Best Reliever In Baseball
June 28, 2011 at 8:22 pm by Ben Duronio under Atlanta Braves
…is not Jonny Venters. It’s probably Craig Kimbrel. This is potentially surprising to some, and gratifying to Braves fans who are able to enjoy the spoils of having arguably the top two relievers in the game on their team, but the numbers do not lie.
Here are the league leaders in fWAR as a reliever with their innings pitched totals attached:
Craig Kimbrel 1.6, 39.0
Jonny Venters 1.4, 49.0
David Robertson 1.3, 31.1
John Axford 1.1, 35.1
Mariano Rivera 1.1, 30.1
There are arguments for other pitchers not on the list, such as Sergio Romo, Joel Hanrahan, Al Alburquerque, or even two-time former Brave Kyle Farnsworth, but Kimbrel’s argument is probably strongest when looking at who the best pitcher out of the bullpen truly is.
Let’s start with Venters. Jonny has been tremendous, and if not for his poor performance on Sunday he would probably sit at 1.7 fWAR — he sat at 1.6 entering Sunday. Even if you take that performance out and add in a base runner free inning, a 1.7 fWAR in 49 innings is not better than a 1.6 fWAR in 39. It’s close, but if you put Kimbrel’s performance in 49 innings he has the higher number. Regardless, Kimbrel’s performance in 10 fewer innings has been more valuable than Venters’ has, according to Fangraphs WAR.
The league leader in xFIP is Sergio Romo, who has pitched only 24 innings and has faced 68 right-handed batters compared to just 23 left-handers. In contrast, Kimbrel has faced 79 righties and 82 lefties. Romo has basically been used as a right-handed specialst, where he has succeeded tremendously. However, it is tough to be labeled “best reliever in baseball” while being used in this fashion. Romos’ splits against lefties are pretty gross (1.028 OPS against), so his usage is not necessarily incorrect. After Romo, the next lowest xFIP is Kimbrel at 2.19.
The same story goes for FIP, where Romo leads and Kimbrel is right behind. Of the four relievers who average over 14 strikeouts per nine, the only pitcher under 5.6 walks per nine is Kimbrel, who sits at 4.15. That’s not a great number, but he has shown the best control out of the other members of the insane strikeout rate club.
This could all change with an outing or two, putting Venters or Robertson in the lead of most categories, but Kimbrel has somewhat quietly become the top reliever in the game. His five blown saves likely have a lot to do with this. Many look at saves and blown saves to determine how effective a closer is, which is a pretty flawed way to look at their production. While Kimbrel’s peripherals are second to none, he has been hurt by a .325 BABIP, which has lead to a few blown saves. That number should sink, which will in turn likely lower his ERA. The walks have also caused a few poor outings, but 18 walks in 39 innings is hardly a legitimate problem combined with Kimbrel’s strikeout rate.
Results wise, meaning runs allowed, Kimbrel has lagged behind a number of relievers in both leagues. His 2.77 ERA ranks 60th in all of baseball compared to his fellow bullpen peers. All things being equal, that number is sure to drop. If he continues to strikeout batters at this rate while allowing under 4.5 walks per nine, he should see his results improve dramatically. Looking purely at results is not the best way to determine who the best pitcher or reliever in baseball is, which is why defense independent stats were created in the first place. For what he can control, no bullpen arm in baseball has been better than Kimbrel.
Loading ...
Wow! I wouldn’t have guess that without looking at the stats. Just out of interest where did EOF come in on the list?
Future form and injuries aside we have finally found long term solutions with our bullpen and 1st base! This bullpen will be great for years to come.
It is impossible to argue with these objective statistics, but I think there will be commenters who will try to do just that.
@Mike/2
I will try to do just that. (Though I’m not sure how to back it up until fielding f/x goes public. Statisticians?)
Obviously, there is no more sure out in baseball than the strikeout. How many of them result in the runner reaching base? A miniscule percentage.
But I’d be curious to know what sorts of batted balls (where the ball is hit, at what speed, in the air/on the ground) have out percentages approaching that of a strikeout. We could call them “error/hole-negligible” outs. Or we could not call them that. Whatever.
Venters, while not at Kimbrel levels, is no slouch when it comes to strikeouts, but I would wager that the types of batted balls he generates are much more likely to be outs than the types of batted balls Kimbrel generates.
I don’t have any statistics to back it up, but fWAR hugely values FIP–and if it assigned a similar value to “error/hole-negligible” batted balls, I think Venters’ fWAR would skyrocket. No offense to Craig, but Big Dick Johnny Venters is the best in the game.
Of course, neither of them has allowed a home run this season. Sick.
Kimbrel allowed an opposite field home run to Hernandez, the catcher from the Reds. I was at that game, and disgusted.
True: Considering only innings pitched, strikeouts, walks, and home runs, Craig Kimbrel has had the best results of any reliever in the majors this year.
Not true: Craig Kimbrel is probably the best reliever in baseball.
I’d take the “over” on 2.77 ERA the rest of the way (and hope I’m wrong).
Why do we hit and run with Uggla? I know it wasn’t actually a h&r, because there was a 3-2 count with 1 out, usually not a bad spot to run, but still, cmon. Its Uggla.
Agreed. Not sure what percentage of Uggla’s at bats end in a strike out or pop-up, but I would venture to say, over the course of the season, that is not going to be a “profitable” decision.
@ 3
I agree – part of the reason why Kimbrel’s BABIP is high is because when hitters make contact, they make solid contact. Whereas for Venters, its usually soft ground balls or bloopers. I don’t recall any of the hits that Venters gave up Sunday were solidly hit (of course that doesn’t excuse the two walks or him not letting that grounder go through to A Gon).
See stats (2011):
1) XBH% – 3.7% CK and 1.6% JV
2) GO/AO – 0.69 CK and 5.33 JV
3) LD% – 19% CK and 7% JV
I’m glad that the Braves have 2 of the top relievers in game, but if I’d have to choose one it’d be JV.
Something I’ve learned recently: LD% is subject to so much scorer bias it’s completely useless.
Yep. I mean, he’s rolled over about 1,346,183,984 outside fastballs to the left side to justify the thought of a H&R.
How about a hit and bunt next time in that situation? At least he’d either put in on the ground and move the runner or foul it off and make an out, but one out is better than two. And if he pops it up, it’s still the same result as a K and a runner thrown out.
I was going to point out my sarcasm in the preceding paragraph, but as I think about, it actually sounds like a better idea. Uggs CAN’T be as bad of bunter as he is a hitter right now.
And it would also fit out FG team philosophy.
our*
@ 9
I see you point as far as bias goes, but I think most who’ve watched the majority of games would agree that Kimbrel has given up more “solidly hit” balls %wise than JV.
What happened to Seattle being a hard place to hit the ball out of the park?
Dan uggla should give back to the braves every dollar he hasn’t earned this season… As of now, if he weren’t already a millionaire, it’d be close to putting him on the streets. Send the man to gwinett and let him get his head straight, please!
There must be a strong wind tonight in Seattle since Uggla just hit one out.
@3!
I am a big fan of both Venters and Kimbrel as they have been extremely good so far this season. The issue which has happened many times before is, can Venters maintain his .230 babip? Last year he was around .280 and did very good while allowing one home run. This year it’s down to .237 with zero home runs. I know Moylan had a zero HR year and I believe Isringhausen did about a decade ago as well, but I believe those are the only two of people with like 40 or more innings.
Last year Venters was very good with sustainable peripherals (except HR rate). This year he’s been better because he’s doing better IN EVERYTHING. Can he hold an 80% GB rate and 85% strand rate? Can he give up zero home runs with a .230 babip over like 85 innings? Maybe, but those are numbers that you usually don’t bank on.
Last season Venters was very good but anyone who was skeptical based it on his history and 2010 being a small sample to predict the future, not because he out performed his numbers.
Kimbrel so far has put up a season like Venters’ last year in that he is doing extremely well BUT ALSO having sustainable rates (outside of HR/FB, and honestly fuck that for relievers) to go along with it. The main difference in Venters last season and Kimbrel this year is Kimbrel had dramatically more hype and pedigree. Venters was basically a failed starter called up to fill the pen (Unless I’m wrong and someone honestly predicted he would become a BP god) while Kimbrel was the “closer of the future”.
@15 Dan actually had a nice swing on that one. He seemed to keep his body closed and just make contact…
Secondly, I know that Alex is pretty good defensively but offensively he has contracted whatever Dan is (hopefully) getting over
Scorer bias shouldn’t be a big deal when comparing Venters and Kimbrel, since they’ve had pretty much the same scorers/stringers.
Kimbrel has allowed six extra base hits in 27 hits allowed. It isn’t like he is getting creamed out there. There have been blown saves with dinks and dunks.
How come FG tells Alex to bunt when he is a hot hitter, but let’s him swing away right now when he is cold? If excess bunting if your philosophy, at least do it somewhat correctly. If there ever was a decent time to bunt (not saying there ever is), that was it because of how cold Gonzo is and how DP prone he is.
Also, I am not a pro hitter, but I did play college ball. Using my experience I’d guess that Alex should have been smart enough to take that slider outside (rather than flail at it) after Pineda just walked 2 straight and threw him a first pitch ball. I guess he was just too excited by not getting the bunt sign for once with runners on, so I can’t blame him.
okay. sherrill coming in i understand, but i dont understand the long leash. and now we’re bringing in proctor. i swear i’ve never said damnit to this many managing decisions in my 15+ years of following braves baseball. Bobby may have done some of the same stuff, but i guess i think he earned the right to do things his way IMO. FG has nowhere near anything close to the experience and success Bobby had for me to not question his decisions, or at least accept them.
at least it worked out okay.
Heyward robbed at the plate.
@2
It is very possible to argue with those statistics. FIP, and therefore fWAR, screw over Venters by ignoring balls in play. Of course if you only look at walks and strikeouts, Kimbrel is going to look better. Venters’s best weapon is his ridiculous sinker, which induces crap contact. Yes, balls in play depend in part on the defense behind the mound, so it is hard to compare two pitchers without ignoring them, but that doesn’t make Kimbrel better than Venters, it only makes it easier to measure Kimbrel’s primary strength (the strikeout) than it is to measure Venters’s primary strength (the bad-contact ground ball). Other stats that don’t use FIP like BR WAR rank Venters above Kimbrel.
@25
Unless it’s been changed, my main complaint with BR WAR is that both pitching and batting use a team component. For Batting I believe they have a hitters WAR representative of his creation of that team’s total runs, while their pitching WAR is basically Runs allowed modified by a team defensive adjustment.
Pitching especially is a problem because it adjusts based on the team’s overall defensive ability, not on the defensive capacity of the players involved in the plays made.
Imagine Venters getting ground balls every hit to Alex Gonzalez, and imagine all the hits going to Dan Uggla. Gonzalez is better defensively and more likely to make the play. Uggla will get to some, miss some or misplay some. BR WAR would actually rate Venters worse should runs score because of a problem by Uggla instead of Gonzalez, because of the aggregate team defensive adjustment.
Fangraphs takes all potential outside influences out of the equation. Is it perfect? No, but I am not familiar with a way to factor in batted ball profile to get a sensible outcome, OR the true effect that batted ball profile has on the pitcher’s ability.
Jeff, I’m pretty sure I just outlined my wish-it-existed-so-I-could-prove-how-ballin’-JV-is batted ball statistic above. It would iron out all of that specific fielder crap and not be subject to the stupid line drive scorer stuff Peter mentioned.
The idea is to categorize all batted balls by a few factors: where they’re going, how fast, at what kind of trajectory. You group together similar (how similar is a question of subjectivity) batted balls and find out how many of them resulted in outs or not, and then make a percentage. Some percentages will approach 100, others won’t. For a pitcher you can turn this into a performance indicator rather than a results metric by comparing the types of batted balls he generates with historical batted ball out percentages.
If this were ever carried out, yhere would be some sort of normal range for major league pitchers. But I’d guess that Johnny Venters would be an outlier on the “good” side of things.
The pitfalls of fWAR are so well documented that I have trouble taking an article seriously that focuses on nothing else. When your creator feels the need to defend his stat publically over and over, and still the best he can do is say “we think its better than some other measures” then you really can’t trust it as the end all be all. Strikeouts, walks and homeruns. That is all fWAR tells us. It completely ignores the fact that there may be luck involved in those three things (especially in small sample sizes where batter sequencing, left/righty matchups, umpire calls, etc. can have a huge impact on overall stats), and ignores all non-luck aspects balls in play. Dave Cameron and the fangraphs guys are really smart, but even they admit that FIP is not supposed to be used on its own.
I see you point as far as bias goes, but I think most who’ve watched the majority of games would agree that Kimbrel has given up more “solidly hit” balls %wise than JV.
I don’t trust your eyes/brain to capture everything and you shouldn’t either.
Well first that is partially I assume that CAC meant with scorer’s judgment on certain balls. Even people that do things like total zone and UZR don’t agree on some of this stuff.
Field f/x, I believe, would serve a great way of measuring this, but is not fully implemented, nor will it be easily accessible once it is.
Anytime you are recording batted ball profiles and measuring their ability to be turned into outs, you are also measuring the fielder’s defense as well which adds noise to the outcome. In any situation under those guidelines, a better fielding team will make a pitcher look better than a poor fielding team. It doesn’t make the pitcher bad or even worse than a high k pitcher, but the objective is to best measure individual value away from external influences.
It is very difficult to measure only a pitcher when you are taking fielded outs into the issue. If you find patterns in how a pitcher gets his struck balls to go, isn’t that less pitcher value and more of a scouting report? If Hudson gets ground balls directly down the third base line on 60% of his RHB ground balls, does that apply value to Hudson over having them go between short and third, or is that more of information the fielders use to better field their zones? Unless Hudson could reliably keep the ball away from the bad fielders and force it closer to the good fielders, I do not know how that could be seen as anything more than a fielding doing a thorough job.
where they’re going, how fast, at what kind of trajectory
All we need is angle, speed, and landing spot. They try to classify them by binning them into ground ball/line drive/fliner liner/fliner fly/fly ball/pop up right now, but they do such a poor job of it we might as well just wait until hit f/x before we start fucking around with batted balls.
Venters isn’t going to keep inducing a ground ball 80% of the time a batter puts it in play, nor will batters continue to hit .329 off Kimbrel when they put it in play. You can argue about whether good/bad luck caused this or something else, but it’s not going to continue. It will regress to the mean in both cases.
Venters has probably been better than Kimbrel so far, but in terms of true talent level you could make an argument for either one.
CAC/32
Your last line there summed it up perfectly for me. The way I see FIP and therefore fWAR, it measure who would be better going forward, not necessarily who has had more value to the team thus far. I think you end up tying yourself in knots unnecessarily by trying to argue that Kimbrel has had more value to the Braves so far than Venters because he has a 1.7 vs 1.5 fWAR even though their ERA’s are quite a fair way apart. (Yes I know ERA isnt’ the best measure of value).
Another way I like to look at value for a team is WPA, and there Jonny Venters leads the team with 3.05, and Kimbrel is at 0.67. So I think its fair to say that Venters has outperformed Kimbrel so far, but Kimbrel’s projects to be the better reliever going forward. At least that’s how I understand it.
On a side note, I have always found it slightly puzzling that for valuing pitchers, less is more (just focus in K, BB and HR) yet for hitters we focus on everything. We don’t value hitters just by looking at K, BB and HR, so I don’t think it would hurt to add more information when looking at pitchers value. My 2 cents.
@29
fWAR is used to attempt a pitcher’s true talent level, i.e. the things pitchers have the most control over. Defense is already taken into account for for the position players. Why judge the talent level of the pitcher for the defense behind him?
FIP is a good judgement. The median FIP in 2011 is 3.89. The median ERA in 2011 is 3.91.
@Jeff/30
“Anytime you are recording batted ball profiles and measuring their ability to be turned into outs, you are also measuring the fielder’s defense as well which adds noise to the outcome. In any situation under those guidelines, a better fielding team will make a pitcher look better than a poor fielding team. It doesn’t make the pitcher bad or even worse than a high k pitcher, but the objective is to best measure individual value away from external influences. ”
Right, which is why my made-up measurement would not take into account the defensive result of the play specific to any instance, only how often similar batted balls have historically been turned into outs. The noise you’re talking about should correct itself in the volume of using the entire history of out percentages on types of batted balls over lots of games from lots of seasons from lots of teams both good and bad on defense.
And yes it could be good for scouting, too? I’m not sure why that’s a problem when we’re talking about statistics used to project a pitcher’s future performance.
@CAC/31
You’re probably right. I don’t know enough about what fielding f/x is able to do on its own without human influence. But I do think it’s worth classifying a groundball as a groundball. I haven’t figured out all the words to explain why yet.
“It will regress to the mean in both cases.”
What mean? Has Kimbrel even thrown 70 innings yet in the majors? Has Venters thrown 80? We’re talking about two extreme pitchers (so far). How useful are the historical means for these guys?
None of those questions are posed rhetorically.
But I do think it’s worth classifying a groundball as a groundball.
Sure, if they actually classified them correctly. There’s tons of hit/out bias at the GB/LD boundary. If it’s a hit, they classify it as a LD, if it’s an out, they classify it as a GB.
How useful are the historical means for these guys?
http://en.wikipedia.org/wiki/Regression_toward_the_mean
Im sure like me that most Braves fans think about the Tex trade a couple of years ago how solid this team could have been for years to come with Elvis Andrus at short and the addition of Nafital Feliz in the bullpen
I agree, I think Kimbrel is the best reliever in baseball, at least going forward. When you’re getting nearly 2 K’s per inning, you will be dominant. That’s a great equalizer for allowing an occasional walk or hit.
I think Kimbrel is still learning to pitch too, so those 5 blown saves are part of the process to figure out what to do when you don’t have your best stuff or location on a particular night.
Jacob/37,
I don’t know if Neftali Feliz would ever have pitched out of the bullpen if he hadn’t been traded.
@28
There can be luck in things like bb rate/k rate and home run rate in the same way there is luck in everything that happens in the game. FIP and xFIP are typically better arebecause they limit what they measure to things you want to know. I would not consider lefty/righty matchups to be “luck” and umpire calls have to be considered washed out in any specific sample otherwise the basis of all baseball stats goes out the window. Umpiring has to be considered mostly a constant otherwise all baseball stats go out the window as useful.
@35
To include batted ball profile into the value of a pitcher, you have to prove that what he does is a skill and that it has value. Groundball rate is mostly seen as a skill, but it’s value is tied to the quality of the defense behind him. The theory is that other than the “type” of hit (FB/GB/LD) the pitcher does little to judge where it goes, and I am willing to assume no one has found a pitcher reliably able to get the balls to go to certain places more than can be considered the hitter’s influence.
FIP and xFIP do most likely under value ground ball pitchers in favor of power pitchers, but that is because of trying to measure something which has great influences beyond the pitcher’s known control.
Pitchers don’t strike batters out at will either. They throw in a way that tends to strike batters out. Tell Craig Kimbrel to strike the next batter out and there’s no guarantee he’d do it–there’s luck there, too. Still, Kimbrel is certainly a strikeout pitcher. It’s a tendency. I’d like to look for a tendency of pitchers to induce high out-percentage batted balls. Again, it would have nothing to do with what that specific defense does behind the pitcher on that specific batted ball, but everything to do with what all defenses in the past have done with like batted balls.
If we find that some pitchers do tend to induce high out percentage batted balls at a much greater rate than other pitchers over a large sample size, I see little choice but to call that a skill akin to the “skill” of striking batters out. Just because we haven’t measured something yet does not mean it does not exist.
It could all be baloney, but wouldn’t it be neat to find out?