Above is from a presentation that Paul Bessire (tinmanPB, former Director Content and Quantitative Analysis at WhatIfSports) gave at a conference in sports simulation analytics.

@saintonian- this is indeed the perfect setup to this experiment, where you have an AL team with two 1.100 OPS guys instead of two .950 OPS guys, or even one 1.100 and one .950. (NL is harder to figure with no DH so it's a different theoretical calculation.) IMO your particular circumstance is the one that theoretically realizes + value, whereas the other two scenarios would not be.... Just an estimation on my part though, I'm sure it's possible to quantify the EXACT ops inflection point and it's probably in-between those two numbers

Posted by MikeT23 on 6/5/2017 3:44:00 PM (view original):
It probably is. PJF cannot fathom that he's wrong. What happened is people began failed C in RF without the bat to support it. Has to be an elite bat. You have one.

No no, we're on the same page with this, you're agreeing with me, we're agreeing with each other. He indeed has two of the bats. I'm arguing you yourself are using bats that don't support it, and you're ... agreeing with me. His bats are 1.100, yours are .950. I'm saying .950 isn't enough, based on rough estimations

Posted by loudawg10 on 6/5/2017 3:47:00 PM (view original):

wow dude nice find, been curious about this for a long time

Sadly, we're not agreeing. I've done the research. I won't get all the way into it, I hate the long-winded post, but we have all the data.

Hits, XBH, BB, etc, etc.
Errors, positive/negative plays, putouts, assists.
League averages for RF in both of the above.

It's not hard to figure out the break even point. I put together 5 seasons of data before I tried it. IIRC, in Moonlight Graham, I needed about .160-170 OPS+ from the "average" RF. And the average RF was just a tick over .700 OPS in MG.

The next extension of this question is whether it's better to play the guy in RF or get a real RF and play the guy at C, with the attendant increase in CERA from having a PC in the 30s.

Since my incumbent catcher is aging and headed for free agency, this might be a good thing to test next season.

The entire process is a plus/minus game as I've stated many times. If you have a great RF(and he still won't produce much in the way of defense) who strokes, there's no sense in giving up the extra outs. But a guy with 70 across the board D usually doesn't hit like a COF in HBD. And he doesn't do much on D.

Often it's a matter of "What else you got?" Because, if you're playing some gold glove RF who has a .587 OPS, you're better off putting a bad C with .800 OPS out there.

You're only using "Errors, positive/negative plays, putouts, assists" and not peripheral data that is not provided on the Team State page (Total Base volume of 2B and 3B that may or may not exist independently of +/- ratio based on lack of range.) I don't know that it does exist, just like you don't know that it doesn't exist. Experiencing the consequences of being wrong is worse than reaping the benefits of being right ... so it's a fundamental question of whether "Errors, positive/negative plays, putouts, assists" are all there is, or whether contextual data is hidden within individual box scores but not published within Team Stats

Did you assemble 5 seasons worth of individual box scores or 5 seasons worth of team stats data? Guessing the latter

Posted by saintonan on 6/5/2017 4:05:00 PM (view original):
The next extension of this question is whether it's better to play the guy in RF or get a real RF and play the guy at C, with the attendant increase in CERA from having a PC in the 30s.

Since my incumbent catcher is aging and headed for free agency, this might be a good thing to test next season.

Bingo, I like the way you think ... but probably better to play a real C at C than a real RF in RF

I also used a formula to determine how many runs an extra out an inning caused. I used MLB data for that. Assuming my RF f'd up 1 ball a game, 50 times a year, I wanted to know how many runs that would lead to.

Posted by MikeT23 on 6/5/2017 4:22:00 PM (view original):
I'm using sufficient data to "gamble" in a sim game.

I used a formula, (PO+A+POS PLAYS) - (ERRORS+NEG PLAYS) / INNINGS PLAYED x 1458
It gave me the POSITIVE PLAYER PER SEASON for any position. The average RF was around 215-220 every year.
The C in RF was really scattered but usually 170-200.
So the simple math was "Is this bat good enough to make up 50 extra outs?" Or, essentially, two games worth of runs?

This isn't rocket science. And, no, I did not use individual box scores.

Doesn't that formula double count plus plays, since presumably all ++ result in putouts? From my numbers +- are a lot less common. Along those lines it might be more useful to frame it in terms of total bases rather than raw net plays. Sadly, those 31 minus plays this season resulted in a lot more than 31 extra total bases for my opponents. Framing it in terms of TB would help separate infield and outfield defense and also account for the impacts of +- plays versus ++.

That is correct. I was trying to give as much credit as possible to the average RF(I think the record for plus plays in RF in most worlds is 20ish). I also assumed the "extra credit" would somewhat offset the negative play the C/RF would make resulting in extra bases.

For this exercise, I didn't need exact. I just needed good enough. After all, no player produces identical, or near identical, results season after season.