Arm/range studies

[drew6013]

So....
Who would be better in left field defensively, Eiether or Bay? a 3(-1)e10 or 3(+1)e3?

[Mean Dean]

That's what I'm trying to figure out :) Not quite there yet. This thread will deal solely with how many runs arms are worth. It's actually not nearly as time-consuming to figure out how much range is worth, and I think I've pretty much got it, but I'm not gonna put it all together until I'm done with the arms.

[LMBombers]

I would guess Bay is better than Ethier defensively.

[palacekillers]

DeanTSC - I haven't even made it to the actual simming part yet. I can't even figure out how to assign the players to their teams. I clicked the Draft-o-matic thing but didn't understand it right away and gave up. Admittedly, it was late, and I had spent a lot of time trying to figure out other things first. Maybe I just need to take a fresh stab at it.

Any suggestions though?

[Mean Dean]

Here's what I'd do:[list:3b2893552b][*:3b2893552b]Use League → Duplicate League to duplicate the 2006 Major Leagues.[*:3b2893552b]Go to your new league (League → Update League to re-name it, so you don't get it confused with the real major leagues), select all the teams, and do Team → Release Players. All MLB players are now in your league free agent pool.[*:3b2893552b]Delete all but 12 teams from the league; I suggest leaving all AL teams if you're simming a DH league, and all NL teams if you're not.[*:3b2893552b]Team → Update Team to change the team names.[*:3b2893552b]Draft-O-Matic, with all teams on manual control. Then just go thru and draft each player onto the team they should be on. (Player → Draft Selection, or more conveniently, CTRL-D to draft)[*:3b2893552b]League → Update League and set it up to be one subleague, three divisions, put the teams in the right divisions, make sure the DH settings are right.[*:3b2893552b]Then if you want to tweak, you can edit the computer managers (Team → Update Computer Manager and Team → Update Default Lineups) and such, but that is the basic idea. And like I said, if you want to sim the league more than once, you can just re-start it. If you want to sim a different TSN league, you can duplicate this league and at least avoid re-doing some of the setup stuff.[/list:u:3b2893552b]

[Coffeeholic]

An interesting study Dean! Hat's off to you for taking the time to run the sims and then post the results. :)

Of course, your results would have much more valididty if you could increase your sample size. Is it very time consuming to sim a season once you have the parameters in place?

You could conceivably alter all kinds of different variables to see how they would effect play... pretty neat... I might have to buy the CD game and give it a whirl.

[Mean Dean]

I dunno, I think 50 is a decent sample size. I would like to bump it up to 100 seasons eventually, but I think it's more useful right now to get plausible estimates about these things than it is to get something really exact.

It actually doesn't take too long to sim a single season; about 10 minutes. But that still means 50 takes a while (especially since I'm not literally simming a season every 10 minutes, I do have a life to lead ;)... and naturally in order to compare two different ratings, you have to do it twice... and especially with throwing arms, which range all the way from -5 to +5 (so, 11 possibilities), you probably want to test more than two of them eventually... so, it all adds up.

[RAYMONDSTRAWN]

Like all scientific and mathematical studies, I challenge someone else to repeat this study to see if the results are duplicated.

What about the Bill Jame Pythagoreon formula (yeah i don't know how to spell) here, wouldn't it tell you how this theoretically should affect winning percentage? I think sharing the runs is interesting, what about the average winning percentage for the different fifty game seasons.

Exciting stuff. thanks for doing it. Hope someone else will duplicate it and report their results here.

[Mean Dean]

Thanks! It's tough to incorporate these findings into the Pythag stuff with true exactitude, because it's going to end up depending on how many runs the team scores and allows. As I mentioned earlier, I am running these tests in the context of a team that is not allowing all that many baserunners, and thus is only giving up 650 runs a year (about four a game.) It's possible that on a team that gave up more baserunners, these defensive differences would add up to more runs (while still representing the same percentage value of your defense.) To determine that, you would have to run all of these same tests with a control group of a team with worse pitching/defense, which I'm really not sure I'm ever gonna be up for :)

However, a general rule of thumb is that 10 runs equal one additional win. So you can go by that, if you want to have a pretty good idea of how many wins we're talking about.

I will have LF/RF arm stuff very soon.

[LANCEBOUSLEY]

i ran this test with a 2005 league i have on my cd-rom.

used +5,-5 and 0 for the arm ratings of grady sizemore.(cause he was on my team)... i ran 10 seasons for each and got pretty much the same results.....

a +5 arm costs ya wins on pace higher than a -5 gets you wins.... comparing to a 0 arm.... i think the +5 was on the order of around 5 wins a year relative to a 0. assuming 10 runs per win... i only looked at runs allowed as well.....