Skip to content

A.J. Ellis: Red Sox manager John Farrell trusting gut over numbers in the World Series

Oct 25, 2013, 8:10 PM EDT

Getty Images Getty Images

Dodgers catcher A.J. Ellis is providing analysis of the World Series to the L.A. Times. In a column posted yesterday prior to Game 2, he wrote about how Red Sox manager John Farrell is trusting his gut over the numbers in a short series lasting between five and seven games. The Red Sox are one of the teams most overtly reliant on analytics, which made Farrell’s decision to start Jonny Gomes in left field against two tough right-handed starters, as opposed to the left-handed Daniel Nava.

Ellis writes that it is Gomes’ playoff experience and energy, rather than proficiency against right-handers, that Farrell wanted in the first two games of the World Series against the Cardinals:

These immeasurable factors of chemistry are loathed by sabermetricians, who tend to scoff at these claims. Undeniable even by the most stubborn supporter of these new metrics is that in the postseason, the Red Sox are 1-3 when Nava starts games and 7-0 when Gomes does, heading into Game 2. To players, managers and fans, that is the only stat that matters in October.

They are 7-1 now, of course, as the Sox lost Game 2 by a 4-2 margin. Gomes is hitless in seven trips to the plate in World Series play. Farrell announced earlier that Nava would be starting Game 3 against Cardinals starter Joe Kelly.

As a self-described “sabermetrician”, I have two thoughts on this: Ellis is relying on a very small sample size, of course. Four starts for Nava and eight starts for Gomes is representative of very little. Secondly, while it is curious for Farrell to abandon the very tactics that helped bring his team to where it is today, it is unlikely to make any noticeable difference in a short series. Nava’s proficiency against right-handers, or Gomes’ deficiency against them, won’t be the sole reason why the Red Sox succeed or fail once you account for myriad other factors, plus the ever-present effect of randomness, even more stark within a smaller sample. In the big picture, there’s nothing wrong with Farrell trusting his gut, but it does set him up to be the fall guy if things don’t work out.

  1. kaleidoscopictreats - Oct 25, 2013 at 8:30 PM

    “it is unlikely to make any noticeable difference in a short series”

    What, exactly, is unlikely to make any noticeable difference? Do you mean Farrell’s decision to abandon his analytic tactics and rely on his gut, instead? If this is what you mean, why would you think this? Abandoning the very tactics that got his team _into_ the world series would very likely make a difference _in_ the world series.

    • Bill Baer - Oct 25, 2013 at 8:37 PM

      Accurately assigning credit and blame is more difficult in smaller sets of games because of all the competing factors.

      Think about it this way. Let’s say you have a deck of cards, but you take out one black card and put in one extra red card, and you draw randomly once. You draw a red card. Is it due to the fact that you put in that extra red card, or just randomness? If you draw randomly 1,000 times and get 520 red cards, you can be more confident that the extra card impacted the results.

      • kaleidoscopictreats - Oct 25, 2013 at 10:20 PM

        I don’t think this is the right way to think about this, at all.

        Take H: The Red Sox will win Game 2. How confident should you have been in H initially? Maybe 0.5, 0.6, since the teams are fairly evenly matched.

        Now what do you think your confidence in H should have been had you learned, E: Farrell didn’t use a highly successful regular season tactic – prior to the game? Higher than 0.5/06, of course.

        Maybe I’m betraying my own ignorance right now, but I just don’t see how the fact that this is a seven game series has any bearing on how confident we should have been in H, given the regular season history. And if we should have been more confident in H given E, then it’s reasonable to think that Farrell’s decision affected the game now. Do we know? Nope, but we’re trading in confidences not knowledge.

      • kardshark1 - Oct 25, 2013 at 10:24 PM

        Every decision impacts the results. The goal as a manager should be to get the maximum out of every little decision that he can control. Even something as little as having a player shift 3 inches to his left impacts the series. If player A will increase your odds of winning by .0016% over player B, then there is ZERO reason to go with player B, even though it’s such a small number. If you disagree and think player B increases your chances, then fine, go with B. But it sounds like Ellis agrees player A (Nava) gives them a better chance, but since it’s such a little difference, it doesn’t really matter. It absolutely matters. If I had a .0016% to win the lotto or a 0.00% chance, I would choose .0016% every time.

      • Bill Baer - Oct 25, 2013 at 10:36 PM

        @ kardshark

        That’s not what I’m arguing. I agree with maximizing strategies to win. I’m pointing out that the little tweak being the direct cause of victory is hard to discern over such a small period of time. Making a big deal out of Farrell’s use of Gomes over Nava assumes a direct causal relationship between Gomes and their 1-1 start. That’s all I’m saying.

      • paperlions - Oct 26, 2013 at 12:06 PM

        You guys should see how small of an effect sabermetricians will argue about.

        Recently, MGL pointed out that the difference of Carlos Martinez facing Ortiz compared to Siegrest or Choate was about 1 run over 9 innings, or about 0.025 runs per AB. Of course, that was based on generality of splits for the players involved as specific matchup effects never have large enough sample sizes to evaluate. But he was outraged at Matheny’s horrible decision, which, on average, results in a 0.025 run difference.

        The reason such small difference are argued about is exact what Bill is saying. Over a short series or a single game or AB, the effect is small, negligible, or impossible to detect, but over an entire season the effect of always making the right decision is huge…and it is those data that should be considered when making a decision.

  2. junglerat524 - Oct 25, 2013 at 8:35 PM

    JONNY GOMES !!!!!!!!!!!

  3. sabatimus - Oct 25, 2013 at 9:14 PM

    I trust my gut when it comes to Stephen Drew, too—I wind up vomiting.

  4. lawrinson20 - Oct 25, 2013 at 9:51 PM

    I kinda disagree.

    One hit can make the difference in any of these games, and Nava at the plate against a right-hander gives you better odds. It’s not as if Gomes is ‘hot.’ This ‘intangible presence’ stuff can only go so far, and i trust Nava’s .300+ average over The Magic that only Farrell sees.

  5. louhudson23 - Oct 26, 2013 at 4:57 AM

    If baseball were a mathematical exercise,then all of this would be relevant. But it is a game played by humans. Unless a single result occurs 100% of the time,the results will always include other statistically less likely results.Which at bat, in which game, in what situation? We are all absolutely clueless until after the fact,which is best the reason to watch.It is not to wait for further data to plug into our spreadsheets….All of this is great if we are making hundreds of bets per year or writing insurance policies…but we aren’t …we are watching a game where emotion,luck,skill,execution in a given moment,situation and game dictate,not after the fact calculations….the rarity of a result does not nullify in any way the fact that it occurred. It doesn’t even mean it won’t happen again on the very next play….watch,enjoy, and root, root, root for the home team……

    • cktai - Oct 26, 2013 at 8:04 AM

      Just because you aren’t 100% certain of the result doesn’t mean you should strive to maximize your chances. Take traffic lights for example. You can run a red light and there is a decent chance that you won’t get hit. At the same time you can wait for a green light and still be involved in an accident. Just because you are not certain of the result doesn’t mean you shouldn’t maximize your chances of being accident-free by waiting for a green light.

      As you said, the game is about emotion, luck, skill, and execution. There are so many factors involved. It seems stupid to not try and control those few factors that we can control.

    • paperlions - Oct 26, 2013 at 12:15 PM

      At least learn something about the things you say don’t matter so you don’t sound so completely ignorant when you talk about them…or better yet, avoid talking about them.

      There are not “after the fact calculations”, these are data available ahead of time. If you really think understanding the factors that go into winning baseball don’t matter, you should tour all of the MLB franchises that have large analytical departments that work year around to develop their own proprietary evaluation techniques and point out to them that they are wasting time and money.

      Your first two stops should probably be Boston and Saint Louis, because they are among the most aggressive in applying analytical approaches to understanding baseball, but I’m sure they are both in the WS despite that, and not because of it. Indeed, StL has been applying analytics to the draft and player development for quite some time to determine which kinds of players are undervalued (i.e. which players kinds of skills/talents/traits players have the provide more value to MLB franchises than their draft position generally suggests they should).

      No need to stop by Philly, as they don’t even own a calculator.

      • Francisco (FC) - Oct 26, 2013 at 1:26 PM

        No need to stop by Philly, as they don’t even own a calculator.

        Stop exagerating. Of course they do. Otherwise Ruben can’t count RBIs beyond 21 without using his toes and unzipping his fly.

      • paperlions - Oct 26, 2013 at 1:39 PM

        I have it on good authority that they have both an abacus and a young whipper-snapper with a slide rule.

      • louhudson23 - Oct 26, 2013 at 2:51 PM

        Calculations require data require data,There was zero data till they played a game. And no new data till they played the next one.After the fact….

      • paperlions - Oct 26, 2013 at 6:42 PM

        You see, Lou. Games have been played before. 1000s of them. In addition, the exact same players have playing in games before, 100s to 1000s of them. And that data can inform decision making so that you can make better decisions. You can’t possibly be this stupid. What you are suggesting is akin to saying that you don’t know the best route to drive to work this morning because you are not at work yet and that none of your prior experience if useful to help you make navigation decisions.

        The data from a single game is actually pretty useless, especially because of the extreme amount of chance that contributes to individual outcomes in particular ABs, which is why sample size matters.

  6. theskinsman - Oct 26, 2013 at 5:30 AM

    Whatever the reason, I’m glad to see Nava back in the line up. Dance with who brought you,and all that…

  7. Francisco (FC) - Oct 26, 2013 at 10:59 AM

    These immeasurable factors of chemistry are loathed by sabermetricians, who tend to scoff at these claims. Undeniable even by the most stubborn supporter of these new metrics is that in the postseason, the Red Sox are 1-3 when Nava starts games and 7-0 when Gomes does, heading into Game 2.

    Correlation does not equal causation. Boy we humans love to see patterns, too bad we can’t tell when it means absolutely nothing.

  8. gloccamorra - Oct 27, 2013 at 3:31 PM

    I can see the headline now: “Red Sox Lose Series When Manager Gets Stomach Flu.”

Leave Comment

You must be logged in to leave a comment. Not a member? Register now!

Top 10 MLB Player Searches
  1. D. Wright (3049)
  2. J. Fernandez (2505)
  3. Y. Cespedes (2414)
  4. G. Stanton (2380)
  5. D. Span (2290)
  1. F. Rodney (2110)
  2. Y. Puig (2097)
  3. M. Teixeira (2006)
  4. G. Springer (1970)
  5. H. Olivera (1938)