Average MLB Starters and How Braves Compare

[Managuarantano's Volunteers]

About how good do our starters need to be this year to make the playoffs, and how good were they last year?

I wanted to do a comparison of our starters with the rest of the league for last year, 2017, to check what realistic expectations are for our 1/ 2/ 3/ 4/ 5 starters this year as well as the next few years if we want to be division winning or in the wild card hunt (within 4 games of wild card, because 5 games added 3 below 0.500 teams in the AL). (I also did numbers for the rest of the league and the AL while I was at it)

Methodology: Basically the same as this article from 2006 by Jeff Sackmann, from which I also got the idea: https://www.fangraphs.com/tht/how-good-is-your-4-starter/

Quick summary: Looking at ERA since it is a measure of actual performance. Assuming 32 starts per starter, and assigning them to the best starters by ERA (If a guy starts 8 games with a 2 ERA, I will give him 8 starts of a #1 starter, followed by 24 starts of the guy with the next best ERA). I also cut two starts from the worst starter of each team to make math work out – 162/5= 32.4 so you have to remove two starts (was not mentioned in the Fangraphs article so I do not know his methodology). I also added xwOBA because it's a measure of expected performance on suggestion from Enscheff, though these calculations may be slightly off as I cut the two worst starts based on ERA from every team rather than xwOBA (the starts likely would be bad xwOBA too if they were ~16 ERA on already small differences, so the error from this is probably minimal).

Ex for our 2017 Braves on ERA (the numbers to the right of ERA are ERA x Games Started so that the average remains proportional, and the average is the sum of those divided by 32):

Name/ GS/ ERA
Max Fried/ 4/ 3.44
R.A. Dickey/ 28/ 4.26

Average 1st starter: 4.16

Name/ GS/ ERA
R.A. Dickey/ 3/ 4.26
Jaime Garcia/ 18/ 4.30
Sean Newcomb/ 11/ 4.32

Average 2nd starter: 4.30

Name/ GS/ ERA
Sean Newcomb/ 8/ 4.32
Julio Teheran/ 24/ 4.49

Average 3rd starter: 4.45

Name/ GS/ ERA
Julio Teheran/ 8/ 4.49
Mike Foltynewicz/ 24/ 4.74

Average 4th starter: 4.68

Name/ GS/ ERA
Mike Foltynewicz/ 4/ 4.74
Luiz Gohara/ 5/ 4.91
Lucas Sims/ 10/ 5.79
Matt Wisler/ 1/ 6.00
Bartolo Colon/ 12/ 8.14

Average 5th starter: 6.41 (Man Bartolo was horrible)

Ex for our 2017 Braves on xwOBA:

Name/ GS/ xwOBA
Luiz Gohara/ 5/ 0.295
Jaime Garcia/ 18/ 0.309
R.A. Dickey/ 9/ 0.322

Average 1st starter: 0.310

Name/ GS/ xwOBA
R.A. Dickey/ 22/ 0.322
Julio Teheran/ 10/ 0.330

Average 2nd starter: 0.325

Name/ GS/ xwOBA
Julio Teheran/ 22/ 0.330
Sean Newcomb/ 10/ 0.332

Average 3rd starter: 0.331

Name/ GS/ xwOBA
Sean Newcomb/ 9/ 0.332
Mike Foltynewicz/ 23/ 0.337

Average 4th starter: 0.337

Name/ GS/ xwOBA
Mike Foltynewicz/ 5/ 0.337
Lucas Sims/ 10/ 0.340
Matt Wisler/ 1/ 0.351
Bartolo Colon/ 12/ 0.365
Max Fried/ 4/ 0.366

Average 5th starter: 0.353 (Man, Bartolo was still terrible)

With Methodology more or less explained, let's move on to the general results.

For ERA:

MLB/ NL/ AL
1st starter: 3.32/ 3.34/ 3.30
2nd starter: 3.92/ 3.91/ 3.94
3rd starter: 4.38/ 4.28/ 4.47
4th starter: 4.99/ 4.84/ 5.15
5th starter: 6.01/ 5.92/ 6.10

For NL playoff race:

Division Winner/ Wild Card or Close/ Other
1st starter: 2.60/ 3.25/ 3.67
2nd starter: 3.05/ 3.64/ 4.36
3rd starter: 3.52/ 3.99/ 4.71
4th starter: 4.24/ 4.40/ 5.28
5th starter: 4.98/ 5.14/ 6.66

For AL playoff race:

Division Winner/ Wild Card or Close/ Other
1st starter: 2.88/ 3.11/ 3.47
2nd starter: 3.36/ 3.84/ 4.13
3rd starter: 3.61/ 4.31/ 4.76
4th starter: 4.46/ 4.81/ 5.43
5th starter: 5.16/ 5.98/ 6.40

Lines for starters (if below the line they are considered a starter of the better number, i.e. 2.50 is a 1st starter/ace, 4.00 is a 2nd starter, etc.):

MLB, NL, AL
1st/2nd: 3.62, 3.62, 3.62
2nd/3rd: 4.15, 4.09, 4.20
3rd/4th: 4.68, 4.56, 4.81
4th/5th: 5.50, 5.38, 5.62

For xwOBA:

MLB/ NL/ AL
1st starter: 0.293/ 1st: 0.291/ 0.296
2nd starter: 0.315/ 2nd: 0.311/ 0.319
3rd starter: 0.328/ 3rd: 0.323/ 0.333
4th starter: 0.340/ 4th: 0.344/ 0.344
5th starter: 0.362/ 5th: 0.367/ 0.367

For NL playoff race:

Division Winner/ Wild Card or Close/ Other
1st starter: 0.270/ 0.291/ 0.298
2nd starter: 0.285/ 0.303/ 0.323
3rd starter: 0.297/ 0.315/ 0.338
4th starter: 0.312/ 0.328/ 0.349
5th starter: 0.345/ 0.341/ 0.369

For AL playoff race:

Division Winner/ Wild Card or Close/ Other
1st starter: 0.263/ 0.289/ 0.307
2nd starter: 0.301/ 0.311/ 0.326
3rd starter: 0.310/ 0.331/ 0.340
4th starter: 0.324/ 0.337/ 0.351
5th starter: 0.348/ 0.363/ 0.374

Lines for starters (if below the line they are considered a starter of the better number, i.e. 0.250 is a 1st starter/ace):

MLB, NL, AL
1st/2nd: 0.304, 0.301, 0.308
2nd/3rd: 0.321, 0.317, 0.326
3rd/4th: 0.334, 0.330, 0.338
4th/5th: 0.351, 0.347, 0.355

Analysis

For 2017, based on the "lines" above for ERA, we essentially had 3 3rd starters, a 4th starter, and a 5th starter performances (MLB and NL). We only had one starter position better than the division winners' 4th starter average, and we had 2 positions better than the wild cards' 4th starter average (all NL).

Based on xwOBA, we would expect a 2nd starter, a 3rd starter, 2 4th starters, and a 5th starter from the batted ball profiles of our pitchers last year. We only had one starter position better than the division winners' 4th starter average, and we had 1 position better than the wild cards' 3rd starter average and 1 better than its 4th (all NL)

So far this year (super small sample size alert - 5 starts per position), we have had 3 1st starter position performances and 2 3rd starter performances.

As you might guess, our top position is better than division winners' 1st starters (five starts from Soroka, Sanchez, Folty have 1.97 ERA averaged), our 2nd is right in line with the first starters (Folty and McCarthy 2.64), our 3rd position is better than division winners' 3rd starters (McCarthy and Wisler 3.35), and our 4th is right at division winners' 4th starters (Newcomb 4.23), and our 5th is better than division winners' 5th starters (Teheran 4.50).

Compared to Wild Card teams from last year, we are also better so far (I expect heavy regression toward the Wild Card team area): we roughly have had 2 first starters, a second starter, and 2 fourth starters.

Based on xwOBA in all the MLB data set, we have had a 1st starter so far in Newk (0.290), a second starter in Folty (0.316), a third starter in Folty/McCarthy (0.325), a 4th in Teheran (0.342), and a 5th in Soroka/Sanchez/Wisler (0.387).

Comparing us to division winners, we have a second starter in Newcomb, a 4th starter in Folty as well as the Folty/McCarthy defacto 3rd spot, a 5th starter in Teheran, and a worse than 5th in Soroka/Sanchez/Wisler.

Comparing us to the wild card race from last year, we have a 1st starter, a 3rd starter, a 4th starter, a fifth starter, and one worse than 5th.

Keep in mind, however, that this is a very limited sample, and I expect some starters to do better (Soroka) and others to do worse (Sanchez) as the season progresses. Our xwOBA however does look pretty good so far, especially if we can get some softer hit balls from the 5th starter position. ERA very obviously looks good so far.

General Comments

1. Kinda disappointed that my results for ERA were basically the exact same as Sackmann in 2006. I had thought ERA would be worse in the wound ball era, but I was wrong in comparison to 2006.

2. Let me know if you guys want me to do something else with the numbers or have ideas for the analysis. I am definitely open for suggestions on how to improve it, etc.

3. I'll continue to keep up with how our starters are doing throughout the season in relation to this.

4. Please use this for other data analyses if you would like.

References:

Jeff Sackmann for his article in the Hardball Times in 2006: https://www.fangraphs.com/tht/how-good-is-your-4-starter/

Fangraphs database for starters' ERA in 2017

Baseball Savant for xwOBA for 2017 (thanks to Enscheff for the idea/reference to this)

 

[Enscheff]

Awesome. This post will be what I reference whenever someone talks about so-and-so being a TOR, or a #3, or any other spot in the rotation.

I love that you broke it down as all teams vs playoff teams.

Average 1st starter: 0.310

Name/ GS/ xwOBA
R.A. Dickey/ 22/ 0.322
Julio Teheran/ 10/ 0.330

Average 2nd starter: 0.325

Name/ GS/ xwOBA
Julio Teheran/ 22/ 0.330
Sean Newcomb/ 10/ 0.332

Average 3rd starter: 0.331

Name/ GS/ xwOBA
Sean Newcomb/ 9/ 0.332
Mike Foltynewicz/ 23/ 0.337

Average 4th starter: 0.337

Name/ GS/ xwOBA
Mike Foltynewicz/ 5/ 0.337
Lucas Sims/ 10/ 0.340
Matt Wisler/ 1/ 0.351
Bartolo Colon/ 12/ 0.365
Max Fried/ 4/ 0.366

Average 5th starter: 0.353 (Man, Bartolo was still terrible)

With Methodology more or less explained, let's move on to the general results.

For xwOBA:

MLB/ NL/ AL
1st starter: 0.293/ 1st: 0.291/ 0.296
2nd starter: 0.315/ 2nd: 0.311/ 0.319
3rd starter: 0.328/ 3rd: 0.323/ 0.333
4th starter: 0.340/ 4th: 0.344/ 0.344
5th starter: 0.362/ 5th: 0.367/ 0.367

For NL playoff race:

Division Winner/ Wild Card or Close/ Other
1st starter: 0.270/ 0.291/ 0.298
2nd starter: 0.285/ 0.303/ 0.323
3rd starter: 0.297/ 0.315/ 0.338
4th starter: 0.312/ 0.328/ 0.349
5th starter: 0.345/ 0.341/ 0.369

For AL playoff race:

Division Winner/ Wild Card or Close/ Other
1st starter: 0.263/ 0.289/ 0.307
2nd starter: 0.301/ 0.311/ 0.326
3rd starter: 0.310/ 0.331/ 0.340
4th starter: 0.324/ 0.337/ 0.351
5th starter: 0.348/ 0.363/ 0.374

Based on xwOBA, we would expect a 2nd starter, a 3rd starter, 2 4th starters, and a 5th starter from the batted ball profiles of our pitchers last year. We only had one starter position better than the division winners' 4th starter average, and we had 1 position better than the wild cards' 3rd starter average and 1 better than its 4th (all NL)

So far this year (super small sample size alert - 5 starts per position), we have had 3 1st starter position performances and 2 3rd starter performances.

I quoted the parts I'm going to take away from this.

I had SPs roughly graded by xwOBA as follows:

Grade xwOBA
80 0.260
70 0.280
60 0.300
50 0.320
40 0.340
30 0.360
20 0.380

Your work suggests I was generally right, but only if we are considering "contenders". Since the goal is to build a contender, I think that's a reasonable assumption to make.

The next step is to look at including IP into this somehow.

 

[Managuarantano's Volunteers]

Awesome. This post will be what I reference whenever someone talks about so-and-so being a TOR, or a #3, or any other spot in the rotation.

I love that you broke it down as all teams vs playoff teams.



I quoted the parts I'm going to take away from this.

I had SPs roughly graded by xwOBA as follows:

Grade xwOBA
80 0.260
70 0.280
60 0.300
50 0.320
40 0.340
30 0.360
20 0.380

Your work suggests I was generally right, but only if we are considering "contenders". Since the goal is to build a contender, I think that's a reasonable assumption to make.

The next step is to look at including IP into this somehow.

I have all the data in Excel, and it's easy to drag data from Fangraphs into Excel. I pretty much just need an idea on the best way to incorporate IP into the system so it isn't a waste of time (aka so I don't have an "I could do this better" eureka moment midway through).

Good ol' forced vacations due to political unrest giving me some time to mess around.

 

[Enscheff]

And the overall conclusion I draw from your fine work is...

The goal of a WS contender should be a rotation comprised of 4 above average pitchers who can produce the following xwOBA or better:

#1: 0.270
#2: ~0.300
#3: ~0.300
#4: 0.310-0.320
#5: random below average SP

 

[Managuarantano's Volunteers]

One other fun part for me (possibly more interesting for less statsy people) was seeing how wide the variation was in ERA and xwOBA between some teams. Like Dodgers vs Orioles, for instance.

Dodgers ERA/xwOBA
1st: 2.21/.259
2nd: 2.98/.278
3rd: 3.48/.288
4th: 3.88/.295
5th: 4.56/.334

Orioles
1st: 4.24/.323
2nd: 4.66/.342
3rd: 5.46/.354
4th: 6.83/.361
5th: 7.72/.387

That convinced me to break it down into playoff vs non playoff teams really quickly

 

[Managuarantano's Volunteers]

Hey Enscheff, nsacpi, or thewupk (and other stat inclined people), I posted this over at TalkingChop for the heck of it, and Ivan said that it would have been better to do this analysis with xFIP- rather than xwOBA. Is there a specific reason to use one over the other? Or are they similar enough that it doesn't really matter?

 

[Enscheff]

Hey Enscheff or nsacpi (and other stat inclined people), I posted this over at TalkingChop for the heck of it, and Ivan said that it would have been better to do this analysis with xFIP- rather than xwOBA. Is there a specific reason to use one over the other? Or are they similar enough that it doesn't really matter?

I prefer xwOBA, but isolating pitcher performance from the contributions of the defense is the main concern.

I suspect xFIP- would yield very similar results.

I would not have bothered with ERA.

 

[nsacpi]

And the overall conclusion I draw from your fine work is...

The goal of a WS contender should be a rotation comprised of 4 above average pitchers who can produce the following xwOBA or better:

#1: 0.270
#2: ~0.300
#3: ~0.300
#4: 0.310-0.320
#5: random below average SP

I think this is the bottom line. Given where the Braves are right now, we can be a bit relaxed about the 5th starter. Maybe the 4th one too. But because pitchers sometimes get injured late in the season, sometimes that 4th or even 5th starter becomes more important for playoff-bound teams.

Having said that, I like our starting pitching depth. If we end up in the playoffs, I think we should be able to find four pretty good starters from Folty, Newcomb, Teheran, McCarthy, Soroka and Gohara.

 

[Managuarantano's Volunteers]

I prefer xwOBA, but isolating pitcher performance from the contributions of the defense is the main concern.

I suspect xFIP- would yield very similar results.

I would not have bothered with ERA.

Ok, cool. That was kind of what I was thinking.

Yeah, I did the ERA portion more for the heck of it and to show friends/people who aren’t used to the “new” stats yet more than for its intrinsic value.

I think this is the bottom line. Given where the Braves are right now, we can be a bit relaxed about the 5th starter. Maybe the 4th one too. But because pitchers sometimes get injured late in the season, sometimes that 4th or even 5th starter becomes more important for playoff-bound teams.

Having said that, I like our starting pitching depth. If we end up in the playoffs, I think we should be able to find four pretty good starters from Folty, Newcomb, Teheran, McCarthy, Soroka and Gohara.

Agreed, though I think we really ought to have four good starters with the amount we’ve invested in pitching through trades, etc. (Similar to the point y’all usually make)