Value of a good arm?

Doc: Check out this excellent thread from TSN archives about the impact of the OF arms, with a focus on the CF arm.

http://forum.365.strat-o-matic.com/arch ... 1&t=251327

Here's a portion of the first post in the thread:

This is a study of the effects of CF arm rating. I used the 2006 Mets as my control group. The following players started every game: Paul Lo Duca C, Carlos Delgado 1B, Jose Valentin 2B, Jose Reyes SS, David Wright 3B, Endy Chavez LF, Carlos Beltran CF, Shawn Green RF. I made Green into a 1 fielder because I didn't want the computer to replace him for defense, since I plan to do this same study for RF as well, so I wanted to make sure he played every inning or damn close.

In 50 simulated seasons, this team allowed an average of 650 runs per season.

I then changed Beltran from a -3 arm, to a 0 arm.

In 50 sim seasons, the team with this change made allowed 652 runs per season. (In case you're [i:385d085c75]really[/i:385d085c75] bad at math, that's a difference of two runs

I then changed Beltran to a +3 arm.

In 50 sim seasons, that team allowed 667 runs a season. (Actually exactly 666.66, which is spooky In any event, that's a difference of 17 runs.

Using an ATG team with DH, where scoring is approximately 50% higher compared to real MLB baseball teams, instead of the '06 Mets as a control group, may show slightly higher differences in runs allowed based on CFer arms, but the conclusion that there's little difference in runs allowed over the course of a season by a team with a CF arm of -3 compared to a team with a 0 CF arm seems likely to remain valid.

I hope this helps. - Bernie

Very wonderful info there!

Thanks

In 50 simulated seasons, this team allowed an average of 650 runs per season.

I then changed Beltran from a -3 arm, to a 0 arm.

In 50 sim seasons, the team with this change made allowed 652 runs per season.

I then changed Beltran to a +3 arm. Difference of 2 runs/season

In 50 sim seasons, that team allowed 667 runs a season. (Actually exactly 666.66, which is spooky In any event, that's a difference of 17 runs/season.

In summary:
-3 to 0 : 2 runs
0 to +3: 17 runs

I've seen this result from Strat simulation, but I would not take these numbers straight up. Even though 50 sims is quite an effort, but there is still some randomness that could have affected the average total runs. Perhaps if he did another sim with Beltran having a 0 arm, he might have on average 655 runs allowed, and then the numbers would be:

-3 to 0: 5 runs
0 to +3: 14 runs

Maybe rburgh could do a sim on his own!!

But the pattern is clear: it hurts much more to go from 0 to +3 than to go from -3 to 0, but it might not be as big as the sim lets it mean.

Here is my own way to analyze the value of outfield defense. It's based on an analysis relatively in depth of 20XX seasons, and I'm in the process of making the adjustments for ATG, but I haven't completed the transformation.

Everything is based on runs allowed (or run saved) in a full season. I assume here that every player are 1-injured rated (3 games max---I make further adjustments later for players who are more or less injury-prone).

Basically the idea is to start by how many runs a perfect outfielder can save, and then deduce that number by the runs allowed from a less than perfect range, from the errors, and from a less than perfect arm.

The number I start with is kinda arbitrary, but it's adjusted so that it makes sense relatively to the other positions in the field. The way I set the number, a perfect cf will have less impact than a perfect ss or 2b, but will make a greater impact than a perfect 3rd or 1st baseman.

CF=-14 which assumes that a perfect cf is 1e0(-4)
RF=-12 which assumes that a perfect cf is 1e0(-5)
LF=-11 which assumes that a perfect cf is 1e0(-3)

(Thus, I don't give any value for a lf who is -4 instead of -3)

Then I assume that a player will replicate his e number in a full season X 110% and that each error is worth 0.6 runs, which is the same as assessing the value of errors as 0.66. I'll explain later why I came up with this number, which might seem low to some folks (in his comment, Doc hints that each error costs a full run, rather than 0.6 runs).

I assume that each increase of range costs 7 run for players rated 2 or 3 (make it 7 X 1.5 = 10.5 runs for centerfielders).

Corner outfielders with a range of 4 have a 25 run penalty (instead of 21 runs, again X 1.5 runs for centerfielders)

The penalty for arm goes the following:

+1..+2..+3..+4..+5...+6..+7 +8 +9...+10
0.2-0.7-1.6-2.8-4.3-6.0-8.1-10.3-13.0-15.8 (X2 for centerfielders)

So a left fielder with a -2 arm has a 0.2 run penalty (because it's only +1 from a perfect -3 arm).
A right fielder with a 0 arm has a 4.3 runs penalty (because it is +5 from a perfect -5 arm).
A centerfielder with a +3 arm has a 8.1 runs penalty (because it is +7 from a perfect -4 arm).

Duke is 1e5(-2)

-14 + 5*0.66 + 0.7 = -10 runs. Duke's defense is worth 10 runs.

Really strong arms keep baserunners rooted. Further, the CF's arm is involved in a lot of baserunning decisions. If the bats are relatively equal, I'd go with the better arm, especially from a key offensive contributor. Extra bases can be the difference in close games (and there are a lot of close games to be played).

For myself, I worry less about arms on the flanks. You can substitute for them defensively in a close game.