Pitching WAR

[jcmurr]

Is WAR for SP just a completely broken stat now? Why is IP valued so highly? It's irrelevant for judging pitchers for me in-game, but the fact that it says that one of these pitchers as 100% more valuable in creating wins than the other is crazy.

A few more peripherals for P1 and P2:

QS: 25 vs. 26
CG: 11 vs. 13

[Sharkn20]

Originally Posted by jcmurr

Is WAR for SP just a completely broken stat now? Why is IP valued so highly? It's irrelevant for judging pitchers for me in-game, but the fact that it says that one of these pitchers as 100% more valuable in creating wins than the other is crazy.

A few more peripherals for P1 and P2:

QS: 25 vs. 26
CG: 11 vs. 13

BABIP

Sent from my SM-G960F using Tapatalk

[Rain King]

Yea, the 2nd pitcher has much better Walk and HR rates....and while 1st pitcher has been good he has also been very lucky with his BABIP. What is the FIP for each of these players?

[jcmurr]

Originally Posted by Rain King

Yea, the 2nd pitcher has much better Walk and HR rates....and while 1st pitcher has been good he has also been very lucky with his BABIP. What is the FIP for each of these players?

2.99 vs. 2.54

I assumed the FIP would be better for the higher WAR pitcher. But a 0.5 split equates to 4 WAR?

[NotMuchTime]

No matter how anyone tries to justify the HUGE disparity, it's just not justifiable.

[Argonaut]

It is justifiable. Checkout those elite HR/9 and BB/9 numbers. Half as many homers and less than half as many walks. That pitcher must have great MOV and CON ratings.

[Drstrangelove]

Well, does OOTP use runs allowed (like BBRef does), FIP (like Fangraphs does), or something else to calculate WAR? We might need pitchers' runs allowed, league ERA, league runs allowed or league FIP, to determine if that WAR looks right.


Still, in a low run scoring environment, a difference of .45 FIP, or something similar for runs allowed, can amplify WAR, as it should. E.g., a difference of of .45 FIP (or .45 runs allowed) in a league scoring 3.3 runs a game (NL, 1968) is hugely more valuable than the same difference in a league scoring 5 runs per game (NL, 2000.) So, the FIP difference AND being in a pitcher's league would explain that pretty well.

[jcmurr]

Originally Posted by NotMuchTime

No matter how anyone tries to justify the HUGE disparity, it's just not justifiable.

I agree.

I'm not going to go off about how WAR is an "incomplete stat" because I do think it's reliable. For SPs, I have seen several strange cases like this that make me think some variables to the equation are too highly-valued.

In that vein, I'm not sold on the talk that FIP, BABIP, HR/9 explain the bigger picture. I agree his .220 BABIP is luck. It's outside his historical performance parameters and an outlier to the rest of league data. That should be irrelevant.

More important to me is that given his low BABIP, wouldn't that make his higher HR/9 seem a bit unlucky as well? Shouldn't that negate the higher FIP number somewhat? Is FIP really the leading variable here?

Pitcher A gave up 18% of his hits as home runs, whereas Pitcher B only gave up 6%. Yet the rest of the data says Pitcher A was more dominant by giving up significantly less hits and struck out proportionately more hitters (+1 K/9). Why is he penalized by one and not rewarded for the other? I understand HRs are the most efficient run producers. But in this case, WAR seems more an extension of long-term performance potential and less an indicator of actual performance.

At the end of the day though, that is all abstract pondering. The question I still do not remain convinced on: were Pitcher B's performances TWICE as valuable as Pitcher A?

[Matt Arnold]

By one metric, he was.

They'd be much closer in ra9WAR, which factors runs allowed.

If you don't like WAR (fip-based), then don't use it to judge people. End of season awards use a mix of fWAR and rWAR to evaluate, so if those 2 are the top Cy Young candidates, it's probably not going to be a unanimous decision.

[Mike D]

Originally Posted by Matt Arnold

By one metric, he was.

They'd be much closer in ra9WAR, which factors runs allowed.

If you don't like WAR (fip-based), then don't use it to judge people. End of season awards use a mix of fWAR and rWAR to evaluate, so if those 2 are the top Cy Young candidates, it's probably not going to be a unanimous decision.

Cool. Can you show me where i can find fWAR and rWAR in game? Thanks!

[Matt Arnold]

Originally Posted by Mike D

Cool. Can you show me where i can find fWAR and rWAR in game? Thanks!

fWAR is what we call WAR. rWAR or ra9WAR shows up I believe in the extended pitching stats when looking on player pages, or is a column you can select in any of the list/table views.

[Drstrangelove]

Originally Posted by jcmurr

I agree.

I'm not going to go off about how WAR is an "incomplete stat" because I do think it's reliable. For SPs, I have seen several strange cases like this that make me think some variables to the equation are too highly-valued.

What is the league ERA that the pitchers are in and what is that team's park factor?

[Drstrangelove]

Originally Posted by jcmurr

I agree.

I'm not going to go off about how WAR is an "incomplete stat" because I do think it's reliable. For SPs, I have seen several strange cases like this that make me think some variables to the equation are too highly-valued.

In that vein, I'm not sold on the talk that FIP, BABIP, HR/9 explain the bigger picture. I agree his .220 BABIP is luck. It's outside his historical performance parameters and an outlier to the rest of league data. That should be irrelevant.

Whether or not a BABIP is lucky or not isn't determined by looking at his career BABIP, but looking at the league BABIP. He's not "lucky" compared to himself, but to the league. In a normal modern league, a BABIP of ~.300 would be normal, so a BABIP of .220 is extremely lucky. FIP assumes that differences in BABIP are luck, not skill.

(This isn't altogether abstract or theoretical. Pitches and the results of batted balls have been measured in the hundreds of thousands and the evidence suggests that pitchers have little or no ability to impact their own BABIP. Actual BABIP relative to the league average is primarily a result of luck and team defensive skill. Average league BABIP is primarily a result of rules and equipment. Add more infielders, add gloves, change the ball, move the fences and it can change average league BABIP.)

Second, BABIP is anything but irrelevant if one wants to determine WAR. A pitcher who's lucky to pitch in a park with slow infields and great infielders shouldn't be compared to a pitcher who works on artificial turf and slow infielders without some sort of fix. FIP is part of that fix.

FIP deems that if you do things infielders can't effect (e.g., BB, HR, SO, HBP) you should be gauged on those, but things you can't effect (i.e., what ever happens to a baseball that doesn't clear the fence after it's hit) should be ignored. (ra9WAR is done differently.)

Originally Posted by jcmurr

More important to me is that given his low BABIP, wouldn't that make his higher HR/9 seem a bit unlucky as well? Shouldn't that negate the higher FIP number somewhat? Is FIP really the leading variable here?

FIP docks a pitcher for giving up homers and walks, credits him for strikeouts. Low BABIP has nothing to do with FIP because BABIP isn't used. It's not used because FIP deems differences in BABIP to be luck.

Originally Posted by jcmurr

At the end of the day though, that is all abstract pondering. The question I still do not remain convinced on: were Pitcher B's performances TWICE as valuable as Pitcher A?

WAR is the marginal value assigned to a player for his performance on the field. Value is determined in a context using a theoretical method. Context might be league, team, park, era. Theoretical method might be ra9WAR, FIP, etc. Whether or not we can get twice the WAR using FIP depends on we haven't been given. We don't know league ERA, park factors, etc.

For example: If the CWS pitchers were in a hitters park (109 PF), and in a league with a low ERA (3.51), then their ERA+, ERA, FIP and WAR would all sync to what was posted. So, in that case, it would be easy for pitcher B to have twice the WAR of pitcher A. Obviously this is just one example.

So from different points of view, the pitcher was both "lucky" and "dominant" by giving up fewer hits, while also being "unlucky" to give up a lot more homers. (His large number of extra walks are lost somewhere.) FIP sees pitcher A as much easier to homer off and much easier to draw walks on. Those aren't luck related.


Is he worth only half the other pitcher? He could be in the proper context.

[chazzycat]

Originally Posted by jcmurr

In that vein, I'm not sold on the talk that FIP, BABIP, HR/9 explain the bigger picture. I agree his .220 BABIP is luck. It's outside his historical performance parameters and an outlier to the rest of league data. That should be irrelevant.

Why would the fact that a player benefited hugely from luck be irrelevant? Is not one of the main goals in statistical analysis to determine the likelihood that a given player will be able to repeat their performance?

If you just want to know which player was more helpful in a given season, you can completely ignore WAR. Just ERA, W/L record, and innings pitched can tell that story no problem. But if you are looking ahead and trying to gauge future performance, you need better stats.

It's the difference between descriptive stats (ERA) and predictive stats (FIP). One tells you what happened...the other tells you what is more likely to happen going forward. They both have their place.

[Rain King]

Results vs. Performance