I’m not a mathematician or a scientist. In fact, the last time I took a course in mathematics was in my last year of high school. It was calculus, and in it I achieved the lowest mark of any of my final year courses. Aside: My teacher’s name was Mr. Wink, and if my mark properly reflected his efforts in teaching me, it would’ve been much higher.

Despite my lack of talent for dealing with numbers, I like to think I understand them. To me, they seem like a source of reason, even when the way they were gathered and processed might be unfamiliar. Yes, when you’re not as clever with digits, you have to have a little bit of trust in the person presenting you with their findings, but no more than the amount of faith you would have in a pilot, a doctor or even a reporter.

Like the rest of those professions, it’s hard for someone dealing with numbers to continue dealing with numbers unless they’re doing it in an accurate fashion. Misrepresentation and frauds are found out quickly. Also like the rest of those professions, every day people rely on the validity of number crunchers to properly do their job.

Major League Baseball’s managers should be no exception.

There are several quotes from Toronto Blue Jays manager John Farrell in this piece from MLB.com that are maddening to anyone who follows the most basic of beliefs in reason or logic, but by far, the most troubling to fans of Toronto’s baseball team is this section that Gregor Chisholm summarized:

Farrell often talks about his tendency to rely on past performance statistics when making decisions with the bullpen. He said approximately 10 at-bats translates into a suitable sample size, but how a pitcher is throwing should also factor into any potential move.

Ten. Ten times. Ten times. I’m sorry, but I’ll have to say it again: Ten times! Let’s not bring up numbers here. Let’s just think about this reasonably. We’re all baseball fans. No matter where we sit on the spectrum measuring belief in the best way to predict an outcome in the sport, we can all admit that there are large elements of luck and variables outside of a pitcher or hitter’s control at play in any single event on a baseball diamond.

So, how could a measly ten times tell you absolutely anything about a potential outcome? Would anyone trust anything that’s only been done ten times before? Would you use a drug that was tested ten times before you took it? Would you believe someone was a psychic if they correctly guessed “heads” on a coin flip more than half of the ten times you tried it? So, why would you possibly trust ten pitcher/batter matchups to tell you anything at all?

It’s true that I haven’t been very impressed with the efforts of the Blue Jays new manager, but to this point I’ve merely complained as a powerless observer. But I don’t want to merely complain in this fashion any longer.

During lunch, I went over to the Chapters book store on the corner of John and Adelaide and purchased a copy of The Book: Playing The Percentages In Baseball. I will be couriering this copy to John Farrell with a letter urging him to read several sections, and at the very least, to think critically about the decisions he makes as a manager.

Even if he laughs it off as the ramblings of some nerdy spread sheet jockey or doesn’t open the package up at all, I can, at the very least, breathe a little bit easier and feel as though I’ve made some effort to improve some of the more questionable in game decisions.

As a manager of a Major League Baseball team, I’d imagine that Mr. Farrell has an enormous amount of calculations at his disposal. Yet, there seems to be a disconnect between having these numbers and using these numbers. It’s my hope that taking a look through this work by Tom Tango, Mitchel Lichtman and Andrew Dolphin might bridge that apparent gap a little bit.

Again, I’d like to point out that I’m not suggesting that I’m smarter than anyone or know more about baseball than the team’s manager. I’ve merely read a book that uses statistics to draw conclusions that seem reasonable to me. And I believe that these reasonable findings could be of benefit to Mr. Farrell and any other person making decisions that affect the outcome of something that I’m particularly fond of: a baseball game.

Comments (22)

  1. I’m just surprised there was a copy of “The Book” there.

  2. How often would a big league manager have many more than 10 ABs to work with in terms of hitter/pitcher matchups? With the unbalanced schedule and interleague, that’s probably mostly limited to guys in the division. I’m not saying 10 is an adequate sample size (and I am in the sciences and interpret/work with statistical data daily), and I’d be interested to see what Farrell actually said instead of what Gregor distilled it down to. There certainly are examples of Farrell not relying heavily enough on past performance (i.e. ignoring Dotel’s splits against lefties, for one), but this particular quote seems foolish to get up in arms about. All he seems to be saying is that he uses the limited, flawed data available to try to make bullpen decisions, and I’m fine with that.

  3. Don’t use singular pitcher/batter matchups. It’s dumb to try. Look at other splits involving pitcher types, hitter types, platoons, which offer so many more samples. It’s ridiculous that someone at the height of their profession wouldn’t be aware of the problems in basing any decision after such a small sample.

  4. I heard Farrell reads the first ten words of a book and then judges if it’s any good. Hope there isn’t a dedication.

  5. I would say 10 at bats could be enough. if a guy is 8 for 10 against the pitcher with 3 Hr, it might be worth thinking about.

  6. @BillCosby: I’d rather see him at least use vs L / vs R splits over individual matchup splits. Farrell’s copies of these individual matchup splits should be doused in kerosene and set ablaze. They are that useless. This is far from the first time this has come up this year either, so it is a repeating pattern that needs to be nipped in the bud.

    Good call Dustin. Way to show some moxie. ;)

    • As far as matchup goes, what about pitcher tendencies? Ground ball pitchers or fly ball pitchers. Power pitchers versus finesse. That sort of stuff more than just LEFT or RIGHT.

  7. I just don’t take this as saying he doesn’t use that data, or as a suggestion that he doesn’t understand that it’s important. The “past performance statistics” bit leads me to believe that he didn’t say that matchups are the be-all-end-all, but that he uses them when he feels there is an adequate sample size to draw from (which he is, of course, incorrect about).

  8. My interpretation goes like this: Any sample size tells you more than zero does. Numbers are there for a reason, so use the ones you have. Besides it’s his first go at managing, just like the players are going to take their lumps, Farrell will take his. I understand that two games don’t make a strong case for a player but I think uncertainty of what they can control is what some managers try to avoid.

  9. @Drew: Good point.

  10. Sweat makes a point that math actually backs up, based on degrees of freedom. 10 at bats lowers the degrees of freedom in the sample which reduces its ability to be representative of the population (by increasing std deviations). But that doesn’t mean that its “not an indicative sample size” in any and all cases. If someone’s 8 for 10 against a pitcher with 3 HR and he’s a consistent .230 hitter, then 10 at bats could be plenty to say that the number is a statistically significant outlier. I’m not saying John Farrell actually incorporates this into his comment, but he’s not necessarily a complete idiot for saying what he said…

  11. Any sample size is far more dangerous than no sample size. It creates a completely unreal expectation.

  12. Casey Stengel did a better job with matchups than most modern managers. Honestly, if Rajai Davis is up and the opposing pitcher features any kind of breaking pitch, I’m pinch hitting. If Jeter is coming up, I’m going to the ‘pen for the hard thrower or sinkerballer. We have charts pitch by pitch as to the hitters abilities. Handedness is the matchup for people who can’t read.

    This is not just Farrell. This is most MLB managers. The revolutionary thinkers do not get these jobs, because a GM would rather win with Dusty Baker, than chance losing and looking like an idiot with Joe Maddon, at least 90%of the time.

  13. You could also suggest that Mr. Farrell go and talk about sample size and other such mathy things with the best player in baseball, who happens to be on his own team. After reading that Jeff Passan article about Joey Bats (here for those who missed it [well worth the read]: http://www.thepostgame.com/features/201106/number-crusher-how-blue-jays-slugger-jose-bautista-experimented-his-way-greatness ) I realized that Jose Bautista’s biggest aid in reaching the point he’s at now, was in fact not a drug at all (though of course we can never be 100% sure whether or not he’s used them: only he knows that). It was his brain, which after all is like a muscle, in the “use it or lose it” sense. His brain happens to really gravitate towards math, so Farrell while benefiting from his on field abilities, could also tap into and put to use his slugger’s brain. Have Bautista read the book and give Farrell feedback. It can’t hurt.

    You don’t hear stuff like this from every athlete: “”There are no flaws in math,” Bautista says. “You can have 50 people read one paragraph, and they’re going to interpret it in 50 ways. You can’t find anybody who would say two plus two doesn’t equal four.”" I would argue that you can find people who will say two plus two doesn’t equal four, but perhaps only in the figurative sense. The article is pure gold from one of the best young baseball writers around.

  14. Small sample size doesn’t make it irrelevant. You are looking at it the wrong way, you are not trying to make predictions, you are trying to evaluate RISK. If a player is 3/10 all singles you can assess it is unlikely he’ll hit a home run, but if a player is 1/2 with 1 hr you can assess that there is high risk he might hit another home run.

  15. Parkes – Let it go, Farrell’s here to stay. BTW, I hear he openly despises the work you do on DJF’s and Getting Blanked.

  16. 10 PA is irrelevant in baseball. The guy could have been in the middle of a hot stretch/cold stretch every single one of those 10 PA. It has no predictive value whatsoever when evaluating options to make a move.

  17. The stupid mistake that farrell made was that he just looked at rep’s numbers vs wieters but didn’t even consider wieter’s tendencies overall. He’s opsing .947 vs lefties and .659 vs righties so obviously you’d want a righty to face him. Stats do not replace common sense and we’ve clearly seen by now that farrell has none.

  18. @mudpie: Exactly. You have to pick the right split/tendency to look at, and at the moment this is a weakness for Mr. Farrell, hence the cheeky play by Parkes.

  19. @TomJackson – that was a great rticle on JoeyBats.

    I’m not sure if there are studies on this but I wonder what the affect is on a player, early in his development, telling him to put heavy focus on, for example, increasing his walk rate and lowering his strikeout rate etc.

    The Blue Jays told Brett Lawrie to improve his strikeouts and walks and he did in on command. I’m not saying this is possible for all players…but maybe a guy like Brandon Wood could have benefitted massively if he were told to focus heavily (quite possibly to the detriment of his huge raw power) on lowering his K rate because that would greatly improve his value and increase his chance of being a regular in baseball.

  20. “Don’t use singular pitcher/batter matchups. It’s dumb to try. Look at other splits involving pitcher types, hitter types, platoons, which offer so many more samples. It’s ridiculous that someone at the height of their profession wouldn’t be aware of the problems in basing any decision after such a small sample.”
    The whole issue is not worth getting worked up about. I would be shocked if Farrell doesn’t first consult terms that provide greater sample-size – ie. batter-handedness. Batter-pitcher provides a data point, and one that is not likely to be limited to pure batting average, although it is easy to distill it down to such. Imagine, if you will, two right handed relievers with similar RHPvs RHB splits for his last few seasons. Late in a game, a certain veteran right-handed batter steps to the plate. Righty pitcher A has faced him six times, surrendering a single and a double. Righty pitcher B has faced the veteran 13 times, also surrendering a single and a double. Both pitchers are rested.
    Do you ignore the history? Do you not think the history is also a scouting report? What factors caused the hitter to have greater success over a smaller period than against the other pitcher. It’s a data point, one to consider, but like all data points, never one to adhere to religiously.

  21. Parkes, thanks for addressing this issue. I asked about it in last weeks livechat during Getting Streamed On. Just wanted to say thanks. Hopefully Farrell reads the book you sent him, they probably have some downtime on the planes between road games. I have never read it, but as someone who has a background in stats, i am pretty skeptical of small sample sizes. I’ll see if my library has that book and I’ll give it a read.

    Maybe Stoeten should actually start printing and mailing his “Dear John” letters to Farrell.

Leave a Reply

Your email address will not be published. Required fields are marked *