DIPS, Defense, and 1974: A Case Study

[pstrickert]

Originally Posted by swampdragon

So in general there's a 20 point spread. That's an acceptable amount. If they all looked like Baltimore's, I'd think we were very close to getting it right.

IRL in 1974, BABIP in the AL ranged from .269 - .292. That's 23 points between the highest and lowest team BABIP. So, in your view, it doesn't matter where a team falls within that range in OOTP, since it'll still be within about 20 points of real life. The team with the highest team BABIP in real life could have the lowest team BABIP in OOTP (and vice-versa) . . . but that's no problem. That's close enough. I guess you're easy to please.

[swampdragon]

Originally Posted by pstrickert

IRL in 1974, BABIP in the AL ranged from .269 - .292. That's 23 points between the highest and lowest team BABIP. So, in your view, it doesn't matter where a team falls within that range in OOTP, since it'll still be within about 20 points of real life. The team with the highest team BABIP in real life could have the lowest team BABIP in OOTP (and vice-versa) . . . but that's no problem. That's close enough. I guess you're easy to please.

Careful. What I said was that if they were all like Baltimore's, I'd feel pretty good. That's 10 either way, not 20. DIPS theory postulates that some of the variation is simply luck. If this is offensive BABIP you're talking about, we're not going to improve those numbers by making defensive imports better. If you run the simulation on Baltimore 100 times, presumably you'd get a bell curve which would be centered on the right number. That's exactly what you want. The right expectation coupled with enough variability to make life interesting.

[Scruff]

This is an interesting thread.

I slow skimmed it, and don't think I saw this mentioned, but in order to use something like the Lahman database to come up with accurate fielding ratings, you are most likely going to have to start with the real life major league team and work your way backwards to the individuals.

This is what Bill James does with Win Shares. I don't think you would have to import players to their real life teams to do this. You can create the players ratings based on his real life team within the database and then import the player into a draft, or whatever after the ratings have been created, right?

For the Hall of Merit project at baseballthinkfactory.org, Dan Rosenheck has created a spreadsheet that basically rates every player from 1901 on for batting, base-running (not SB), fielding, pitching wins. The ratings he's come up with correlate pretty well with the current play-by-play metrics for modern players and he then applies the formulas back for the years where we don't have play-by-play. I would think even just using that spreadsheet as a jump-off point would help, heck the ratings could be directly imported into Lahman by adding a column or two (I believe he has a column that adjusts them based on playing time as well). It might be tougher for guys that played multiple positions, but I think it could be done.

But if you aren't using PBP data, you pretty much have to work backwards from the team, or it's not going to work.

[pstrickert]

Originally Posted by swampdragon

Careful. What I said was that if they were all like Baltimore's, I'd feel pretty good. That's 10 either way, not 20. DIPS theory postulates that some of the variation is simply luck. If this is offensive BABIP you're talking about, we're not going to improve those numbers by making defensive imports better. If you run the simulation on Baltimore 100 times, presumably you'd get a bell curve which would be centered on the right number. That's exactly what you want. The right expectation coupled with enough variability to make life interesting.

1. I'm not sure I follow. "10 either way" is a spread of 20, isn't it?

2. By team BABIP, I mean the BABIP for the team's pitching staff. Team defense must factor into that, right?

3. The past two years, I've simmed the '74 season probably 100 times. Did I check the BABIP stats each time? Of course not. But I have noticed that Baltimore's defensive stats rarely match up with their real-life defensive stats. I think it's no coincidence that the BABIP for Baltimore's pitching staff also rarely (if ever) approaches their real-life BABIP.

[Garlon]

I edited the DB and gave all players a default rating at their positions so that all 2bmen have the same defensive rating, all OF's have the same ratings, etc when using base fielding ratings on Career. I found that pitcher BABIP against were basically right on target with their real life stats.

This suggests that right now the spread of fielding ratings in OOTP historicals is too great. On the 100 scale we probably need a spread of no more than 20 points between the best and worst defenders at a position.

[Spritze]

Originally Posted by Garlon

On the 100 scale we probably need a spread of no more than 20 points between the best and worst defenders at a position.

Done

I was bored.

[RchW]

Originally Posted by Garlon

I edited the DB and gave all players a default rating at their positions so that all 2bmen have the same defensive rating, all OF's have the same ratings, etc when using base fielding ratings on Career. I found that pitcher BABIP against were basically right on target with their real life stats.

This suggests that right now the spread of fielding ratings in OOTP historicals is too great. On the 100 scale we probably need a spread of no more than 20 points between the best and worst defenders at a position.

That's a very interesting result. It supports a contention I've made for several version now that reserve middle infielders and catchers by definition should have good defensive skills (ratings). IRL there is no other reason for being there. If they could hit they would start. At 1B and 3B you can see it going both ways.

In the OF there is logic in keeping 1 bad fielder (AL excepted) who can hit but unless you have a true platoon situation reserve players don't start 'cause they can't hit.

[Scruff]

Originally Posted by Spritze

Done

I was bored.

Does this mean if we use the neutralized DB to import players today we'll have this update? Do I have to do anything to get it? Very, very new to OOTP, so want to make sure I'm doing this right, as I'm just now planning to start a universe from 1871 . . . thanks for any help.

[Scruff]

Originally Posted by RchW

That's a very interesting result. It supports a contention I've made for several version now that reserve middle infielders and catchers by definition should have good defensive skills (ratings). IRL there is no other reason for being there. If they could hit they would start. At 1B and 3B you can see it going both ways.

In the OF there is logic in keeping 1 bad fielder (AL excepted) who can hit but unless you have a true platoon situation reserve players don't start 'cause they can't hit.

There are always exceptions, but in general this is true.

You don't have players that are replacement level hitters and replacement level fielders. Well, you do, but they are called AA players.

Typically replacement level hitters at a position are at least average fielders or they wouldn't have jobs.