Fangraphs: Ubaldo has been worth +0.6 fWAR this year

[BradyBunch]

Would you consider a starting pitcher with a 5 ERA anything but bad?

Check out Mike Mussina's numbers in 1996. He had a 4.80 era and went 19-11. Had a whip of 1.3+, worth 5 WAR and he still was 5th in CYY votes.

We are back in HR era again so we have to shift our view of stats.

[waroriole]

Which is EXACTLY why I don't look at fWAR, only rWAR. Fangraphs does a ton of things well, but calculate WAR is not one of them in my opinion. They try to be a little too smart for their own good. Any stat that nets Jimenez as plus value needs to be immediately discontinued because it's majorly flawed.

Yes!!!! All of this.

[Il BuonO]

This is a random stat I came across this morning that surprised me. Essentially, Ubaldo's FIP (4.90) and xFIP (4.96) are not all that terrible. Arguably, he's been terribly unlucky, with a .380 BABIP. So, fangraphs assigns him a +0.6 fWAR.

And yet, that just flies in the face of common sense for anyone who has watched him pitch. BB-ref has him at -1.6 rWAR, which seems much closer to the mark than fangraphs' +0.6 fWAR.

I thought you are not a fan of FIP? Fippity doo dah or some such thing you said regarding its accuracy.

And please explain how what you cite is different than Arrieta (4.61 FIP) the year they traded him.

[DrungoHazewood]

Which is EXACTLY why I don't look at fWAR, only rWAR. Fangraphs does a ton of things well, but calculate WAR is not one of them in my opinion. They try to be a little too smart for their own good. Any stat that nets Jimenez as plus value needs to be immediately discontinued because it's majorly flawed.

Or maybe just look into the reasons why there is something that doesn't match up with your observations. If you discontinue use of all metrics that sometimes have outliers that need further explanation we'll be out of metrics quickly.

[Nevermore]

What I'd like to find are the actual mathematical formulas to calculate fWAR, rWAR, FIP and other similar metrics. I'm a math major and I'd like to be able to analyze what exactly I am seeing. Are these linear equations, quadratic equations, cubic equations, or quartic equations? Can you differentiate these functions to measure rate of change?

Are there sites that provide this information?

[Frobby]

I thought you are not a fan of FIP? Fippity doo dah or some such thing you said regarding its accuracy.

And please explain how what you cite is different than Arrieta (4.61 FIP) the year they traded him.

I only mentioned FIP by way of explaining how fangraphs came up with +0.6 fWAR. I think it is pretty clear from my other posts in the thread that I do not agree with their valuation of Ubaldo, and implicitly, that I don't agree with their heavy reliance on FIP.

I'm not sure what this has to do with Arrieta. Clearly his FIP with the Orioles was not at all predictive of what happened when he got to Chicago.

[Spy Fox]

What I'd like to find are the actual mathematical formulas to calculate fWAR, rWAR, FIP and other similar metrics. I'm a math major and I'd like to be able to analyze what exactly I am seeing. Are these linear equations, quadratic equations, cubic equations, or quartic equations? Can you differentiate these functions to measure rate of change?

Are there sites that provide this information?

Fangraphs has all the math provided:

Position player fWAR.

Pitcher fWAR.

I can't find actually equation-by-equation math for BBRef but they do have pretty detailed outlines of their process of determining position player WAR and pitcher WAR.

[LookitsPuck]

It's why I hate FIP. Unless you're giving up homers, it doesn't really ding you much.

[DrungoHazewood]

I only mentioned FIP by way of explaining how fangraphs came up with +0.6 fWAR. I think it is pretty clear from my other posts in the thread that I do not agree with their valuation of Ubaldo, and implicitly, that I don't agree with their heavy reliance on FIP.

I'm not sure what this has to do with Arrieta. Clearly his FIP with the Orioles was not at all predictive of what happened when he got to Chicago.

I think most of the arguments surrounding continued faith in Jake Arrieta in early-to-mid 2013 centered on the fact that his FIP/xFIP over his last 150 innings in Baltimore was about 2.00 runs better than his ERA.

[DrungoHazewood]

It's why I hate FIP. Unless you're giving up homers, it doesn't really ding you much.

Well, unless you walk people. Or don't strike out many. It's kind of the point that it doesn't really ding you much for things that are heavily influenced by things that you can't really control.

[Hallas]

Well, unless you walk people. Or don't strike out many. It's kind of the point that it doesn't really ding you much for things that are heavily influenced by things that you can't really control.

I think it should ding (or reward you) for quality of contact. Exit velocity and GB/LD/FB are a start to that.

With exit velocity, it seems that a correlation between it and future performance has yet to be determined, in large part because the available data is so new.

GB and FB rates are pretty repeatable and also have some correlation with pitcher performance. FIP does take into account FB in the sense that it uses expected-HRs based on the number of flyballs you allow, but it doesn't adjust for other batted ball types, nor does it adjust for in-play flyballs, even though batted ball types are controllable by a pitcher.

I would probably like a modified version of fWAR that took into account batted ball outcomes more than I like rWAR. The approach that the current incarnation of fWAR took - ignoring factors that it didn't understand - rubbed me the wrong way.

[Nevermore]

Fangraphs has all the math provided:

Position player fWAR.

Pitcher fWAR.

I can't find actually equation-by-equation math for BBRef but they do have pretty detailed outlines of their process of determining position player WAR and pitcher WAR.

I looked up these formulas. They are all linear equations, using linear algebra or arithmetic. Which makes them very static. What I'd like to see are formulas that would measure change in performance. This would either involve probability theory or statistics. Or else differential equations. I'm still trying to figure out how to do this. You would have to compare several calculations of WAR (or ERA, WHIP or whatever). Graph these and look at the pattern that the graph makes.

[waroriole]

It's why I hate FIP. Unless you're giving up homers, it doesn't really ding you much.

According to FIP, there is no difference between a ball hit off the top of the wall and a slow roller back to the mound. It's all just happenstance.

[DrungoHazewood]

According to FIP, there is no difference between a ball hit off the top of the wall and a slow roller back to the mound. It's all just happenstance.

According to ERA there's no difference between:

1) a clean inning with three Ks on nine pitches and

2) an inning where you walk two, get a double play on a liner to short, walk two more, and have the CFer rob a homer at the 420' sign in center.

It's all just results, right? If you want to fixate on outliers why are we even discussing metrics? Why not just stop at standings?

[waroriole]

According to ERA there's no difference between:

1) a clean inning with three Ks on nine pitches and

2) an inning where you walk two, get a double play on a liner to short, walk two more, and have the CFer rob a homer at the 420' sign in center.

It's all just results, right? If you want to fixate on outliers why are we even discussing metrics? Why not just stop at standings?

They both clearly have issues. You don't see me in here taking up for ERA, but many very smart people are ignoring the glaring problems with FIP.

Combining ERA with WHIP seems a much more effective tool than FIP.