You Can Count on It

“The most important thing we want our guys to understand is we need to get into better hitting counts. When you get into better hitting counts, you become better hitters. If you look at the numbers in the NL the past few years, guys in 2-1 counts were hitting .340, when the count was 3-1 they were hitting .330. What we’ve tried to accomplish is to be more patient so we can get ourselves into better hitting counts.”
—ex-Pirates manager Lloyd McClendon, Tribune-Review April 8, 2005

Unfortunately for McClendon, this philosophy didn’t translate into more runs for the Pirates, who were 28th in the majors (behind even the Royals) when he was fired this week. But that doesn’t mean that McClendon wasn’t right.

Earlier this week I talked about how pitch data is often used to evaluate players by looking at pitches per plate appearances and things like the percentage of swings and misses, first pitch swings, fouls and taking balls. And while all of those measures are interesting, what I think gets to the heart of the matter is how well a player controls the strike zone and is therefore able to select a pitch to hit rather than having to react to a pitcher’s pitch. So this week I’ll take a look at how we might measure that skill using pitch data.

Getting in the Zone

The Retrosheet data that I used for the previous article does not include pitch location, so a direct analysis that looks at how often a hitter swings at pitches out of the strike zone is not possible. In the strictest sense then, measuring the ability to control the strike zone is out of our reach. However, the counts at which a hitter’s plate appearances end are recorded. As a result, we can group the counts into hitter’s counts, pitcher’s counts and neutral counts based on McClendon’s reasoning that hitters who end up in favorable counts do so because they’re more selective and avoid swinging at pitches out of the strike zone early in their at-bats.

For this article I’ve grouped the counts as follows:

Count  Hitter Pitcher Neutral
 0-0                     X
 1-0      X
 2-0      X
 3-0      X
 0-1             X
 0-2             X
 1-1                     X
 1-2             X
 2-2             X
 2-1      X
 3-1      X
 3-2                     X

The categories in which the counts fall are pretty self explanatory, although some might argue that a 3-2 count is a hitter’s rather than a neutral count. Although the on-base percentage (OBP) recorded at 3-2 is high, the batting average is typically not, so it seems to lend itself to a neutral count. And although hitters do hit fairly well in 0-0 counts, I placed it in the neutral group since neither the pitcher nor the hitter has an inherent advantage on the first pitch of the at-bat, with the pitcher not pressured to throw a strike and the hitter not pressured to swing.

To then see which players were more proficient in getting themselves into advantagous situations, I took a look at the 473 hitters who amassed 500 or more plate appearances (not including intentional walks and hit-by-pitches) in the years 2000 through 2004 and counted the number of plate appearances for each hitter that fell into the three categories. I then calculated the percentage of plate appearances in each category, along with their on-base plus slugging (OPS) normalized for league, year and park (NOPS).

The top 20 hitters in getting into hitter’s counts were:

                      NOPS      PA  HCount  PCount  NCount    HPct    PPct    NPct
Barry Bonds            179    2704     973     683    1048   0.360   0.253   0.388
Mark Grace             104    1640     547     498     595   0.334   0.304   0.363
Brian Giles            129    3227    1043    1009    1175   0.323   0.313   0.364
Chipper Jones          125    3163    1018     959    1186   0.322   0.303   0.375
Luis Gonzalez          123    3111     995    1142     974   0.320   0.367   0.313
Carlos Baerga           98     510     161     198     151   0.316   0.388   0.296
Bernie Williams        114    3059     954    1011    1094   0.312   0.331   0.358
Lenny Harris            85     860     264     314     282   0.307   0.365   0.328
Rafael Palmeiro        114    3268     985    1140    1143   0.301   0.349   0.350
Albert Belle           106     607     181     182     244   0.298   0.300   0.402
John Olerud            111    3089     919    1132    1038   0.298   0.366   0.336
Mark McLemore           95    2108     626     721     761   0.297   0.342   0.361
Scott Hatteberg         99    2382     707     940     735   0.297   0.395   0.309
Barry Larkin            99    1830     543     719     568   0.297   0.393   0.310
Kenny Lofton           100    2730     803     982     945   0.294   0.360   0.346

On average these hitters are putting the ball into play around 30% of the time when ahead in the count. The interesting thing about this list is that it doesn’t just include good hitters like Chipper Jones, Luis Gonzalez and Bernie Williams, but it also includes players with high walk rates such as Lenny Harris, Kenny Lofton and Mark McClemore. As you might expect, this indicates that getting into hitter’s counts, while good for the batting average and slugging percentage, is also a big boon to eventually reaching base via walks.

The “top” 20 hitters in getting into pitcher’s counts were:

                      NOPS      PA  HCount  PCount  NCount    HPct    PPct    NPct
Alex Gonzalez           91    2259     367    1231     661   0.162   0.545   0.293
Joe McEwing             87    1144     239     620     285   0.209   0.542   0.249
Rod Barajas             80     906     163     479     264   0.180   0.529   0.291
Shawon Dunston          96     571     101     301     169   0.177   0.527   0.296
Gerald Williams         85    1179     180     621     378   0.153   0.527   0.321
Angel Berroa            93    1293     237     678     378   0.183   0.524   0.292
Gerald Williams         85    1171     180     614     377   0.154   0.524   0.322
Homer Bush              75     750     141     390     219   0.188   0.520   0.292
Alfonso Soriano        107    2749     545    1425     779   0.198   0.518   0.283
Vance Wilson            93     691     146     358     187   0.211   0.518   0.271
Dee Brown               79     832     149     426     257   0.179   0.512   0.309
Doug Glanville          85    2240     419    1146     675   0.187   0.512   0.301
Mark Ellis              94    1010     191     513     306   0.189   0.508   0.303
Tomas Perez             93    1025     197     517     311   0.192   0.504   0.303
Jared Sandberg          92     698     109     352     237   0.156   0.504   0.340
Tony Womack             87    2773     570    1397     806   0.206   0.504   0.291
Bill Hall               90     606     120     305     181   0.198   0.503   0.299
Pat Meares              83     779     173     392     214   0.222   0.503   0.275
Adrian Beltre          111    2952     592    1483     877   0.201   0.502   0.297
Mike Difelice           81     828     143     415     270   0.173   0.501   0.326

This is a group you don’t want to be in unless you have the uncanny ability to hit waste pitches out of the strike zone like Alfonso Soriano and, to a lesser extent, Adrian Beltre. The Alex Gonzalez in the list is the one from Florida, not Tampa Bay. You’ll also notice that many of these players did not garner a large number of at-bats due to their low NOPS. And those low NOPS numbers come home to roost when comparing the two lists; the average NOPS of the first was 112 while that of the latter was just 89.

I also found that when looking at seasonal data the correlation between a high percentage of plate appearances in pitcher’s counts and NOPS got stronger (more negative) as the plate appearance threshold I used went up. In other words, as hitters garner more plate appearances in a season, it is more important for them to avoid getting into pitcher’s counts. This is just what we would expect if there were a true correlation here and not simply randomness. In a smaller sample size a hitter is more apt to get lucky in a few at-bats, which raises his NOPS and masks the relationship between getting into hitter’s counts and being successful at the plate.

Controlling the Zone

But controlling the strike zone is really a combination of getting into favorable hitter’s counts and avoiding favorable pitcher’s counts. To attempt to measure this I then divided the HPct by the PPct so that this ability could be represented by a single number. The leaders in the resulting CZoneRate were:

                      NOPS      PA  HCount  PCount  NCount    HPct    PPct    NPct CZoneRate
Barry Bonds            179    2704     973     683    1048   0.360   0.253   0.388   1.425
Mark Grace             104    1640     547     498     595   0.334   0.304   0.363   1.098
Chipper Jones          125    3163    1018     959    1186   0.322   0.303   0.375   1.062
Brian Giles            129    3227    1043    1009    1175   0.323   0.313   0.364   1.034
Albert Belle           106     607     181     182     244   0.298   0.300   0.402   0.995
Bernie Williams        114    3059     954    1011    1094   0.312   0.331   0.358   0.944
Luis Gonzalez          123    3111     995    1142     974   0.320   0.367   0.313   0.871
Mark McLemore           95    2108     626     721     761   0.297   0.342   0.361   0.868
Rafael Palmeiro        114    3268     985    1140    1143   0.301   0.349   0.350   0.864
Matt Lawton            103    2853     835     982    1036   0.293   0.344   0.363   0.850
Moises Alou            116    2902     801     950    1151   0.276   0.327   0.397   0.843
Lenny Harris            85     860     264     314     282   0.307   0.365   0.328   0.841
Frank Thomas           121    2328     682     833     813   0.293   0.358   0.349   0.819
Kenny Lofton           100    2730     803     982     945   0.294   0.360   0.346   0.818
Lance Berkman          129    3052     839    1030    1183   0.275   0.337   0.388   0.815

When looking at all 473 players as a whole there is a connection between controlling the strike zone and recording a high NOPS. Below is the scatter plot of all 473 players and their NOPS plotted against the CZoneRate. The yellow line is the linear regression line that shows a pretty healthy positive (correlation coefficient of .456) correlation between the two measures.

Although I’m guessing it won’t offer much consolation at this point, tell your friends that Lloyd McClendon was right.