Sox get Peavy

[Orange Juiced]

Then what's the point of this? We're taking about Jake Peavy, not a hypothetical pitcher. He has starts where he gets lit up. You can't pretend they don't happen.

 

Go back to my first post on this topic, where I presented my point about Peavy. My argument is that his stat line looks much better when you take out that one horrific outing. He's not likely to repeat that horrific outing, though he will likely have a couple of not-so-good ones. And once that horrific outing (which he's not likely to repeat) is removed, he's actually been a really good pitcher this year.

 

I was forced to use that illustration that seemed a bit extreme to point out that we can't just look at the final stat line to show the quality of a pitcher. Some guys are consistently what they are. Other guys get those numbers because they are generally better than that first group, but just have the occasional blowup for some reason. So yes, in the end, over the course of a season, the stat lines may look similar. But that doesn't mean that they're both equal.

 

I think Peavy is more likely to be around a 3.70 pitcher than a 4.20 pitcher, was my point.

[User Name]

If you want to evaluate an SP by taking away his worst starts, you have to take away his best starts as well in order to remove all outliers.

[sk7326]

A 3.7 ERA is NOT utterly dominant 9 out of 10 starts.

 

But can our 5-man rotiation churn out 5 solid starts over a turn consistently? With Peavy, by all means yes. Heck, with Workman they've been doing it the last month. Yeah, this move doesn't turn them into Tampa Bay - but looking at how close those games were, they didn't have to. No soft spots in the rotation now - whether or not Buchholz comes back to his pre injury form.

[BornToRun]

If you want to evaluate an SP by taking away his worst starts, you have to take away his best starts as well in order to remove all outliers.

 

Time for me to do some math.

Damn.

[Orange Juiced]

If you want to evaluate an SP by taking away his worst starts, you have to take away his best starts as well in order to remove all outliers.

 

Just his one worst and one best. And his worst impacts his numbers far more than his best.

[Orange Juiced]

But can our 5-man rotiation churn out 5 solid starts over a turn consistently? With Peavy, by all means yes. Heck, with Workman they've been doing it the last month. Yeah, this move doesn't turn them into Tampa Bay - but looking at how close those games were, they didn't have to. No soft spots in the rotation now - whether or not Buchholz comes back to his pre injury form.

 

Right. I like this part. Every night the Sox should be throwing a guy that is pretty darned good. In terms of the last two months and getting to the playoffs, this will be huge.

[User Name]

If you want to evaluate an SP by taking away his worst starts, you have to take away his best starts as well in order to remove all outliers.

 

By this criteria, he'd have a 4.09 ERA. Slight improvement, but not a .50 ERA improvement.

[BornToRun]

Took away the two 6 ER starts as well as a 1 ER CG and one of 7 IP and 1 ER. Came out to 57 2/3 inning with 24 ER allowed which is a 3.74 ERA.

 

Edit: I think similar performance is a fair expectation when you factor in our good CF/RF defense with his low WHIP and good K/BB ratio.

[User Name]

Just his one worst and one best. And his worst impacts his numbers far more than his best.

 

For this case yes, since you are only mentioning his worst blowup. I am merely presenting the rule of thumb. For every one of his worst starts you discount, you have to discount one of the best.

[User Name]

Took away the two 6 ER starts as well as a 1 ER CG and one of 7 IP and 1 ER. Came out to 57 2/3 inning with 24 ER allowed which is a 3.74 ERA.

 

Taking the two 6 ER ones is not plausible. He does not have two equally good starts to discount for analytical purposes, which is strange.

[Orange Juiced]

For this case yes, since you are only mentioning his worst blowup. I am merely presenting the rule of thumb. For every one of his worst starts you discount, you have to discount one of the best.

 

Right. And if you do that, he's much closer to a 3.70 pitcher than a 4.28 pitcher. That was my point in the first place.

[BornToRun]

Taking the two 6 ER ones is not plausible. He does not have two equally good starts to discount for analytical purposes, which is strange.

 

How do you calculate what would constitute equal value in a good start compared to the bad ones? I'm not arguing, just curious.

[Orange Juiced]

How do you calculate what would constitute equal value in a good start compared to the bad ones? I'm not arguing, just curious.

 

You can't, and that's now how it's done. Because in any statistical sample, the outliers on either end may not be equal outliers. For example, if we took 10 NFL players and measured their average weight, and we took away the lightest guy and the heaviest guy, it would look different if the heaviest guy was Jerod Mayo (around 250 lbs) versus Vince Wilfork (around 340 lbs).

 

So that shows that the outliers may have a greater or lesser effect on the average. So you can't look to take away equal data - what you do is take away an equal number of data points from either end and then do the math.

 

So take away Peavy's best and worst outing, and you end up with a guy around 3.75.

[User Name]

How do you calculate what would constitute equal value in a good start compared to the bad ones? I'm not arguing, just curious.

 

Usually what happens is that pitchers have several "outlier" starts. Those are starts you know a pitcher can't consistently get, like 2 IP, 6 ER blowups or CG shutouts. But Peavy's only "good" outlier is a 9 IP 1 ER outing, and his two best after that are nearly identical 7 IP,1 ER outings. Those aren't outliers.

 

I think the 4.09 ERA resulting from removing his best and worst start is the best you can when looking for a "true" average.

[User Name]

You can't, and that's now how it's done. Because in any statistical sample, the outliers on either end may not be equal outliers. For example, if we took 10 NFL players and measured their average weight, and we took away the lightest guy and the heaviest guy, it would look different if the heaviest guy was Jerod Mayo (around 250 lbs) versus Vince Wilfork (around 340 lbs).

 

So that shows that the outliers may have a greater or lesser effect on the average. So you can't look to take away equal data - what you do is take away an equal number of data points from either end and then do the math.

 

So take away Peavy's best and worst outing, and you end up with a guy around 3.75.

 

You can. If you can reliably pinpoint what an "outlier" is for a pitcher. With Peavy having two nearly identical 7 IP, 1 ER in a small sample, they can't be considered outliers. However, both 6 ER outings can be.

[BornToRun]

You can't, and that's now how it's done. Because in any statistical sample, the outliers on either end may not be equal outliers. For example, if we took 10 NFL players and measured their average weight, and we took away the lightest guy and the heaviest guy, it would look different if the heaviest guy was Jerod Mayo (around 250 lbs) versus Vince Wilfork (around 340 lbs).

 

So that shows that the outliers may have a greater or lesser effect on the average. So you can't look to take away equal data - what you do is take away an equal number of data points from either end and then do the math.

 

So take away Peavy's best and worst outing, and you end up with a guy around 3.75.

What has me excited is that his peripherals are solid and his big issue is lefties who are slugging .488 off of him (.306 on base) and I'm thinking that our huge right field with Vic out there and Ells in center should help him with that.

[BornToRun]

You can. If you can reliably pinpoint what an "outlier" is for a pitcher. With Peavy having two nearly identical 7 IP, 1 ER in a small sample, they can't be considered outliers. However, both 6 ER outings can be.

I'm confused lol

[User Name]

Also, i don't know where you came up with 3.75 ERA Orange Juiced. If you take away his best and worst outings, you end up with a 4.09 ERA.

[BornToRun]

Usually what happens is that pitchers have several "outlier" starts. Those are starts you know a pitcher can't consistently get, like 2 IP, 6 ER blowups or CG shutouts. But Peavy's only "good" outlier is a 9 IP 1 ER outing, and his two best after that are nearly identical 7 IP,1 ER outings. Those aren't outliers.

 

I think the 4.09 ERA resulting from removing his best and worst start is the best you can when looking for a "true" average.

I think I get it. So you're saying that 7 innings with 1 ER isn't good enough to or rare enough to compensate for the terrible starts he made before going on the DL?

[BornToRun]

Also, i don't know where you came up with 3.75 ERA Orange Juiced. If you take away his best and worst outings, you end up with a 4.09 ERA.

 

I took out the two bad ones, the CG and one of the 7 innings with 1 run starts and got 3.74, and a 4.00 when you took out the other 7 1 ER.

[User Name]

I think I get it. So you're saying that 7 innings with 1 ER isn't good enough to or rare enough to compensate for the terrible starts he made before going on the DL?

 

In theory yes. I'm not going to sit here and tell you i'm right, but it certainly helps clear the water if you try to keep the analysis as simple as possible. That's the problem with trying to analyze data in small samples. Too many outlier landmines.

[Bellhorn04]

You can't, and that's now how it's done. Because in any statistical sample, the outliers on either end may not be equal outliers. For example, if we took 10 NFL players and measured their average weight, and we took away the lightest guy and the heaviest guy, it would look different if the heaviest guy was Jerod Mayo (around 250 lbs) versus Vince Wilfork (around 340 lbs).

 

So that shows that the outliers may have a greater or lesser effect on the average. So you can't look to take away equal data - what you do is take away an equal number of data points from either end and then do the math.

 

So take away Peavy's best and worst outing, and you end up with a guy around 3.75.

 

I think for this kind of analysis you should also be looking at things like Quality Starts and Quality Start %.

[sk7326]

One thing to note here btw is that a bunch of good outings can offset even a bunch of bad ones - just because a team is FAR more likely to win a game giving up 6 runs than they are likely to lose a game where they give up 2. All runs are not precisely created equal.

[Bellhorn04]

Quality Starts is a crude measure but at least it gives you some idea of how many good starts the guy has given you out of his total starts.

[User Name]

My main problem with QS is that 6 IP, 3 ER outing (4.50 ERA) is considered a QS while a 9 IP, 4 ER outing (4.00 ERA) is not.

 

However, i agree with Bellhorn in that it's a decent tool for evaluating the overall effectiveness of a pitcher.

[User Name]

One thing to note here btw is that a bunch of good outings can offset even a bunch of bad ones - just because a team is FAR more likely to win a game giving up 6 runs than they are likely to lose a game where they give up 2. All runs are not precisely created equal.

 

Yeah but on average, the best and worst one will have a similar impact on a pitcher's overall ERA. Unless the worst start is a true disaster.

[a700hitter]

Did you even actually read my post and follow my train of thought? Because this comment gives the impression that you didn't.

I read it. I don't agree with it. Her's a question that I ask people when they tell me that a pitcher had a good outing except for one bad inning. Do you know what you call a pitcher who has one bad inning (4 or more runs) every start? A bad pitcher.

[User Name]

That's not what he's trying to say though.

[a700hitter]

If he wanted to say that he is pitching better lately, that is a legitimate argument, but the stats he put up didn't support that argument. He didn't show a progression of increasing good outings. He just pulled out the bad start. That's just not a valid measure of anything.

[BornToRun]

I read it. I don't agree with it. Her's a question that I ask people when they tell me that a pitcher had a good outing except for one bad inning. Do you know what you call a pitcher who has one bad inning (4 or more runs) every start? A bad pitcher.

 

I think that can be argued against by using the gospel of SSS.