correct" results.

http://www.ootpdevelopments.com/boar...t=park+factors

[Reed]

This has been discussed in the past and you may be comparing apples to oranges.
Do a google search of OOTP forums park factors and you will find various threads on it.
Also, when you use park factors, do you use neutralized stats?

[GoPedro99]

I also have brought this up many times. My issue is not the park factors themselves necessarily, it is the ridiculousness that LH and RH factors are almost identically in all parks. You can not tell me a RH hitter in PNC park(410 LEft Centet gap) has the same HR factor as a LH hitter(relatively short RF fence)

[Sweed]

This is the reply from the thread I linked..

Originally Posted by pstrickert

Mike Garlon developed the formula for park factors. This is how he explained it:

I created all of the park factors used for OOTP. They are indeed based on the park effect on runs. What is done for each season and each park is take the park's 3-yr factor on runs and then create an appropriate increase or decrease in hits and HR that reproduce that park's impact on scoring runs. Each season this value needs to be recalculated using a quadratic equation because the distribution of singles, doubles, triples, and HRs and their impact on scoring changes each season in history.

I'm sure many think that it seems odd that the doubles and triples factors are always at 1.00, but actually when you change the BA factor to say 1.05 you are also raising doubles and triples by 5% as well simply because you are generating that many more hits overall, so we don't need an additional factor for doubles and triples as it is implied.

So basically it's one factor that is being used across BA/2B/3B/HR in order to replicate the scoring environment. If scoring is up say 10% in a given park, you need only boost all hits by a fraction of that amount, calculated through a quadratic equation, in order to reproduce the same scoring affect the park should have.

It says to PM Garlon if you have questions but wouldn't a discussion here be better for everyone?

I didn't look at this then, but with this thread, have taken another peek. I seems to me with this new method we no longer have "doubles" or "triples" parks. IE parks with big outfields where gappers are more common but homers are down due to longer distances? According to the explanation the doubles and triples will increase as the batting average does and that's all well and good but... doesn't that leave most (except the few outliers in vXX) parks with the same ratio of double and triples to hits? IE If doubles are 20% of hits then it's 20% of hits in most parks?

Isn't it the same too for rhb and lhb homers in most of the stadiums that have identical factors for both lhb and rhb as GoPedro99 brings up. Is this really correct?

Like I said in my first post, this must have been tested and found to work but it's such a drastic change more explanation would be nice. Too the eye it looks like it would work for correct totals but not for splits in HRs and extra base hits stadium to stadium?

[Sweed]

Double reply due to the "can't get to OOTP fourm" issue. Deleted.

[pstrickert]

PM Garlon and point him to this thread. I'm sure he'll be glad to respond to questions.

Garlon has, in fact, created an era_ballparks.txt file with discrete park factors. Then again, it is only for historical leagues. I don't know how park factors are determined for the current MLB season.

[Sweed]

Originally Posted by pstrickert

PM Garlon and point him to this thread. I'm sure he'll be glad to respond to questions.

Garlon has, in fact, created an era_ballparks.txt file with discrete park factors. Then again, it is only for historical leagues. I don't know how park factors are determined for the current MLB season.

Done

[Garlon]

The park factors that are in the game are based on the Quadratic formula in order to recreate the scoring environment of the park. These are based on 3yr factors and all of the statistics used can be found on Retrosheet and they do not have vs LHB/RHB data on there.

The 2B and 3B factors are always set to 1.000 with the quadratic factors, but you have to remember that if you increase BA by 3%, then 2B and 3B will increase by that much as well even if they are set to 1.000.

If you want a different set of park factors calculated a different way from the same data, then you can try this file and rename it to era_ballparks and place it in the My Documents>OOTP Baseball 20>database folder, but be sure to make a copy of the original game file before replacing it with this version. This file I am posting has discrete BA/2B/3B/HR factors.

If you are referring to starting a 2019 game in OOTP and not a historical game, then I cannot comment on those values that were put into that file as I had nothing to do with the creation of that.

[t-bone shuffle]

Trying to make sure I understand this information correctly.

Is the sole desire here to produce an approximately accurate run scoring environment for each ballpark (and hence league-wide)? Regardless of the statistical accuracy of each offensive category for each individual park?

Effect on individual player output? Should we expect, or not (for example) that an LH/fly ball/extreme pull hitter at Yankee Stadium (again, for example) would, or would not produce more homeruns than his actual statistical output at a different home field might?

Am I to understand that using a set of edited factors, the ballpark scoring environment cannot be expected to be accurate?

Previously, I understood that as long as each ballpark factor for the entire league totaled 1.0, the overall statistical outputs (league and stadium) should be reflective of the year/era, etc. Is this not, or no longer the case?

Are random factors like wind speed/direction, temperature having any affect on the outcome of a batted ball?

[Garlon]

We do not have this LHB/RHB data for parks, if we had the data then I'd have put that into the file. What we do have for all parks going all the way to 1871 in the Runs Scored per game at home and on the road for both the home team and their opponents. From this data and with the use of the quadratic formula the intent was to recreate the scoring environment for the parks.

We now have more discrete data for ballparks going back to 1906. So in the file I posted above, if you want the relative BA/2B/3B/HR factors for parks, which in theory will also produce the same scoring environment you can use that one as well.

You can set every park to 1.05 for HR and they will all actually play like 1.000 if you do this. You can invent your own ratings if you want. Where are you obtaining the data? How is it being formatted for use in OOTP?

Any impact that the wind and temperature had on the ballparks historically should have been accounted for in the stats and hence in the factors. Now, in OOTP here is also a weather file. I use a custom file because I believe that the weather file actually functions as another factor in the game.

As a rule of thumb, if you were to set BA/2B/3B/HR all to 1.05, or 5% above average, expect your scoring to go up at least double that, so more than 10%. So if you use edited factors you may not recreate the proper scoring environment if the factors are not correct for the game.

I think in general people make too much out of the park factors. You have to play half of your games on the road anyway, you can't have a home version and an away version of your team.

Let's consider team A whose park yields 20% more HR and let's consider a player on team B who hits 40 HR playing in a completely average ballpark who then gets moved to team A. Well, in league with 15 teams, those HR factors should add up to 15.000. Team A has a park with 1.200 for HR, that means the other 14 parks must equal the other 13.80. Ok, 13.80/14 = .985. This player now plays half of his games in 1.200 and half in .985. OK, so 1.200+.985 = 2.185 and 2.185/2 = 1.092. So how many HRs does this guy now expect to have this season? 1.092*40 = 43.68. So about 3 more HRs for the season. If he is a guy who hits 9 HRs per year, he might be expected to hit 10 with team A.

[Sweed]

First I want to thank Garlon for taking the time to come here to post a reply.

As for me having park factors that are based on hits, 2b, 3b HR adds personality to each park. So yes, while a guy may only hit 3 more HRs in a season, too me, it becomes more a bout where he hit those HRs. Just my 2 cents.

Originally Posted by Garlon

If you are referring to starting a 2019 game in OOTP and not a historical game, then I cannot comment on those values that were put into that file as I had nothing to do with the creation of that.

This I found interesting and it makes me wonder where the 2019 park factors came from?

I import my game into each new version and continue on. It has become habit for me to import each new versions factors into my game as it covers any new stadiums and adds some fluctuations to stadiums that already exist. I did that for vXX and that is where I first noticed how generic my Cubs Wrigley field had become but.. when I looked at Yankee stadium it still had LH and RH splits for the categories.

As I looked around my league it seemed some stadiums got the "historic ratings" with no LR splits and 2b-3b set at 1.000, while others retained their old splits. I wonder why? Mistake?

In any case my answer for my game was to manually import the park factors, one stadium at a time, from my v19 game. I am now again a happy camper with stadium splits.

I do hope this gets looked at for a fix (as noted there seems to be some of both types of stadiums in vXX), if needed in vXX, and so stadium splits return to v21, for the most current season, as I do like to update them version to version as stated.

[waittilnextyear]

If you guys want an alternative to the ballpark factors that come with the game, you can take a look at my thread linked here:

https://www.ootpdevelopments.com/boa...d.php?t=300927

If you like it, feel free to use it. If not, no hard feelings.

[t-bone shuffle]

First, likewise thanks very much for the response and your work on this.

As for this:

Originally Posted by Garlon

Let's consider team A whose park yields 20% more HR and let's consider a player on team B who hits 40 HR playing in a completely average ballpark who then gets moved to team A. Well, in league with 15 teams, those HR factors should add up to 15.000. Team A has a park with 1.200 for HR, that means the other 14 parks must equal the other 13.80. Ok, 13.80/14 = .985. This player now plays half of his games in 1.200 and half in .985. OK, so 1.200+.985 = 2.185 and 2.185/2 = 1.092. So how many HRs does this guy now expect to have this season? 1.092*40 = 43.68. So about 3 more HRs for the season. If he is a guy who hits 9 HRs per year, he might be expected to hit 10 with team A.

I understand the math behind this but, unless I'm missing something, this isn't accurate in, for example the current MLB divisional setup. Again, just for example purposes, consider the AL East and say that the combined HR factors for these 5 stadiums is 1.13. Wouldn't, for example the Yankees play 117 games in this environment (81 home/36 at 4 other AL East stadiums)? Wouldn't this turn your 40 HR player into an approximately 47 HR player,(+/- depending upon the home stadium)?

Regarding the change in Park Factors as noted in the OP, I can say with certainty that there has been some type of change. I'm currently playing a game that originated with version 15/2014 MLB roster set. As an example the 2B/3B factors at Camden Yards are .990/.880 respectively. I can only assume that the 3 season aggregate was not being used in 15?

Not sure what else to add here other than that it feels as though I/we(?) were misunderstanding the influence of Park Factors. Or that the game engine has moved toward a place where Park Factors are less of a factor?

Thanks for the discussion.

[Sweed]

Originally Posted by t-bone shuffle

First, likewise thanks very much for the response and your work on this.

As for this:



I understand the math behind this but, unless I'm missing something, this isn't accurate in, for example the current MLB divisional setup. Again, just for example purposes, consider the AL East and say that the combined HR factors for these 5 stadiums is 1.13. Wouldn't, for example the Yankees play 117 games in this environment (81 home/36 at 4 other AL East stadiums)? Wouldn't this turn your 40 HR player into an approximately 47 HR player,(+/- depending upon the home stadium)?

Regarding the change in Park Factors as noted in the OP, I can say with certainty that there has been some type of change. I'm currently playing a game that originated with version 15/2014 MLB roster set. As an example the 2B/3B factors at Camden Yards are .990/.880 respectively. I can only assume that the 3 season aggregate was not being used in 15?

Not sure what else to add here other than that it feels as though I/we(?) were misunderstanding the influence of Park Factors. Or that the game engine has moved toward a place where Park Factors are less of a factor?

Thanks for the discussion.

Hope you don't mind my commenting...

First I hadn't thought about those playing in divisions with multiple parks that have higher or lower than 1.000 parks, good point.

Too my eye nothing has changed with the park factors as we know them.
With the "new" way it has become a more generic across the league setting (Wrigley is 1.002 in every category?)where things should even out. In other words I don't think this is any different than playing a league where all factors had to add to 1.000 as you mentioned in your earlier post. It seems to me to be a different path to the same end, IE proper statistics.

My problem is it gets there in a generic way in that it doesn't matter which park the LHB 40 hr guy hits his homers, he's still going to get his +\- a few?

Over the coarse of a long running game maybe this doesn't matter statistically but, as I noted before, it matters to me as IMHO the parks have lost their individualized atmosphere. What's the point of looking forward to a trip to Fenway if righties don't hammer the Monster while lefties have long shots to the RF gap die in a fielder's glove? Losing handedness IMHO is huge in taking away the immersion factor of the game.

Maybe I'm over simplifying it?

As Garlin noted he did not do the 2019 park factors but, in my game, they seem to have been mixed in randomly. Wrigley is nerfed at 1.002 across the board and like all others done in this way has no l\L\R splits. While other parks still show splits and different modifiers in each category. I have to think this was a mistake? Surely they should have splits in the modern 2019 game?

In any case, as noted, I've returned to my factors from v19 through manual import and am happy to have found this discussion before starting my newest season.

[endgame]

Know you're talking primarily historical, but this has its place there, too. This a ten-year-old, or greater, request, and it still stands. It's easy to say that ballpark factors are given too much weight or attention, but for those who are interested in them, they provide both a reference point and a kind of immersive quality that adds an additional element of consideration. The format attached is perfect for that kind of review.

Originally Posted by endgame

Suggestion:
Add Park Factor histories available by team or by ballpark. This wouldn't require any more than last season and the history over the last two to three seasons. I'll try to attach a sample from Scouting Notebook. Include, as pictured, how this park ranked in the league. *

Reason
Enhance the experience, history, and personalization of the ballpark, making it even more evident (as goes stats) how the park treats pitchers and hitters alike, including error likelihood. Brings about a real sensation of what kind of #'s are reflected by the park factors.

Priority
High. IMHO, a feature that would really add another new dimension to the game.

[t-bone shuffle]

Originally Posted by Sweed

Hope you don't mind my commenting...

First I hadn't thought about those playing in divisions with multiple parks that have higher or lower than 1.000 parks, good point.

Too my eye nothing has changed with the park factors as we know them.
With the "new" way it has become a more generic across the league setting (Wrigley is 1.002 in every category?)where things should even out. In other words I don't think this is any different than playing a league where all factors had to add to 1.000 as you mentioned in your earlier post. It seems to me to be a different path to the same end, IE proper statistics.

My problem is it gets there in a generic way in that it doesn't matter which park the LHB 40 hr guy hits his homers, he's still going to get his +\- a few?

Over the coarse of a long running game maybe this doesn't matter statistically but, as I noted before, it matters to me as IMHO the parks have lost their individualized atmosphere. What's the point of looking forward to a trip to Fenway if righties don't hammer the Monster while lefties have long shots to the RF gap die in a fielder's glove? Losing handedness IMHO is huge in taking away the immersion factor of the game.

Maybe I'm over simplifying it?

As Garlin noted he did not do the 2019 park factors but, in my game, they seem to have been mixed in randomly. Wrigley is nerfed at 1.002 across the board and like all others done in this way has no l\L\R splits. While other parks still show splits and different modifiers in each category. I have to think this was a mistake? Surely they should have splits in the modern 2019 game?

In any case, as noted, I've returned to my factors from v19 through manual import and am happy to have found this discussion before starting my newest season.

I'm with you on all of the above.

[Saul Goodman]

Would be good to get an explanation on why the park factors have seemingly been normalised on the 2019 start compared to previous years. I'd flip back to the 2019 park factors, but worried that something under the hood has changed which has forced them to make park factor changes to how they are now.

[Reed]

Just curious, those of you that use park factors, do you use neutralized stats for players? If not, then you are double weighting park factors.
I guess it does not matter if you are playing fictional.

[Sweed]

Originally Posted by Reed

Just curious, those of you that use park factors, do you use neutralized stats for players? If not, then you are double weighting park factors.
I guess it does not matter if you are playing fictional.

I've imported my game from version to version since v4 (2002) and have used whatever default is so.... I assume I have not neutralized stats. So by attrition my league is now totally fictional.

In any case to my mind I've never seen this double weighting as a real issue, or at least that doesn't have a major impact though I've done no statistical studies.

I like my stats, I'm good with math, but am not a real statistician, like some here, so maybe I'm wrong but here is my thinking. All of the parks operate in the same environment and that environment is heavily controlled by League Total Modifiers that provide "realistic" stats all within ~2-3% of my desired League Totals.

So with all parks operating in this "double" environment, statistically controlled with LTMs, why does double weighting matter? Isn't it a net zero difference? Have I over simplified it in my mind?