The Starting Rotation

[Baldkin]

Baseball-Reference uses ERA to calculate WAR, and they had WOLF at 2.1 WAR last year, which equates to ~11M.

[logan82]

I am not sure about the total calculation but it relies heavily on strikeouts and walks. I believe it normalizes HR and BABIP. They are trying to separate out a pitcher from the defense behind him.

FIP normalizes LOB% and BABIP more or less.

xFIP normalizes HR% as well.

 

None of them are perfect but FIP is just an odd choice to use for WAR. ERA tells what happened better than FIP and xFIP shows true skill level better than FIP so why use the middle one.

I don't think it is to odd that they don't use ERA since they are trying to separate out defense. I don't think it gives you an accurate value though.

[topper09er]

No, you CAN'T remove what you consider an outlier. Data is data.

Correct, you remove outliers if you have a justifiable reason that it does not reflect the same sample as the rest of your data point, or it was collected under different conditions. For example in structural testing which I have background in, you may throw out an outlier because you determined that the strain gage collecting that data was broken or not calibrated correctly thus the data it gives is not the same as other data points in the sample from funtional strain gages. Or you may throw out a data point because it failed by a different mode thus you essentially did not conduct the same experiment as the other data points, there are many reasons for this. You CANNOT remove a data point simply because it is different from the others.

 

Baseball is different, every start you are facing major league hitters and you try to get them out. The lineups and parks are different but there are factors to account for this like ERA+ or WAR. Unless you have a good reason as to why Wolf facing the Pirates on July 21, 2010 was fundamentally different than any other MLB starting pitcher facing any other MLB lineup then you cant just throw it away because it is an "outlier".

 

This is the other problem, what is an outlier? This is the major problem I have with many sabr inclined people is that they do not understand statistics and treat baseball as if everything has tight normal distributions where the average of any population is a good approximation for the population and all baseball events are independent, random variables. If this was the case you would be able to predict baseball events with some degree on uncertainty, and if you could do that you should move to Las Vegas and become a millionaire. What type of distribution applies to the ERA of a MLB pitcher on any given outing? Normal, log-normal, gamma? Can you predict how many of those will occur this season, how confident are you in that prediction? Basically, can you prove this is an outlier even if it is the only time it has ever happened to Randy Wolf in his whole career?

[dlk9s]

Or, we can all stop trying to impress each other with our knowledge of quantitative analysis and just say, "Hey, you know what? That one ugly game aside, Randy Wolf had a pretty nice 2010 season."

[topper09er]

Or, we can all stop trying to impress each other with our knowledge of quantitative analysis and just say, "Hey, you know what? That one ugly game aside, Randy Wolf had a pretty nice 2010 season."

Not if you compare him to everyone else in the league minus their one ugly game. If you do this he will still be towards the bottom of MLB pitchers. I find it hard to say someone had a pretty nice season if it below average.