So, this discussion concludes I should  continue to place a fairly high value on solid defense. Thanks, I doubt I could force myself to change anyway, I'm that old horse smoking something there at the water trough.

Posted by topoftheworl on 7/31/2014 5:46:00 PM (view original):
That's a lot of outs.  My back of the envelope calculation is that you would need about 200 points of OPS to make up for that.  These outs aren't fungible like say a rangy CF who might "steal" putouts from an RF or LF.  I guess I'm using COF'ers for my first basemen from this point forward.  I understand we are comparing it to a out of position 1B, but still, the number are bigger than I expected.

How do you come up with the 200 points of OPS?

I'd work it backwards.    He's costing his team 55 outs when compared to an average 1B.    Over the course of 162 game season, 4374 outs should be recorded(more or less).   In Mantle, teams are averaging about 4.5 runs per game which will amount to 729 runs.    729/4374=.16666 runs per out.   He's costing his team 9.16 runs a season when compared to an average 1B if he played in Mantle.  

Runs divided by outs times 55.

Tons of work has been done on this.  The value of converting an out into a single, in literally dozens of linear weights studies, centers at around 0.70 - 0.77 runs.  Add about 0.3 runs for each extra base.

Even if none of the misplays involve extra bases, that's about 40 runs a season, which is a lot.  Figure a few doubles down the line and a few overthrows allowing an extra base, and you're talking almost 5 games worth of runs you're giving away.

Another way to look at it-- take 55 singles per 162 games away from his career offensive line.  His slash stats: .197/.282/.416.  Or down 185 points of OPS.

He's killing you.

55/600 =.0916 x2 since those hits effect both obp and slg and you get .183. If even a few turn into extra base hits you arrive 200 points of ops.

I chose 600 at bats out of thin air.

Posted by dedelman on 8/3/2014 1:48:00 PM (view original):
Tons of work has been done on this.  The value of converting an out into a single, in literally dozens of linear weights studies, centers at around 0.70 - 0.77 runs.  Add about 0.3 runs for each extra base.

Even if none of the misplays involve extra bases, that's about 40 runs a season, which is a lot.  Figure a few doubles down the line and a few overthrows allowing an extra base, and you're talking almost 5 games worth of runs you're giving away.

Another way to look at it-- take 55 singles per 162 games away from his career offensive line.  His slash stats: .197/.282/.416.  Or down 185 points of OPS.

He's killing you.

While I agree about removing the hits and seeing what you've got, the .70 out to single number seems too high.    And I think the .1666 I tossed out there is too small.   A player giving up an extra out every 3rd games seems like he'd be giving up more than 9 runs but 40 still seems too high.

Do you know the logic behind the .70 number?

The method is:  for a span of a season or more (has been done for at least 30 seasons) you determine how many runs were scored, on average, in each base-out situation (eg, bases empty, no out; runner on first, one out, etc.).  The you compare the difference between a single and an out in each of those 24 contexts (eg, for bases empty, no out, you compare the difference between the average number of runs scored with bases empty, one out vs runner on first, no out).  Repeat for all 24 contexts.  You obviously need to know things like what percentage of runners go first-to-third on a single to get an accurate calculation.  Basically, the average difference between average runs scored after a single, and after an out, averaged over all 24 base-out contexts, is 0.73 or so.

The guy who has done the most work with this in the last 15 years is Tom Tango.

Well, I have no doubt that Tango has done more with baseball stats than I'll do in 100 lifetimes but that still feels too high.    Obviously there are times when the error automatically results in a run(runner on third, # of outs doesn't matter) but that number says 70% or more, the error results in a run.   More than 2/3 of the time.  That doesn't feel right. 

Mantle averages:

110 games 
490 runs
455 earned runs
61 errors
22 negative plays

An error seems to produce .573 runs.    Not sure how to include negative plays since it wouldn't produce an unearned run.   

It's not just the error-- it's the extension of the inning, the failure to get an out.  That's big.  And sometimes those runs are earned.

MG last season
705/655
85 errors
.588 runs

Coop
689/642
82
.573

I kind of see a pattern.

Posted by dedelman on 8/3/2014 8:11:00 PM (view original):
It's not just the error-- it's the extension of the inning, the failure to get an out.  That's big.  And sometimes those runs are earned.

Aren't all runs after an error unearned?

Posted by MikeT23 on 8/3/2014 7:33:00 PM (view original):
Well, I have no doubt that Tango has done more with baseball stats than I'll do in 100 lifetimes but that still feels too high.    Obviously there are times when the error automatically results in a run(runner on third, # of outs doesn't matter) but that number says 70% or more, the error results in a run.   More than 2/3 of the time.  That doesn't feel right. 

That application of the logic is not quite right. Imagine this. Two outs, no one on. Instead of a third out, poor fielding results in a one base error and there is now a guy at first. The inning should be over. The pitcher then surrenders two walks and a grand slam. The next guy pops up to the catcher. The error resulted in 4 runs without even taking into account the effect of the extra pitches thrown later in the game. It's not that the guy comes around to score 70% of the time! it's just the benefiting team scores .7 more runs a game.

Though I do agree that even taking what I said into account .7, seems too high. I use .6 when doing my calculations for not specific reason other than Mike's.

The average run weight of a single is around .45 of a run, the average out is about -.30.  It's the difference between the two that's about .75 but the average single adds a little less than half a run to run expectancy compared to expected runs prior to the PA .  Of course the value of each in a given situation varies quite a bit depending on # of outs and # and location of baserunners, as you mentioned.