Home » Articles » Opinion » Craig Kimbrel, Small Samples, and the Causal Fallacy

Craig Kimbrel, Small Samples, and the Causal Fallacy

by Nik Gaur

With the playoffs just over a month away, bullpen usage and consistency will naturally be major discussion points. As Craig Kimbrel has struggled with fastball command since his trade to the White Sox, it is equally natural for questions to arise concerning his role. Many fans are asking whether Kimbrel should be used as a ninth inning-only traditional closer, since current closer Liam Hendriks has proven that he is capable of pitching in any high-leverage situation, regardless of the inning.

Nevertheless, the Kimbrel discussions are still marred by small sample sizes, narrow analyses, and the causal fallacy. Moreover, retroactive concerns about the package the White Sox exchanged for Kimbrel are similarly confounding.

Most of the issues concerning Kimbrel’s usage that I have seen are based on either this tweet or some variation of it. The discrepancy is indisputably striking. Yet, Kimbrel has pitched a whopping 9.2 innings during the 8th inning of games this season. He would need to throw hundreds of innings in both the 8th and 9th in order for one to develop meaningful statistical conclusions without luck heavily factoring in. Of course, this never happens with relievers over the course of a full season, which speaks to how difficult they are to objectively evaluate. Nevertheless, it is premature to conclude that Kimbrel absolutely cannot pitch in any inning other than the 9th based on a 9.2 inning sample, especially when one considers the circumstances of those 9.2 innings.

For example, 3 of the 6 runs Kimbrel has surrendered in the 8th inning of games this season came on Andrew Romine’s three-run home run (shown below) on August 6th — a home run that had a .170 xBA (expected batting average) due in part to its meager 94.2 MPH exit velocity. If that hit were instead a flyout, as it typically would be, Kimbrel’s 8th inning ERA would drop from 5.59 to 2.79.

Additionally, in last night’s game against the Blue Jays, Kimbrel actually pitched well. He gave up a leadoff single on a ground ball that had a .210 xBA — while this is frustrating, it will happen — that’s just baseball. He then got a strikeout, a groundout, issued an intentional walk, and then got another strikeout. Unfortunately, the leadoff single came around to score thanks to an advancement on the groundout sandwiched between two wild pitches.

It is difficult for me to agree that Kimbrel threw a wild pitch because he was pitching in the 8th inning rather than the 9th — in my opinion, his issue is that he has been more prone to yanking his fastball to his glove side since the trade, which is something he can correct mechanically (he has done so several times in his career with respect to this exact problem).

Overall, what we are dealing with here is a mix of small sample analysis and bad luck, which Kimbrel was more than due for. For as dominant as he has been this season, a 0.00 ERA in any inning is simply not sustainable. Once you control for luck and external circumstances, Kimbrel is very likely the same pitcher, or at least close to it, in the 8th and 9th innings.

All of this said, I actually wouldn’t mind if Kimbrel and Hendriks swapped roles. Hendriks is better suited as a multi-inning weapon, even if I do not believe that Kimbrel is a materially worse pitcher in the 8th inning. If they flip roles and Kimbrel converts a save or two, I am sure fans will use that as justification for the belief that he cannot pitch in the 8th inning. No matter what happens with Kimbrel this season, I would be wary of the causal fallacy, which is essentially our regular attempt to ascribe meaning (cause) to events (correlation) where outcomes may be purely coincidental. It’s human nature, but until Kimbrel throws hundreds of innings in the 8th so that we can compare to his lifetime performance in the 9th, there simply isn’t any statistical backing for the fan theories.

We see similar examples all the time, with respect to discussions about hitters’ performances with RISP (runners in scoring position). One that I have seen lately is “Andrew Vaughn doesn’t perform well with RISP, therefore he isn’t as good as people think.” Vaughn has 104 plate appearances with RISP, and would need ~1,000 more before one can judge his outcomes without needing to heavily account for luck — this is, of course, assuming that hitters are often significantly better or worse with RISP in the first place, which is already controversial and rarely supported (over significant samples) by data. Another example I see is “Yoan Moncada is a good player, but his struggles with RISP this year negate a lot of that value.” In reality, Moncada’s .277/.414/.465 (.879 OPS, 147 wRC+) line with RISP is very good.

The above complaints are all sorts of things — most conspicuously, poorly-researched, at least for Moncada. But, more importantly, they are other salient examples of fans attempting to draw conclusions from insufficient samples in order to make sense of what they are watching. One could certainly argue that the causal fallacy is also at play, even if the “correlation” itself is arguably non-existent in these cases. Again, this is all human nature, and I, too, am guilty of it regularly, but it is important to keep these ideas at the forefront of our minds in order to more objectively assess what we experience.

One more thing I’d like to rant about on this fine Tuesday afternoon is how Kimbrel’s performance has caused fans (both of the White Sox and Cubs) to prematurely assess the Kimbrel trade as a whole.

If you’ve read my articles over the years or follow me on Twitter, you know that I am not the type to blindly praise everything the White Sox do. That said, even if Kimbrel continues to struggle with fastball command, or if he somehow regresses to his 2019 form, I still will never be upset with the trade the White Sox made to acquire him. Losing 4+ years of Codi Heuer stings a little, mostly because I still believe he has the potential to evolve into an effective and consistent high-leverage reliever, as he flashed in 2020.

Regardless, the White Sox made a win-now move at the cost (aside from Heuer) of a player who could not provide any value for the remainder of the 2021 season due to his injury. Nick Madrigal is a fine player, and I’m sure I will come off to Cubs fans as a salty White Sox fan in denial, but my concerns about his profile date back to 2018. That the White Sox executed the Kimbrel trade in exchange for their (arguably) eighth-best position player who will be unavailable during an upcoming playoff run will never be considered a bad process by me, no matter the result.

I have even read some comments from White Sox fans who believe that the Kimbrel trade will “haunt” the team, and they have likened the deal to the 2017 Jose Quintana trade. Putting the post-trade performances of Quintana and Kimbrel aside, this is another thing I do not understand. If you are like me, and you have concerns about the sustainability of Madrigal’s offensive profile, then the trade is nothing to lose sleep over.

If you are not like me, and you think Madrigal’s profile is either sustainable or can get better, I still do not understand the Quintana trade analogy. Madrigal’s defensive miscues could improve to some extent — I can see him cutting down on the mental mistakes — but his lack of range and iffy arm strength are unlikely to improve. Perhaps he will stop running into outs on the basepaths, and perhaps he will build on what he flashed in 2021 by continuing to hit for high averages and showing enough power to at least keep outfielders’ positioning in check.

But, Dylan Cease aside, the comparisons to Eloy Jimenez are questionable. One could point to Jimenez’s durability issues, but he has played in a higher rate of MLB games (57%) since his promotion than Madrigal has since his own (46%). His defense in left field is very poor, but even as a DH, it is not at all outlandish to say that Jimenez, like Madrigal, could hit .300 or better (he is hitting .296 since the start of 2020), but also with 40 or more home runs over a full season. Silver Slugger Award winners like Jimenez do not grow on trees, and there are likely still levels of his power and consistency that we have not witnessed. Jimenez and Madrigal are not the same class of player, and this will be made more clear in my opinion if both can stay healthy and play a full season now that they have gotten past their rookie years. If anything, my concern about the trade package would be that Codi Heuer might develop into a shutdown closer in the next few years.

Overall, I do think it is very understandable to have concerns about Craig Kimbrel. His fastball command could use improvement, and even if regular-season results hardly matter for the White Sox over the next month, it would at least feel better to know that he has settled into his role heading into the playoffs. But, despite these factors, fans should also be aware of the fluky nature of small sample size analysis. Further, they should be careful when attempting to ascribe cause to correlation, especially when the correlation itself is so fickle.

And, finally, as for the fans who are already likening the Kimbrel trade to the Quintana trade with respect to Jimenez and Madrigal, I hope Eloy Jimenez can remind you just how good he is throughout the rest of the season.

Follow us @SoxOn35th on social media for more!

Featured Photo: Chicago White Sox / Twitter

Notify of

Most Voted
Newest Oldest
Inline Feedbacks
View all comments

There is confusion in this article, though. No, of course, one cannot draw statistical conclusions based on data. But the data supports concerns that people could have had from the outset. One is that Kimbrel, like most pitchers, can be thrown off mentally. Witness how he faltered with the Cubs before getting back on track. One could wonder if one of the greatest closers in MLB history is mentally untracked by getting asked to play a setup role, when he is still the most dominant reliever in the NL. It >>is<< a mental game. Adding brilliance to brilliance (Kimbrel to Hendricks) could undermine both, counterintuitively.

LaRussa did this 37 years with the White Sox. Lamarr Hoyt, Bannister, Dotson, and Burns formed the most incredible starting rotation in baseball, with a virtually perfect second half in 1983 when the White Sox won a championship (in this case, just the division) for the first time since 1959. So the White Sox bring in Tom Seaver to this brilliant rotation. You can predict what happened. Yes, Seaver was ok, and he got his 300th. But management was saying to the rotation that they weren't good enough, and they wound up performing according to that message, i.e.,they wound up underperforming. I don't know if the same thing is happening here, but I think it is. I predict the best days of Kimbrel and Hendriks are behind them, because of this trade.

Do the statistics prove this? No, of course not, not yet. But it is not an unreasonable conjecture. I think the same thing happened with Yermin Mercedes, by the way, who started going downhill when LaRussa publicly excoriated him for hitting a home run off of a position player. Maybe LaRussa is usually a master of player psychology. He certainly is not in these cases.


Should be required reading for every White Sox fan! Thanks again, for another smart, reasoned analysis of this team. A breath of fresh air to much of what I read online.

You may also like