I wonder if it’s possible for somebody to do an analysis on...

[Tx Oriole]

1 hour ago, higgybaby said:

I wonder if it’s possible for somebody to do an analysis on...

What would the orioles record have been had Ubaldo, Miley and Tillman been replaced with Lance Lynn, Alex Cobb and Miguel Gonzalez this past season?

It can’t be as simple as just switch their records out would it?  Would you have to use some advanced metrics to figure it out?

 

That's an exercise in futility. You can't possibly know what Lynn and Cobb would have done in Baltimore. They would have been better but still have the same offense that struggled. 

[Luke-OH]

43 minutes ago, Frobby said:

Oh, believe me, I'm capable of it.   But I didn't really have a point to make in this instance.    

I'm not a huge fan of pitcher fWAR, though.    By that measure, Ubaldo was in positive territory all four years he was here (+4.9 overall).    That is so contrary to what I witnessed that I just can't put much credence in it.   rWAR had him in negative territory in 2014 and 2016, reasonably positive in 2015 and just barely positive in 2017, and worth 0.9 overall.    That just seems a lot closer to the truth.   

rWAR for pitchers is better at telling you how well someone has done, fWAR for pitchers is better at telling you how someone will do in the future.   

[Frobby]

25 minutes ago, phillyOs119 said:

rWAR for pitchers is better at telling you how well someone has done, fWAR for pitchers is better at telling you how someone will do in the future.   

I'm not sure I even believe that.    If Ubaldo pitches half-decently next year for someone else, maybe then I'll believe it.    

[Luke-OH]

20 minutes ago, Frobby said:

I'm not sure I even believe that.    If Ubaldo pitches half-decently next year for someone else, maybe then I'll believe it.    

You can't base statistical measures on one player's performance.  The fact is FIP has shown to have a noticeably larger correlation to future performance than ERA.

[Number5]

14 hours ago, phillyOs119 said:

You can't base statistical measures on one player's performance.  The fact is FIP has shown to have a noticeably larger correlation to future performance than ERA.

I'm not sure how this can be categorized as a fact.  Reasonable arguments can be presented for both sides.  They are both useful tools, but neither FIP nor actual ERA should be taken as absolute indicators of how a pitcher will do next year, IMO. 

[Frobby]

16 hours ago, phillyOs119 said:

You can't base statistical measures on one player's performance.  The fact is FIP has shown to have a noticeably larger correlation to future performance than ERA.

Two things:

1.   Have there been any recent studies on this?   I'm aware there was a study at the time of Moneyball, but I'd like to know just how true this is.

2.    Even if the statement is true over the group of all players, there still may be ways of grouping players for whom it's not true.    I don't know if there are such predictable groups or not, and whether Ubaldo would fit in to one of those.     I wonder what the track record is in cases where FIP and ERA diverge by more than a run, for example.    Maybe I'll play around with this over the winter.

[El Gordo]

Ubaldo, Tillman and Miley combined for 15 QS. Cobb Lynn and Gonzo combined for 46. That should tell you something. Teams generally win more games than their QS total. We were 61 QS last season and 75 W. If you add the extra 31 QS to our 61 we end with 91 QS . Obviously we don't know how those SP would have done in Balto,  but it's safe to say a lot better and most likely we make the playoffs.

[OFFNY]

47 minutes ago, El Gordo said:

 

Ubaldo, Tillman and Miley combined for 15 QS. Cobb Lynn and Gonzo combined for 46. That should tell you something. Teams generally win more games than their QS total. We were 61 QS last season and 75 W. If you add the extra 31 QS to our 61 we end with 91 QS . Obviously we don't know how those SP would have done in Balto,  but it's safe to say a lot better and most likely we make the playoffs.

 

o

 

Thank you.

And the effects are not just singular, also.

In addition to the most glaring statistic that you cited (46 Quality Starts vs. 15 Quality Starts), the Orioles' perennially solid bullpen was also extremely overworked this past season because of the historically bad starting pitching forcing them to log more innings than any other team in the Majors except for 3 (the Reds, the Marlins, and the Blue Jays.) 

 

I suspect that the Orioles' bullpen at-large would have performed better than they had with Cobb, Lynn, and Miguel Gonzalez taking the starts of Jimenez, Tillman, and Miley this past season ........ that would have resulted in a considerably reduced workload AND considerably less pressure on them overall.

 

o

[El Gordo]

1 minute ago, OFFNY said:

o

 

Thank you.

And the effects are not just singular, also.

In addition to the most glaring statistic that you cited (46 Quality Starts vs. 15 Quality Starts), the Orioles' perennially solid bullpen was also extremely overworked this past season because of the historically bad starting pitching forcing them to log more innings than any other team in the Majors except for 3 (the Reds, the Marlins, and the Blue Jays.) 

 

I suspect that the Orioles' bullpen at-large would have performed better than they had with Cobb, Lynn, and Miguel Gonzalez taking the starts of Jimenez, Tillman, and Miley this past season ........ that would have resulted in a considerably reduced workload AND considerably less pressure on them overall.

 

o

Cobb had 15 QS, Lynn 18, and Gonzo 13. Our best was Bundy at 13

[Frobby]

22 minutes ago, El Gordo said:

Cobb had 15 QS, Lynn 18, and Gonzo 13. Our best was Bundy at 13

That's not correct.   Bundy had 19 QS.

[El Gordo]

Just now, Frobby said:

That's not correct.   Bundy had 19 QS.

You're right, read the wrong column. 13 W 19 QS. Doesn't alter may central point.

[Frobby]

1 hour ago, El Gordo said:

Ubaldo, Tillman and Miley combined for 15 QS. Cobb Lynn and Gonzo combined for 46. That should tell you something. Teams generally win more games than their QS total. We were 61 QS last season and 75 W. If you add the extra 31 QS to our 61 we end with 91 QS . Obviously we don't know how those SP would have done in Balto,  but it's safe to say a lot better and most likely we make the playoffs.

For what it's worth, here's how we did when we had a quality start, by pitcher:

Bundy 14/19

Gausman 11/15

Miley 5/8

Jimenez 4/6

Tillman 2/3

Hellickson 2/5

Asher 1/3

Ynoa 1/1

Wilson 1/1

Aquino 1/1

All told, that's 42/62 = 67.7%    In the games we didn't have a QS, we were 33/100 = 33.0%.    So, figure an extra 31 QS would have been worth about 10 more wins, before getting into the issue of saving the bullpen, etc.

 

 

[Luke-OH]

8 hours ago, Number5 said:

I'm not sure how this can be categorized as a fact.  Reasonable arguments can be presented for both sides.  They are both useful tools, but neither FIP nor actual ERA should be taken as absolute indicators of how a pitcher will do next year, IMO. 

 

6 hours ago, Frobby said:

Two things:

1.   Have there been any recent studies on this?   I'm aware there was a study at the time of Moneyball, but I'd like to know just how true this is.

2.    Even if the statement is true over the group of all players, there still may be ways of grouping players for whom it's not true.    I don't know if there are such predictable groups or not, and whether Ubaldo would fit in to one of those.     I wonder what the track record is in cases where FIP and ERA diverge by more than a run, for example.    Maybe I'll play around with this over the winter.

There are definetly players who defy FIP or xFIP due to specific abilities.  Those players can usually be identified over large sample sizes.

There are a ton of recent studies, most saying how FIP and xFIP aren't perfect and need to be improved, but still that they function as a better predictor of future performance than ERA alone in the population of MLB pitchers.  Infield fly balls are being added to recent metrics, since this seems to be a skill pitchers have to induce them and they are pretty much as good as a strikeout.

Here is an article about the predictive values ERA, FIP, xFIP among other things.

The author uses individual pitcher's performances on even numbered days as compared to the same pitcher's performances on odd numbered days in the same season.  He did this for each pitcher season between 2003-2008.  He ran these numbers with ERA, FIP, and xFIP at different IP cut offs.  A perfect predictive metric would have the same number on even and odd days since a pitcher's true talent isn't effected by even or odd days of course.  He used "root mean square error" to test the differences, the smaller the RMSE, the smaller the difference between the samples and therefore the better that metric is at indicating true ability (the predictor of future performance).

These are the results, as you can see FIP is lower than ERA at every sample size level and xFIP is lower than FIP at every level.

IP	NUM	ERA	FIP	xFIP	tRA
10	595	3.21	2.73	2.49	2.61
20	595	2.41	2.03	1.87	1.99
30	475	1.89	1.61	1.51	1.57
40	158	1.83	1.46	1.28	1.36
50	117	1.75	1.41	1.35	1.42
60	105	1.18	1.06	1.02	1.03
70	134	1.47	1.28	1.18	1.25
80	160	1.20	1.01	0.94	1.03
90	153	1.07	0.98	0.91	0.95
100	59	1.20	0.85	0.87	0.86
110	23	1.05	0.82	0.76	0.74