The Yankees And Pace


Robbie doesn't like a pitch, time to go for a walk. (Chris Pasatieri/Getty Images)

If you’ve been following the Yankees for long enough, you’re well aware that their games usually take an eternity to play. They have patient batters who work deep counts and an offense that scores a lot of runs, and playing in the AL East means they face a bunch of teams built the same way. More pitching changes are made these days made as bullpens become more specialized as well, and that lengthens the game quite a bit. The Red Sox had the longest average time of game at 3:13 in 2011, and the Yankees were second at 3:08. That’s a lot of baseball.

Although we have a general idea of why the Yankees played such long games in 2011, I wanted to get an idea of who specifically was responsible. To do so, I used the “pace” data available at FanGraphs. It’s compiled from PitchFX and tracks the time between pitches within a single plate appearance, so not the time between the last pitch of one plate appearance and the first of the next. The average pace across MLB was 21.6 seconds last year, with both leagues right there (AL was 21.6, NL was 21.7). Let’s break it down into the three major components of the team…

Starting Lineup

For the most part, we can deduce who the culprits are. Nick Swisher always seems to take his sweet time between pitches, as does Derek Jeter. Mark Teixeira doesn’t spend much time fidgeting around between pitches, and neither does Curtis Granderson. Let’s look at the pace data…

(min. 350 PA for ’11 and 1,000 PA for ’09-’11)

Cano is second only to Carlos Pena in both samples, and not by a small margin either. Pena leads him by 1.6 seconds in 2011 and 1.9 seconds from 2009-2011. That’s pretty nuts, the guy takes nearly half-a-minute between pitches. Cano ranking so high surprised me, but it makes sense when you stop to think about it. He’s always walking around the catcher and umpire and fixing his gloves between pitches. It makes sense. The rest is as expected, with Jeter, Gardner, and Swisher ranking well up there and everyone else further down the pack.

For what it’s worth, there’s basically no correlation between pace and wOBA. Using the ’09-’11 data, I get an R-squared of 0.0898 when I plot the two stats against each other. An R-squared of one means a perfect correlation while an R-squared of zero means no correlation. It’s worth noting that several former Yankees — specifically Jerry Hairston Jr. and Hideki Matsui — also rank pretty highly in terms of 2009-2011 pace. There are 234 players in our 2011 pool and 225 players in our 2009-2011 pool, just so you know.


I don’t think there is anything more annoying in baseball than a pitcher who takes an eternity between pitches. Josh Beckett drew some ire last year for his painfully slow pace, and Steve Trachsel was not-so-affectionately known as The Human Rain Delay for the same reason. Rafael Betancourt also has a reputation for taking his sweet time between pitches. Let’s see where the Yankees starters ranked in 2011…

(min. 100 IP for ’11 and 300 IP for ’09-’11)

Unsurprisingly, the slowest working pitcher in baseball last year was Beckett at 26.9 seconds. In fact, the four slowest working pitchers last year all have Red Sox ties: Beckett, Brad Penny, Erik Bedard, Jon Lester. Penny is a few years removed from his stint in Boston, however.

The Yankees’ five starters last year are pretty high up there on the leaderboard, which consists of 137 players. Colon and Nova didn’t qualify for the 2009-2011 player pool, which is 126 guys deep. Two of the three Yankees that did are really high up there, and it’s worth mentioning that former Yankees hurler Javy Vazquez is at 22.8 seconds while Andy Pettitte is at 21.3 seconds. Click the link below the table, and you’ll see that the top of the pace leaderboard is dominated by AL East pitchers, which I doubt is a coincidence. The whole lotta grinding out at-bats happens in the division.

The old adage says that a pitcher who works quickly helps keep his defense on their toes, but it doesn’t show up in ERA (R-squared of 0.0006 using ’09-’11 data). That doesn’t mean it’s not true though, the defense may very well be more “into the game” with a quick pitcher, but it hasn’t yielded any significant benefits in recent years. Just FYI, the R-squared between pace and FIP is 0.00007. For all intents and purposes, that’s zero correlation.


Before I dug up the bullpen data, my hunch is that relievers would have a slower pace than starters because it’s the late innings and there are often men on base; these guys then to take their sweet time and think things through before each pitch. Turns out I was right in terms of the league average (by about two seconds), but not as far as the Yankees go…

(min. 30 IP for ’11 and 90 IP for ’09-’11)

Our player pools are 202 and 183 players deep for 2011 and 2009-2011, respectively. For whatever reason, the Yankees bullpen works substantially quicker than its peers and at about the same pace as the team’s starters. Why? Your guess is as good as mine. It could be Mariano Rivera‘s influence on those around him, could be a team philosophy, who knows. The guys they’ve brought in from other teams were also quick workers, so maybe it’s just dumb luck.  That’s pretty interesting though, and it’ll take some work to determine why this is the case and if it’s actually helping any.

The slowest paced reliever over the last three years wasn’t Betancourt (he was second), it was Jonathan Papelbon at a whopping 31.7 seconds. That seems rather excessive. Papelbon and Betancourt (30.8) were the only relievers over 30 seconds while Jonathan Broxton was at 30.0 seconds on the nose.

* * *

Although the Yankees and Red Sox averaged the longest games — and the Mariners averaged the shortest at 2:46 — there was basically no correlation between average time of game and wins in 2011 (R-squared is just 0.066). Some teams just play longer games, and the Yankees are one of those teams thanks to guys like Cano, Swisher, Gardner, Sabathia, and Burnett. Are long games irritating? Sure, but they come with the territory at the moment, so just enjoy the extra baseball.

Categories : Analysis


  1. Tampa Yankee says:

    This post is pathetic and embarrassing. It took too long to read.

    /Joe West’d

    • Ethan says:

      I lol’ed

    • Genghis says:

      It appears to me the Yankees take about a second longer between pitches both at bat and when pitching. Given about 150 x 2 pitches per game, minus the times between batters and between innings, conservatively this adds, at most, 5 minutes over the average game length.

  2. Bill Style says:

    Never noticed the pace data at FanGraphs. I wonder what Knoblauch’s pace was, having to readjust his batting gloves and helmet in between every pitch

  3. Jose M. Vazquez.. says:

    I always thought that Papelbon and Beckett were violating the rule that says you have to deliver the ball in 30 seconds or less. Is it 30 or 20? In any case the umpires are not calling a ball when the rule is broken.

  4. moonimus says:

    Is there any way to check whether or not there is some carryover effect in the playoffs? I know the playoffs are a crapshoot but it would be interesting to see if the length of the games in an already brutal season had any impact going into the postseason. I’m thinking no but would be interested to see those results.

  5. Kevin says:

    Is that data curved for the average pace of the players they are facing? If not, is it possible that the data could be skewed for AL East batters and/or pitchers because of having to face one another? Holding either factor constant independently (Ie the amount of time a player spend futzing with his batting gloves) and increasing the other (The amount of time a pitcher spends picking a pitch to throw)would increase a players score, and visa versa. To get really good data on a player’s pace individually, wouldn’t you need to curve based on the differing pace of the pitchers/batters they were facing?

    Also, I think it would be interesting to see if the idea of “slowing the pace” is valid. For example, you often get the impression, when watching the game, that after a particularly long at bat the game can slow down to the advantage of the team without momentum, while a fast pace benefits the team in possession of momentum.

  6. RkyMtnYank says:

    Probably right about the same as Cano since Robbie does it every single time too.

  7. Monteroisdinero says:

    I like long baseball games because I love baseball. I wouldn’t mind starting all night games no later than 7 PM. Networks be damned.

  8. kevin says:


    Just a technical comment. You say, “The old adage says that a pitcher who works quickly helps keep his defense on their toes, but it doesn’t show up in ERA (R-squared of 0.0006 using ’09-’11 data).”

    This is a common misuse of the R-squared statistic. R-squared does tell how well the regression equation as a whole(in this case, the only independent variable is pace) explains the dependent variable (in this case ERA). R-squared measures how well we’ve specified the relationship, but not how well a particular independent variable correlates with the dependent variable. What we really want to know here is the statistical significance of the t-statistic for the independent variable. If it is significant, then we deem there to be a statistical correlation, EVEN IF THE R-SQUARED IS MICROSCOPICALLY SMALL.

    So, what does it mean if the have a significant t-stat and low R-squared? It means we’ve left other important independent variables out of the regression. It is obvious that if we only regress ERA on pace we’d get a low R-Squared, because we’re leaving out many (almost all) extremely important independent variables. Ideally, you want a regression that already includes all the usual suspects for explaining ERA and then adds “pace.” Then the t-stat on pace would tell whether it is marginally important after accounting for the other important variables.

    • Jose M. Vazquez.. says:

      You must be a mathematician or statistician. I would have liked it if you could simplify your comment a little more if it is possible. I really could not understand all of your commemnt.

      • Ethan says:

        Basically, from my understanding (which could be off). R-squared tells you how much of the variation in the dependent variable (ERA) is accounted for by the independent variable (pace).

        Mike seems to suggest that because the R-squared value is very low, there is no correlation when in fact, it is the value of the t-statistic from the regression that tells you the significance. So even a very small R-squared variable does suggest some correlation (if the t-statistic is significant).

        However, I think Mike was saying that there is no correlation because the R-squared is so small. While this does not necessarily mean there is no correlation, it does imply that the independent variable (pace) does not have a large impact on the dependent variable (ERA).

        If I am incorrect in my understanding, someone please correct me.

        • kevin says:

          The issue isn’t just the impact of the independent variable on the dependent variable — it’s also how well the regression as a whole explains the variation. That’s the function of the R-Squared. If Mike merely regressed ERA on pace and no other independent variables, he’s certainly going to get a low R-squared because so many other more important things explain ERA.

          But as I said, the issue isn’t the impact of the independent variable on the dependent variable; what’s important is the marginal or incremental impact of the independent variable on the dependent variable when the effect of all other independent variables are taken into account. The marginal effect is measured by the beta coefficient in the regression and it’s statistical significance is ascertained with a t-statistic (at a certain p-level).

          So the t-stat is what’s important. If it is significant, then the variable matters, even if the R-squared is low. The t-stat measures the correlation of any particular independent variable on a dependent variable. Now, all of that said, if all he ran was a univariate regression of ERA on pace, then even a significant t-statistic might not be reliable, because so many important variables were left out and if any of the variables omitted are correlated with pace, then pace could pick up the correlation of the omitted variable.

          All of this may be hard to grasp. The bottom line is that running a single-variable regression of ERA on pace and nothing else is bound to give suspect results.

    • Jose M. Vazquez.. says:

      You must be a mathematician or statistician. I would have liked it if you could simplify your comment a little more if it is possible. I really could not understand all of your comment.

    • Ethan says:

      Yeah, it would be interesting to know what the p-value of the regression was as to whether or not it was statistically significant.

  9. Monteroisdinero says:

    How about stats on Garciaparra and Mike Hargrove? Those guys were successful like Robbie but had ridiculous manners at the plate.

  10. Chris says:

    That pace of play thing seems so dumb. How can that be an accurate count of if someone is slowing a game down? So Robbie fouls a ball off his foot and takes 3 min to get back in the box. Over the whole season that could add .1 or .2 seconds to his average time. Do that 3 or 4 times a year and you are looking at half a second to a full second. What about pitchers trying to hold runners on?

    Is “deliering the ball” the same as putting the ball in play? Do throw overs count?

    I think this is all part of the game.

    • Anchen says:

      I think over the course of a season or several seasons the general averages play out. That’s why it is a fairly large sample of data if you want to make any significant correlations. All players foul balls off, although yes, some do at different rates. I think if you take a 3 year average or something in general would give a fairly good representation, which is what statistics is about. And given that Beckett and Papelbon and most of boston is rather slow seems to match with my own feelings about their pace. Although I am surprised to see cc so high relatively since I always felt like he worked pretty fast.

      • Chris says:

        I see your point but the difference between the Yankees and Boston was 5 min. They were being lambasted this season for pace of play. 5 min is nothing.

        My bigger question is not about pace of play at all but how the stats are gathered. Do they stop the clock when there is an injury? Do they count throwing over to a base delivering the ball or do they only count time between pitches? Sometimes after a foul ball the ump dusts off the plate, a batter gets a new bat etc…I wonder if they pay attention to when the ump called time out for those things.

        Its such a specific stat that effects every second of the game that I can’t imagine whoever is in charge of keeping those stats get it right every time.

        The other major flaw in this is who is actually holding who up here? If Cano takes 15 seconds to get back into the box that means the pitcher is starting at 15 seconds every time. Thats not the pitchers fault but it is time billed to him anyway.

        Consider a guy like Tex who takes a strike from Becket. Tex does not move and is ready, yet Becket steps off the mound, licks his fingers, rubbs the ball, looks around then gets back on the mound etc. Now Tex is charged 15 seconds?

        I think its an interesting stat but how its counted and applied is of more importance to me.

        • Anchen says:

          Yeah, it can probably be a bit imprecise, not sure how those get counted, maybe they’ll publish something on their data gathering. There’s probably some judgement calls too, just like scorers, or defensive stats (+/-, UZR, etc) on calling what plays are in a players zone. I think you can make adjustments for the raw stats as it gets more “advanced”, such as weighing in who the players play or even the pitchers they face. I think given a large enough sample size (The yankees play boston 18ish times but there’s still an entire 162 game season so they’ll play other teams quite a bit). Obviously the stat isn’t perfect. I think over a large sample, things like Tex getting charged a few extra seconds when he faces boston that is out of his control balances out or can be factored out. I think over the large sample general trends of who is fast and who is not generally do bear out though.

  11. RetroRob says:

    Mike Hargrove was the original “The Human Rain Delay.” I never heard the term used with Steve Trachsel. I can only imagine what would happen if Trachsel faced Hargrove. All time might stop. Joe West would cease to exist…and that would be a good thing.

    Hargrove would be loved today by the sabermetric crowd for his ability to draw walks, drive up pitch counts, and cause total chaos with a pitcher by his constant delay tactics.


Leave a Reply

You may use <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong> in your comment.

If this is your first time commenting on River Ave. Blues, please review the RAB Commenter Guidelines. Login for commenting features. Register for RAB.