HOOTIE wrote:DK where did you get that sb % of 63.7? It's way too low. Using liniar weights, a sb is worth .17 of a run, and a caught stealing -.45. That means 73% is the break even point.

Using the Moneyball theory, if a player steals second successfully, it increases the chance of scoring by .20. If a player is caught stealing, it goes down by .35. So, you need to make it at least 7 successful swipes to every 4 unsuccessful. 7/11 is 63.636363 (and so on).

Using the Moneyball theory, if a player steals second successfully, it increases the chance of scoring by .20. If a player is caught stealing, it goes down by .35. So, you need to make it at least 7 successful swipes to every 4 unsuccessful. 7/11 is 63.636363 (and so on).[/quote]

Man I hope there is not a math quiz to go with the grammer and spelling quiz at the end of this thread

I just solved all the problems, or some of them, posed in this thread! I have found a way to compare wins without placing bias towards pitchers on better teams. Here it is:

Wins per one run of run support minus earned runs.

In symbols: W/[(R of run support) - (ER)]

That is, for the difference in runs given up by a pitcher and runs supporting a pitcher, how many wins did he get?

Now, this needs to be critiqued and tweaked before it becomes the real deal.

just got done skimming this thread and figured I'd contribute to the confusion (and the poor spalling).

My take on FBB categories is that an optimal set (whether it is 4x4 or 5x5 or anything like that) would only contain categories that are as independent of each other as possible. This is probably next to impossible to do for real baseball, but maybe it can be approximated. Also, an optimal set would emphasize the importance of individual player achievements rather than stats that are greatly dependent on the performance of the team the player is on.

I like to dabble with numbers. I recently made a plot of various stats vs. other stats and found some intersting things.

1) I think someone actually studied this a little more in depth, I saw an article on it, ERA and WHIP are not independent. Larger WHIP usually means larger ERA. There is, of course, scatter but that is to be expected in statistical data.

2) it shouldn't surprise anyone that RBI/PA and SLG% are highly correlated. I don't know how to fix this, but the "best" set of categories would probably feature not use one of the two stats.

3) I don't like OPS as a stat. I do like it much more than AVG though. I think it should be OPS-HITS/PA, the reason for this is that hits get double counted in OPS compared to BB or HBP. (just look up the formulas for OBP and SLG and you'll see that hits figure in both)

4) the original poster is right, IMO, that W are a flawed category. In some sense, maybe, something like QS+CG might work better (or some combination like that). QS are a fickle stat and I don't think they are constrictive enough (I mean, 3ER in 6IP isn't THAT great...)

5) K for pitchers are a good stat, I think. I'd like to maybe see something like 'K with runners at 3rd' as a possible stat. but I see no problems with Ks.

6) I really don't like HRs as a category. My main issue with is is that when someone hits a HR four of the five standard categories are changed (R,RBI,HR and AVG). This is not right. In some sense, maybe the categories should be R-HR, RBI-HR, HR and AVG (I'll live with just the average being affected and getting the RBI for the runners on base already)...

i have some other ideas, but maybe I'll put them in another thread

Bloody Nipples wrote:I just solved all the problems, or some of them, posed in this thread! I have found a way to compare wins without placing bias towards pitchers on better teams. Here it is:

Wins per one run of run support minus earned runs.

In symbols: W/[(R of run support) - (ER)]

That is, for the difference in runs given up by a pitcher and runs supporting a pitcher, how many wins did he get?

Now, this needs to be critiqued and tweaked before it becomes the real deal.

So fire away!

i cant wait to see it on yahoo next year

Lets Go patsox!

yay! sox win!!!
bambino-1918-2004
Rest In Pain ;-)

Nice idea, Bloody Nipples, I can see at what you're trying to get at, but that formula makes it optimal for a pitcher to give up exactly one run less than the runs of run support he got. If he gave up no ER, he'd get a W/(RS-ER) of nearly 0, and Arlo has already pointed out the problem when ER=RS.

jbuk wrote:Nice idea, Bloody Nipples, I can see at what you're trying to get at, but that formula makes it optimal for a pitcher to give up exactly one run less than the runs of run support he got. If he gave up no ER, he'd get a W/(RS-ER) of nearly 0, and Arlo has already pointed out the problem when ER=RS.

W/(ER+RS) makes a bit more sense.

right...

RS/GAME and ER/game might be better, but you still run into the problem of equality and such... you almost want to actually look at:

(RS-ER)/W if W>0.
you remove the possiblity of infinity (unless you consider pitchers with 0 wins... but do you really want to?)

Also, you now have TWO team factors -- W and RS for a personal stat -- not too good, is it? I mean... do you really want the pitcher who gives up 5runs/game ands gets 6 in support (and say, wins 15) to have the same stat as the pitcher who gives up 1 and gets 2?