It is a poorly run baseball team that sees two months of baseball and panics, throwing two years of performance out the window to tear down a rotation. I would be incredibly surprised if there is a feeling of desperation in the front office right now, or even in the clubhouse. Frustration? Absolutely. These days it seems like every line drive finds a glove on offense and every breaking ball finds a barrel on defense.
However, it is perfectly understandable for people to be upset about the outcomes the Indians have received from the back of their rotation. Trevor Bauer and Josh Tomlin own disappointing records and concerning ERAs; the recent complaints of fans are directed at these two, the former in particular. A meme is developing in the fan community that if the Indians want to contend in the postseason — or even make the postseason — then they need to trade for a “quality number three starter”.
I’m not arguing that the Indians need to play better than they are right now in order to have a shot at returning to the World Series. However, when it comes to pitching, I will say this...
The Cleveland Indians already own the starting pitching help that they need
For a while now I’ve had a hunch that the general dominance of Kluber and Carrasco keeps the efforts of the rest of the rotation in a perpetual and unfair shadow. After the barrage of tweets suggesting that Bauer can’t cut it as a starter; that the Indians need to trade someone, anyone for another starter; that the Indians only have two good starting pitchers, I sat down and starting crunching the numbers.
I chunked all of the starting pitchers with at least 30 IP as of Saturday and put them into buckets: 1-30, 31-60, 61-90, 91-120, and 121-150, sorted by fWAR. If talent distribution in Major League Baseball was perfect, we would expect each team to have one starter in each of these buckets. In other words, these are the “ideal” No. 1, No. 2, No. 3, No. 4, and No. 5 pitcher distributions.
Then, I compared each of these buckets to the actual distribution in Major League Baseball. The Los Angeles Dodgers, for example, have three “Ace” quality starters based on 2017 numbers to date. This means they are ranked in the top 30 of all pitchers in baseball. The Indians have two, and the Baltimore Orioles have zero. If you would like to see the full spreadsheet, you are more than welcome to click this link. I will pull from there and highlight the relevant data in tables here.
So, what does a quality No. 3 starter even look like?
Here’s what the average third starter on a baseball team would look like if talent distribution by fWAR was perfect. In other words, your “Ideal” No. 3 starters:
They would have about .8 fWAR, a losing record, 71 IP, and an ERA of 4.46. By xFIP- (take xFIP, set the mean to 100, then each “point” in either direction is a percentage point better or worse than average, with lower being better), they would be almost exactly average overall among pitchers at 99.2.
What does the average No. 3 starter actually look like, and how does that compare with the “Ideal” No. 3?
|Difference vs Ideal||0.04||-0.47||0.50||0.70||0.20||-0.10||-0.08||-0.15||-0.19||-0.26||-5.97|
As you can see, the actual No. 3 starters are worse than their hypothetical counterparts in every way except for the average number of wins they have. In other words, in every way that actually matter for judging true talent, real number threes are slightly below average compared to other pitchers.
You’ll notice that there are only eleven actual No. 3 starters through Saturday that had accumulated more than 1.0 fWAR. Trevor Bauer happens to be one of them.
By fWAR, then, insinuating that the Indians need to find a quality No. 3 pitcher because they don’t currently have one is a little bit crazy. It becomes even more so when you see his xFIP-. This criteria actually likes Bauer enough to suggest that he is an average No. 2 starter, whether you consider the ideal or actual rankings.
|Diff Bauer vs. Actual #3||0.53||1.50||1.70||10.71||2.92||0.24||0.15||0.92||-0.43||-0.94||-21.17|
|Diff Bauer vs. Ideal #3||0.49||1.97||1.20||10.01||2.72||0.34||0.23||1.07||-0.23||-0.68||-15.20|
|Bauer vs Actual #2||-0.05||1.10||1.93||7.40||1.78||0.60||0.43||1.70||0.36||-0.32||-6.90|
|Bauer vs Ideal #2||0.04||0.83||2.00||7.08||2.12||0.31||0.46||1.47||0.21||-0.53||-11.60|
It’s worth recalling at this time that I am assessing pitchers based on fWAR and xFIP statistics. These statistics generally only credit pitchers for things that they are able to control. My understanding is that there is some wiggle room when it comes to whether or not pitchers are better or worse than peers at inducing weak contact (we’ll get to that with Tomlin). Nonetheless, it suggests that Bauer’s ERA is due for regression, and that the outcomes the Indians experience as a result of his pitching performances will only improve.
In fact, as of Saturday, there are very few pitchers who are as poised as Bauer for positive regression. He has the ninth-largest negative disparity between his ERA (5.53) and xFIP (3.65) in all of baseball.
So, when you say that the Indians need to add a quality third starter to its rotation if they want to be competitive, my response is that they already have one. We can expect games in which Trevor Bauer starts the rest of the season to yield better results than we’ve seen up to this point in the season.
Bauer isn’t the only one with an interesting disparity like this. The Indians might actually have two “quality” number three starters.
Josh Tomlin is better than you think
I’ll wait for you to stop yelling.
Okay, thanks. I want to start with the numbers, and then we can take a look at some of the narratives surrounding Josh Tomlin.
|Diff Tomlin vs. Actual #3||0.03||-0.50||3.70||4.81||-0.71||-2.48||0.38||1.46||-0.16||-0.50||-11.17|
|Diff Tomlin vs. Ideal #3||-0.01||-0.03||3.20||4.11||-0.91||-2.38||0.46||1.61||0.04||-0.24||-5.20|
|vs Ideal #4||0.42||0.43||3.70||15.86||-1.16||-2.96||0.14||1.13||-0.57||-0.68||-15.27|
|vs Actual #4||0.39||0.30||3.97||14.52||-0.65||-2.44||0.12||1.04||-0.56||-0.53||-11.73|
Pretty wild, no? Again, we see a situation where the on-field performance of the pitcher, based on the advanced metrics, does not match the on-field outcome. It’s a little more stark with Tomlin than it is with Bauer; Tomlin has the seventh greatest difference between ERA and xFIP in all of baseball.
As is well-documented, the Achilles heel of Tomlin is his tendency to give up home runs early and often. The eyewitness account from fans is that he’ll give up a couple of singles and then a long ball, and by that point it’s just too late for the Indians to do anything.
This season, Tomlin’s HR/9 is 1.66. It is the lowest it has been in since 2014.
Another narrative forming about Tomlin is that because he throws so many strikes — he is perpetually the league’s lowest in BB/9 — it is easy for players to make solid contact and put the ball into play against him. If this were true, we’d expect Tomlin to have a slightly higher BABIP against hitters than the rest of the league.
This season, it’s sitting at .353. His career number, including this season, is .284. For reference, Bauer is in a similar boat; hitters this season ‘bip’ .340 off of him, contrasting with a career number of .297. I don’t want to assign blame for this or try to find a reason as to why the numbers are higher so far this year. I doubt there is one. I suspect the number will regress for both pitchers as xFIP and career values suggest it will.
So far we’ve determined that concerns about both Trevor Bauer and Josh Tomlin are likely overstated. What we’ve looked at, however, is the production of these pitchers when compared to league average. If we want to understand how the Indians chances to reach the postseason look, it might be better to take a look at their production when compared to likely playoff teams.
The Indians have enjoyed production from starters comparable to other likely playoff teams
For this exercise, I elected to take the top three starters from every team and compare them. We expect this group on any team to eat the lion’s share of innings during the regular season, and in some circumstances they may start every single game in the postseason. Here are the results, ranked by fWAR again.
|Indians vs. Average||-0.05||0.53||0.47||-0.62||0.82||-0.04||0.10||0.56||-0.06||-0.24||-5.00|
The Indians are right in the middle of the pack. When fans plead for the Indians to trade for a third starter, they do so because they do not believe the top end of our starting rotation is competitive. These numbers suggest that they are wrong.
Now, one thing we might want to consider is whether or not the Indians should seek to add a fourth starter, relegating Tomlin back to “fifth” status. This is a slap in the the face of Danny Salazar, who is also struggling with bad luck this season. There are other reasons to be concerned about Danny. We expect him to be a well above average No. 3 starter on this team, and at a fringe-ace level. When this is true, it is precisely why the Indians are so devastating; they deploy five starters, none of whom are worse than the average No. 3 starter across baseball.
This season, here’s how Salazar looks compared to both No. 3 and No. 4 pitchers, actual and Ideal:
|Diff Dzar vs. Actual #3||-0.37||-1.50||0.70||-18.29||5.18||1.63||0.60||0.89||0.26||-0.87||-20.17|
|Diff Dzar vs. Ideal #3||-0.41||-1.03||0.20||-18.99||4.98||1.73||0.68||1.04||0.46||-0.61||-14.20|
|Diff Dzar vs. Actual #4||-0.01||-0.70||0.97||-8.58||5.24||1.67||0.34||0.47||-0.14||-0.90||-20.73|
|Diff Dzar vs. Ideal #4||0.02||-0.57||0.70||-7.24||4.73||1.15||0.36||0.56||-0.15||-1.05||-24.27|
It’s clear by the numbers and the eye test that Salazar is struggling this season. His walk rate is up, as well as his HR/9 over previous seasons. Taken together I see this as command issues. However, his stuff is just as nasty as ever, as his K/9 is the highest in baseball among all starters through Saturday. For that reason, it’s also clear by the numbers that luck isn’t doing him any favors.
Regardless, Bad Danny Salazar is about average for a No. 4 starter in Major League Baseball so far this season. Even if he doesn’t fix his command issues or work ethic issues or whatever it is that’s “wrong” with him, he’s due for some regression. If he does fix those problems, the rest of the American League is in an awful lot of trouble.
Remember how we talked about Bauer and Tomlin being quite unlucky, according to xFIP? Well, among starters, Salazar is 13th in difference between ERA and xFIP. His career BABIP is .310; this season it’s been .366. That’s three Indians starters among the unluckiest baker’s dozen in all of baseball.
- The Cleveland Indians are currently in first place in the American League central with a 40-35 record.
- They’ve done this despite having on-field results from three starting pitchers that are well below what advanced metrics tell us to expect.
- If you still believe the Indians should add a starter to compete with likely playoff teams, you’re ignoring that the Indians already compare favorably in advanced and traditional stats with likely playoff teams. That’s before we consider likely regression.
- A trade for a “quality third starter”, which is likely to come from another contending team with a strong incentive to keep that talent, will probably cost the Indians Jason Kipnis or Carlos Santana, along with some other pieces.
- The Indians still have the best bullpen in the American League, and possibly all of baseball, to back these starters up throughout the rest of the season.
Based on my analysis of the numbers, it is clear to me that the Indians starting rotation is perfectly fine. Barring injury, the performance we can expect from the Indians rotation will compare well with the best in all of baseball. Even with injury, this team proved it can still shoulder the load last season.
Am I suggesting that the Indians stand completely pat at the deadline? Not at all. Any team heading toward the playoff should see where it can improve before the end of the season. What I am stating is that starting pitching is not going to be a spot at which the Indians will need help.
Possible reasons to be skeptical
xFIP expects the HR/FB rate to be between 9/10 percent every season, and so it regresses FIP to account for that. Right now, that rate is 13.8 percent, which is easily the highest on record. Last season was the previous record at 12.8 percent. This means that results are up around the league, but perhaps that skews the data somehow.
The other thing that might bork the numbers is if defensive independent pitching stats somehow don’t understand the Cleveland Indians. Corey Kluber went through a stretch where his peripherals didn’t match the result on the field. If Tomlin and Bauer allow a different kind of contact or take a special sort of approach that exempts them from using DIPS like xFIP, that would be remarkable.
We’re also working with a somewhat small sample size, although that’s by design. The people clamoring for a trade are basing that only on what they’ve seen this season. What I’d like to do is revisit this in a couple of months to see whether or not the prediction is correct. If the numbers do not lie, we are likely see positive regression from Bauer and Tomlin (This is also true of Danny Salazar, whom we have mostly ignored in this post). It’s possible that randomness will continue to reign and the numbers don’t improve. That would be both bad and unlikely.
The last thing to consider is that I’m pulling all of these numbers from FanGraphs. They calculate value a little bit differently than other stats houses. It’s possible that they overvalue the Indians staff by some quirk of the way they evaluate pitchers. Baseball Reference is quite a bit less enthusiastic about Josh Tomlin, and a little less so about Bauer and Salazar. Baseball Prospectus’s DRA stat, however, very much agrees with what we’re seeing from FanGraphs.
Long story short: DRA controls for even more things than xFIP. Here’s where the DRAs stand for each pitcher:
Salazar: 3.72, DRA- 79
Bauer: 4.05, DRA- 86
Tomlin: 7.29, DRA- 155
For Salazar and Bauer, Baseball Prospectus corroborates our expectation that regression will be kind moving forward. For Tomlin.... woof. I’m not surprised to see that he’s the pitcher for whom analytics draw the most divergent conclusions. He’s a unique player. It’s not often you see someone with such exquisite command and control also lead the league in dingers allowed.
I look forward to revisiting the numbers again next time. What I’d like to do is gather up the various WAR statistics and take an average of them for each pitcher in order to get the rankings. In political polls, a “poll of polls” tends to be more accurate than any one poll; I imagine it will be the same for WAR and the various DIPS in baseball. I’d also like to leverage the community to see if anyone has suggestions about various regressions or other statistical analyses to run on the data set. Clustering is included in the stats spreadsheet, courtesy of a friend, but there could be much more to do with it.
With a little luck we’ll be sitting around in front of our computers commenting about how pretty the Indians are sitting with a ten-game lead in the AL Central and five ace-caliber starters.