My once a year post on baseball

Dubner worries a lot about whether people comment on our posts, which is definitely evidence that he doesn’t have enough important things to worry about.

Every time I have made a post about baseball, it has unleashed a torrent of comments. So as an Easter gift to Dubner, here is my annual baseball post.

I’ve been working for the last year with a fantastic young man named Ken Kovash. He is unusual in that he got his MBA from the University of Chicago, but while doing so he also took a bunch of economics classes. After getting his MBA, he asked if he could come work with me, and although I told him he was crazy to pass up salaries four times greater than I could offer him, he decided to stick around and work with me anyway. We’ve spent a good part of the last year working on some baseball related projects, as well as some more traditional applications.

While our joint work on baseball is not yet ready for prime time (although baseball genius Nate Silver of Baseball Prospectus makes mention of one piece of it here), Ken has been doing some baseball-related analysis on his own that offers a substantial challenge to the conventional wisdom among sabermetricians. The particular issue Ken has been examining is “protection.” In other words, does it help me as a batter if the person in the on deck circle is really good? The existing research, by excellent folks like J.C. Bradbury who has a nice blog (and has written a book on the economics of baseball which I have not yet read but have heard great things about) argues that protection is a myth. When you look at the outcomes of at bats, you don’t see measurable effects of having a good batter behind you. In Ken’s research, which you can read here, he makes a nice economics-style argument. There is a lot of randomness in how an at bat turns out which makes it hard to detect an impact of protection even if it is really there. There is a lot less noise, however, in some of the inputs to an at bat, like whether the pitcher throws strikes. By focusing on how the pitcher pitches to the batter, rather than how the at bat actually ends, Ken is able to cut through the noise to find strong evidence that it seems to matter who bats behind you.

After a year, Ken is pretty tired of me. So let me make a pitch on his behalf. Here is a guy with great work experience prior to his MBA, an MBA from the U of C, and a set of data skills honed through a year of working side-by-side with some top economists. I think there are very few folks anywhere who have the combination of skills that Ken has. As much as I would like him to stick around, somebody should steal him away from me! Ideally, it would be a major league baseball team that is looking for a guy who lives and breathes baseball, knows the existing sabermetric literature cold, but also brings a business perspective that goes far beyond his ability to manipulate statistics. Failing that, I think just about any firm would want to have Ken on their payroll. If you want more details, you can contact him directly at kkovash[at]gmail[dot]com.


Hi Ken!

- Eric K.


That's a nice plug for Ken, but by putting his email address on the web, you've doomed him to a lifetime of spam. I suppose a $250,000 salary would make up for that, but still...


The real point of protection is to prevent a good hitter from being walked. The data shows that putting good hitters behind a player means he'll see more strikes, reducing the number of unintentional walks he'll get. The pitcher will also intentionally walk the hitter much less. Putting your worst hitter behind your best hitter will mean he'll rarely get to swing with runners in scoring position.


I finished my last year of collegiate baseball in June of 2006 at Brigham Young University. I've been pitching my whole life and I can tell you that there is truth to what Victor is saying. On the other hand there is never a rule that establishes what strategy will take place in a certain situation. It's all circumstantial. So while I agree with Ken in some situations, I would have to agree with J.C. Bradbury in the majority. For example, if there's a guy on second and I absolutely need a double play ball with a good hitter up... I'm going to walk the guy, no matter who is batting behind him! But then again, if the batter has gone 0 for 3 that day with 3 strikeouts and has looked like garbage, I'm going to pitch to him because I know I can beat him. Also maybe I walk the guy, and a really good hitter is on deck... There is always the option of switching things up and substituting in a different pitcher that has had good success against this really good hitter. I'm tellin you, there is just too much a team can do in any situation. You can't control a study like this, all you can do is look at the numbers and I'll bet that while protection can play a small factor, for the most part, it's irrelevant. One last thing, I know who's on deck when I'm pitching, but that doesn't mean I'm thinking about it. My focus is getting the out at hand!



Thanks Steven for pointing out Ken's study and mentioning my own. Actually, Ken's results fit with the results of the study (co-authored with Doug Drinen) I present in the book. Protection has two parts: 1) preventing the batter from walking 2) and allowing the batter to see better pitches to hit within the zone.

Consistent with Ken's study, we find that a good on-deck batter reduces the likelihood of a walk. However, we also find that a good on-deck batter also lowers the likelihood of getting a hit, extra-base hit, and hitting a home run. While good on-deck batters reduce walks, pitchers appear to compensate by throwing pitches that are more difficult to hit within the strike zone. I'm curious to see if Ken has fastball speed in his data to see if pitchers throw harder when the batter is protected. The overall magnitude of the effect is tiny, and thus the gains from protecting a batter are a wash.

Ken and I briefly corresponded last night, and we're going to look into this a bit further when out schedules let up.



Well, as my Easter gift to you and Dubner, I have made a comment, be it a very short comment.


The last Indians World Series win was my birth year, 1948.
Hope springs eternal.


Seeing a highly statistical approach validate (conventional) conventional baseball wisdom makes me think things have come full circle.

Please convince Ken to go work for the Red Sox. Failing that, please convince him not to work for the Yankees. Thanks.


When I played college baseball, my first couple of seasons I batted in the second spot in the lineup in front of a pair of All Americans; I was the protected. I saw lots and lots of fastballs for strikes. My latter two years of playing I was hitting in the middle of the lineup and saw lots of breaking balls and changeups; I was the protector. However, my results numbers didn't really show a major difference despite the difference in how I was pitched, just like the data referenced above. I suspect that this is because the results for any particular batter are much more dependent on individual talent level than on how they are pitched. Over the course of a season, every hitter will see a large number of fastballs and breaking balls. And, like the BYU pitcher mentioned, individual situations play a tremendous role in the way a pitcher approaches an at-bat.


I haven't done much statistical analysis on baseball, but as a fan for 17 years, I would argue that the guys batting before a prime-time hitter matter more than the guys batting behind him.

Put a .350-.400 OBP hitter in front of Albert Pujols and a .250-.300 OBP behind him. I would bet that Albert would see more good pitches with that lineup than vice-versa.


It seems to me that the next hitter's OBP (or other statistic) against that particular pitcher would matter more than his overall OBP. Some pitchers have great success against particular batters, and vice versa.

Also, if the pitcher is a left-handed specialist brought in to pitch to a left-handed batter, the OBP of the next batter should only matter if he is a left-hander, too.

Finally, the exact location of the pitch and the resulting ball-strike call by the umpire are outcomes themselves, which are both subject to noise. I don't know whether these stats are kept anywhere, but I would guess that there are more visits to the mound by the catcher and/or a manager/coach when a pitcher falls behind to batter with a good hitter behind him.


l would think it evens out , yes the pitcher would be bear down , but the hitter also knows the ball will be close or in the strike zone , but l would like to see the speed difference of pitches , also would like to see lead off hitters included and pitchers of opposing teams stats on pitch speed

as for Ken thought Barbie would always support him , or Malibu Barbie ?


But what about Pete Rose and the Hall of Fame? Is there really any evidence that he bet on baseball much less bet on his own team?

And look at poor Barry Bonds. Who among us can really say that he took steroids? I mean just look at him. He doesn't look like the Guvernator.


I wonder what caused this post to be so quiet. Was it Easter? Maybe your blog demographic has shifted?


Didn't really incite a maelstrom, but you did manage to entice yet another gem from RandyfromCanada, which is all I really care about. Always leaves me feeling that much closer to enlightened.

Merry Easter, Mr. Dubner.


you lub me don't you !


I love your blog, and I read regularly, but I don't comment often. If you'd mentioned soccer, I might have something to say to this post, but that's not the reason I' m commenting.

As a budding social scientist, I feel like there is something you can do about your comment numbers. Analyze the comments you do recieve on your posts. You already did a crude sort of survey by asking people about commenting, now look at the data you have and run some statistics on them! Maybe it has to do with the time you post your topics? The topics themselves? Do the same people comment over and over, or do you get new commenters everyday. Are people turned off by the TypeKey sign in, or are they just lazy?

When you post some of these statistics, maybe I can be coaxed into commenting again. ;)


I still say Pete Rose deserves to be voted into the hall of fame. And anyone who says otherwise is a bum. Because Pete never bet on baseball.


This is referring to the novemeber 6, 2005 article

There are plenty of things with this article that I both agree and disagree with. First off, it is true that the opportunity cost of voting can be high for some people and that a single vote will most likely not affect the final outcome of a big election. However, it is also true that the federal government and the state-legislatures tax the citizens of America. This article completely ignores the fact that that Americans pay many expenses to the government. This leads me to disagree with the article in that this idea of not voting is rational. I also believe that the people who believe in the IDEA are irrational and do not believe that each individual is irrational for not voting. I mean, if irrational people consistantly make irrational decisions, most likely those "irrational voters" will vote for legislation and/or leaders that are irrational. Then, consequently, the so called "rational people who don't vote" are all of a sudden economically punished for believing in the idea that their vote "would not affect the election". Now they have unnecessary taxes and other expenses that are irrational. In actuality, it is the idea that a single vote won't affect the overall election that would cause an irrational decision. If that idea wasn't floating around in the heads of economists, maybe we would have had more rational leaders in the past. Certainly if those rational people who believe they won't make a difference in an election, collectively vote, a huge difference would be made. Therefore, I believe that the idea of not voting is irrational, and the individuals that don't vote may be rational, but will end up in an irrational situation.



Hey Ken, its eric representing the 3312. Holla