What about historical fielding?

[Charlie Hough]

If the real issue is that good fielders in OOTP are having a disproportionate effect on offense, defense or stat distribution, to the point that overall league output or core individual player stats are far removed from real life, then I think those stats should probably be the focus rather than total zone runs.

Markus and company will probably be moved by discrepancies in core league totals or primary individual stats rather than sabermetrics such as total zone runs.

There is so much arbitrariness in how total zone runs are calculated, especially for most of baseball history, that I couldn't imagine using it as a basis for measuring real fielding performance or comparing real life data to OOTP results. Here's the definition and explanation of the methodology provided by Baseball Reference:
https://www.baseball-reference.com/a...tal_zone.shtml

Do we know what method OOTP is using to calculate total zone runs and whether that's equivalent to the different methods used to calculate this stat for players before 1989 vs. the different methods and inputs used for more recent seasons? My guess is that OOTP is using only one method, whereas Baseball Reference is using multiple methods based on the year and available data, as explained in the definition, and that could be the source of the discrepancy right there.

If not, I still think the focus needs to be on the actual discrepancies in overall league output and the distribution of core individual player stats rather than sabermetrics such as total zone runs. For example, if good fielders are committing far too few errors while those errors are all getting dumped on other players, then where are those numbers? If good fielders are having too much influence on overall results, then where are the stats showing that overall league batting averages are too low or that a batting average against a specific team is way too low?

I've been away for a few years, so my apologies if people have been discussing and presenting evidence in these other stats. But I think that's where the hope lies in convincing anyone of anything.

[Garlon]

TH overall league output is controlled by the league totals modifiers. The good fielders are disproportionately good and help their team in run prevention far more than they should be able to do so. There is also a defensive efficiency statistic in OOTP that tracks player efficiency on plays they are making. For Cal Ripken who is at the top of that list for SS his efficiency was 1.20, meaning he made 20% more plays than the average SS. That means with over 500 assists he made more than 100 plays above average. With the average SS play being worth about .75 runs that also turns out to the 75 runs saved he has that season.

The only calculation that I think they have wrong for runs saved might be for catchers. I think they are undervaluing errors, passed balls, and caught stealing for some reason.

The conversion of additional plays made to runs saved for the other positions seems to be accurate.

The average player at a position with save 0 runs above average. When you have these players putting up impossible values that means there must be many other players performing significantly below average. The defensive efficiency for players gives you good information for translating the player performance. For infielders thus refers to assists and for outfielders this refers to put outs. So if an outfielder has 300 PO and has an efficiency of 1.05, you take 300/1.05 = 286, and this means the average outfielder was only expected to make 286 of those plays, so this player is 14 plays above average. Outfielder plays are worth about 0.50 runs and infielder plays are worth about 0.75 runs roughly.

[Charlie Hough]

I understand that you're seeing sabermetric stats that seem to indicate good fielders are disproportionately good. But where are the corresponding stats to show that weaker fielders are disproportionately bad? Without those numbers or some indication that this is skewing the distribution of league totals in some way, I suspect you'll have a hard time convincing Markus to change anything. At the end of the day, if individual stats and league totals are still coming out reasonably well despite this, it's probably going to be a hard sell.

[Lukas Berger]

Originally Posted by Charlie Hough

If the real issue is that good fielders in OOTP are having a disproportionate effect on offense, defense or stat distribution, to the point that overall league output or core individual player stats are far removed from real life, then I think those stats should probably be the focus rather than total zone runs.

Markus and company will probably be moved by discrepancies in core league totals or primary individual stats rather than sabermetrics such as total zone runs....But I think that's where the hope lies in convincing anyone of anything.

What Charlie is saying above basically exactly matches my thoughts. What I'm reading here (and what I recall reading from Garlon on this when he's discussed this in the past) seems to be saying simply that we're calculating fielding results differently than Total Zone is and we have more players on the high end of the runs prevented curve than they do.

Which even if true, my question would be, so what?

Total Zone isn't some sort of gospel fielding metric that any other metric should be benchmarked against. In fact, iirc some of what I've read about it in the past, it's kind of a relatively crappy metric in comparison to some others.

So all Garlon's really pointing out is that OOTP is measuring fielding differently than one other specific (and flawed) metric is. Which has nothing to do with anything. It's irrelevant.

Every individual metric measures things differently. For example, WAR is different from Baseball-Ref to Fangraphs to OOTP to anyone else who calculates WAR. Zone Rating in OOTP being different than Total Zone is not in and of itself a problem any more than the different WAR's are a problem.

To convince us to make any changes, you'd first have to prove to us that Total Zone, calculated as you're referencing it, is significantly better and more accurate at measuring the real impact of fielding than OOTP is to the point where we should be using it as a benchmark. And/or that the way we're calculating things now is screwing up the results in-game.

Right now in all of the above, the only point I can really see being made is that non-OOTP Total Zone is different, not that it's better. Which doesn't move me to recommend changes and it definitely won't move Markus to make them.

[joefromchicago]

How does OOTP measure total zone runs? Is it the same as how Baseball Ref does it? I ask because I can't figure out how anybody measures it

[Lukas Berger]

Originally Posted by joefromchicago

How does OOTP measure total zone runs? Is it the same as how Baseball Ref does it? I ask because I can't figure out how anybody measures it

I assume not, but I do not know for sure. Markus and Matt are probably the only ones that do.

[Garlon]

This is a summary of the 1984 results for SS from the statistics previously posted in this thread.

What is good is that OOTP does an excellent job of keeping track of the fielding results. The sum of the team SS Zone Rating for the entire league comes out to 0, which is correct.

In the game, SS made 12856 assists, and in MLB that season they made 13126 assists, so this is very good too as it means there is a very realistic distribution of balls being hit toward the SS position.

The average +/- play in OOTP at SS is worth approximately 0.50 runs. It seems that OOTP is also basing the individual play value on both the difficulty of the play (what % of the time does the average SS make that given play) and the base-out situation as well. So a difficult missed play with 2 outs and nobody on base is different than an easy missed play with bases loaded and 2 outs. So I give them big credit in this area as well as this is about all that you can possibly ask for when tracking fielding results. Excellent work.

When you start looking at team and player results, things are not working well. We see that Cal Ripken has an efficiency rating of 1.299, he made an extra 155 more plays than the average SS given those same opportunities and he saved 75 runs for the season. All of that is impossible. I mentioned previously that an efficiency rating for a SS above 1.10 is very extreme at the major league level. We also see that there are players like Ivan de Jesus at .738 efficiency and Jackie Gutierrez at .770 efficiency and there is no way that any major league SS could ever perform that poorly. For reference, Jeter is the worst SS ever by far, and he was maybe a 0.90 efficiency.

If you look at the Orioles and the Red Sox you see that the difference alone in SS defense in OOTP puts them at 125 runs apart for the season, with the average win that season requiring 8.52 runs, the defense from SS alone puts these two teams over 14 wins apart. These two teams were the best and the worst at SS that season, but the difference between them was about 50 runs, so OOTP is off by a very wide margin here. This is not a matter of a slight difference in the way the ZR calculation is being made, there was a difference of 301 team SS assists in the game that season, the real difference in assists that season was 165. So we know OOTP was off by an additional 136 assists, and if OOTP is valuing these on average at 0.50 runs, then that is about 68 runs scored and if we add that to the difference of 50 that these two teams really had that gets us to 118, which is very close to the 125 we see in OOTP that season.

Ripken in OOTP made +155 assists at SS that season, but realistically how many could he have made? I made a database of fielding for 1871-2019 that takes into account Team Defensive Efficiency, Team Pitcher Strikeouts, Team Caught Stealing, and Team Groundball Percentage in order to determine how many plays above and below average each player made. Since historical team defensive efficiency is known, team +/- plays can also be determined for every season. After accounting for these variables, I have Ripken at an astounding +53 assists that season. If we value the assist at 0.50 runs, the same as OOTP, that gives us +26.5 runs saved, and on BBR they have Ripken at +23 runs saved that season.

[Lukas Berger]

Originally Posted by Garlon

Ripken in OOTP made +155 assists at SS that season, but realistically how many could he have made? I made a database of fielding for 1871-2019 that takes into account Team Defensive Efficiency, Team Pitcher Strikeouts, Team Caught Stealing, and Team Groundball Percentage in order to determine how many plays above and below average each player made. Since historical team defensive efficiency is known, team +/- plays can also be determined for every season. After accounting for these variables, I have Ripken at an astounding +53 assists that season. If we value the assist at 0.50 runs, the same as OOTP, that gives us +26.5 runs saved, and on BBR they have Ripken at +23 runs saved that season.

Ok, this is interesting and it's a start.

That being said, this is an extreme outlier season in OOTP based on the info you posted, and I don't know how much it tells us to simply look at one or two of the biggest outliers.

It's kind of like if in a sim Babe Ruth hits 80 HR's in a season that he actually hit 55 in reality, while the worst regular HR hitter hits 25% less than he did. Unrealistic yes, but possible and unavoidable given random variance and the fact that we're not fudging the numbers to prevent outliers.

So while it's somewhat unrealistic by one definition, it's not imho in and of itself proof of something being broken.

What happens when you do a similar breakdown for one of the more 'typical' high end seasons in OOTP, with a ZR of more like 50-55, one of the Belanger seasons for example?

How many 'extra' assists is OOTP giving the league leader in those seasons compared to actual league leader in the same season?

[Lukas Berger]

I'm also more interested in the raw numbers, rather than the numbers in comparison with your db. Because comparing to your db is really just switching the comparison from OOTP vs Total Zone to OOTP vs 'the Garlon metric'. And if you built your db with a similar methodology to Total Zone, than it only makes sense that OOTP won't match up as well with your metric either. But that's not an indication OOTP is wrong, just an indication of different methodologies. I could basically copy/paste what I said above here again, just replacing 'Total Zone' with 'The Garlon Metric'

So what about the raw numbers for the season by season numbers for top level SS defenders in OOTP like Ripken or Ozzie or whoever. How many raw assists/chances are you seeing for them season by season over the course of a career in comparison with their real-life raw assist/chances numbers?

And what about the raw numbers in OOTP for low-end defenders, like Jeter or whoever?

[Lukas Berger]

I'm also curious what are the raw numbers of assists/chances for Ripken in the outlier season you're quoting above as compared to real-life?

[Garlon]

Player, Career Assists OOTP vs Real, Career Innings Played vs Real Career Innings Played

Ripken: 7981 vs 6977; 20592 vs 20232

Smith: 9304 vs 8375; 22324 vs 21785

Belanger: 6148 vs 5786; 15215 vs 15334

Aparicio: 8522 vs 8016; 22643 vs 22407

Vizquel: 8412 vs 7678; 23424 vs 22960

Dent: 4887 vs 4332; 12010 vs 11758

Tulowitzki: 4407 vs 3889; 11319 vs 10945

[Lukas Berger]

Originally Posted by Garlon

Player, Career Assists OOTP vs Real, Career Innings Played vs Real Career Innings Played...

Hmm, interesting. Looks like we are indeed on the high side here.

I'd like to see a full list though and also see what the year by year numbers look like as well. Can you upload the .dat files for the whole league so I can look it over a bit more closely?

[Garlon]

Ripken

[Garlon]

The best defensive SS are twice as far from average as they should be and the worst are twice as far from average as they should be and this holds for any season. I can try to send you the files, but I am not sure exactly how to locate the exact files you need. You can also try creating your own league in 1901 and use real lineups and transactions . Make sure to have players retire according to history enabled too. Then have OOTP simulate from 1901-2019 and you will see these results too. The defense is off by such a wide margin at the positions that it really ruins the game. The import process needs to be revised and minimum and maximum values for each of the positions used. For example, at SS no player should have less than 118 range or arm ratings, and nobody should be higher than 198. Put the league average for both range and arm for SS at 158. This provides +/-40 points on the scale above and below average. That is all that is required to produce realistic results. I did this when creating the Baseball Greats mod on the Workshop. I found that going any lower than this produces unacceptable results. Even using 118-198, the bulk of all players should be between 138-178 for any season.

[Lukas Berger]

Originally Posted by Garlon

I can try to send you the files, but I am not sure exactly how to locate the exact files you need. You can also try creating your own league in 1901 and use real lineups and transactions .

I would just need the files that end in .dat from the .lg folder of your OOTP saved_games directory.

If it's a problem, don't worry, I can run some sims myself. Was just hoping to save a bit of time if possible!

[David Watts]

Bump

[Rain King]

I know why you were searching for these threads, but I just want to make sure you understand that Garlon was largely involved in the revamping of Historical defensive ratings before (I believe) OOTP 22.

So, most of his ideas have likely been implemented at this point.

[David Watts]

Originally Posted by Rain King

I know why you were searching for these threads, but I just want to make sure you understand that Garlon was largely involved in the revamping of Historical defensive ratings before (I believe) OOTP 22.

So, most of his ideas have likely been implemented at this point.

I bumped them so that Actionjackson could easily find them. Action wasn't as active on the boards when Garlon made these posts.