What are the keys for pitching at Coors Field?

Jordan Lyles (Getty Images)

Para la traducción al Español, haz click aquí.

For more than 20 years now, pitching in the high altitude of Denver has been a difficult riddle to solve for virtually every member of the Colorado Rockies pitching staff. In fact, only five times out of twenty-one seasons,have the Rockies cracked the top 10 in ERA-, a stat that compares ERA with league average after being adjusted for park and league. An ERA- of 110 for example, means 10% worse than average, while 85 means 15% better. Just seven times – a third of the seasons – they managed to finish with an ERA- under 100. That is bad pitching, no matter how you look at it. There are a number of reasons why successful pitching has been so hard to find for Colorado, and some of them have to do with altitude.

The ball carries farther in the thin air. Knowing this, and in order to reduce the amount of home runs, they built a spacious outfield for Coors Field, which led to a lot of singles, and balls hit to the huge gaps. Also, a humidor was put in place in 2002 to store balls and keep them from hardening and jumping lively off the bat. Last but not least is the fact that thin air reduces the Magnus Effectwhich is what makes a baseball’s trajectory bend, as a result of their rotation. All of this combines to a very strange, awkward place to pitch.

The Humidor

The Humidor (AP)

The Rockies have, however, found some amount of pitching success in the last decade. Of the five times they finished among the 10 best ERA- in the majors, four of them were between 2006 and 2010. So if we take data from those years, and mix it with some awful pitching years like ’04, ’05, ’12, and even ’13, we have a wide array of results that we can analyze, and try to see what the keys are to have success as a Rockies pitcher.

The first thing we’ll do is establish a measuring stick for pitching success.  For that purpose, we’ll use FIP, which is a stat that resembles ERA, but is calculated using only events that pitchers have control over, namely: HR, BB, HBP, and K. This is especially fitting for analyzing pitching at Coors, where more hits are given up, resulting in more runs surrendered, and higher ERAs.

The next step would be to select a timeframe to focus our study. As a general rule, the more data you have, the more reliable the results. However, as I mentioned above, the humidor was introduced during the 2002 season. That changed pitching in Coors quite a bit, so I think it’s convenient to draw the line there and select 2002-2013 as our timeframe. For the purpose of this study, we’ll use only starting pitchers stats.

The last step consists of taking a set of peripheral stats of every individual qualifying season by a Rockies pitcher, and calculate a Pearson Correlation Coefficient between each stat and our success indicator, FIP.  Ideally, the set of stats with the highest correlations to FIP, will be the keys for success.

There is an important fact, though, that has to be taken into account. There are certain stats, that translate into success no matter what team or park the pitcher is playing on. So in order to compare and isolate the ones that better fit Rockies pitching, we’ll do the same procedure for all Non-Rockies pitchers in the same timeframe.

As a rule of thumb, a correlation coefficient of 0.70 or more (shaded red) indicates a very strong relationship between the stat and FIP, a coefficient between 0.40 and 0.69 (shaded orange) indicates a strong relationship, between 0.30 and 0.39 (yellow) indicates a moderate relationship, and 0.29 or less (white) indicates a weak or negligible relationship. Additionally, correlations can be positive or negative. A positive correlation means a direct proportion between the variables, while a negative one means an inverse proportion.

Without further ado, here are the results, in descending order of Pearson Correlation Coefficient magnitude:

Correlation of assorted stats with FIP
Stat Rockies Non-Rockies
HR% +0.76 +0.65
K/9 -0.48 -0.61
wFB -0.48 -0.60
BB/9 +0.42 +0.43
FB% +0.39 +0.24
wSL -0.35 -0.33
GB% -0.34 -0.24
wCB -0.25 -0.28
wCH -0.24 -0.28
SwStr% -0.16 -0.52

wFB, wSL, wCB, and wCH are pitch type linear weights values for fastball, slider, curveball, and changeup. Pitch type linear weights usually range around -10 to +10, and indicate success with a particular pitch, based on changes in run expectancy, you can read more about them here.

It’s not a surprise that three of the four highest correlations are for HR%, K/9, and BB/9, since these stats are the main components of FIP. In any environment, these three stats are going to have a big influence on a pitchers success. However, there are some differences between the two columns in that table that are worth noting.

Suppressing home runs seems to be more important for Rockies pitchers than Non-Rockies, as evidenced by a stronger correlation with FIP. Being the homer-friendly park that is Coors, for a pitcher to have the ability to keep home runs to a minimum there is highly convenient, more so than in parks like AT&T Park in San Francisco. This has a lot to do with two other stats that differentiate pitching in Colorado from the rest of baseball’s venues: fly ball and groundball percentages.

The current popular theory is that groundball pitchers tend to do better in Coors than fly ball pitchers. The reason is pretty obvious: more groundballs means less fly balls, less fly balls means less home runs, and less home runs means lower FIP. As much as this idea seems foolproof for every pitcher in baseball, the numbers suggest that there is very little correlation between GB% or FB % and FIP among Non-Rockies pitchers, while there is a moderate to strong correlation among Rockies pitchers.

On the flip side, strikeouts and swinging strike percentage are less of a factor for Colorado hurlers. In the case of K/9, there’s a strong correlation for both Rockies and Non-Rockies pitchers, but it’s considerably higher for pitchers not wearing purple. In the swinging strike percentage, the gap is pretty big. Making batters swing and miss is a very important key for pitchers in general to have success, it just doesn’t seem that way for the Rockies, at least among starting pitchers. Swing-and-miss pitchers are generally fly ball pitchers, and we’ve established that they don’t do well at Coors, so it makes sense that SwStr% has a negligible correlation with FIP for Rockies pitchers.

Summarizing the results shown above, it’s important to keep GB% high, FB% low – in order to keep HR% low – and it’s not as important to strike batters out or make them miss.  It’s clear that the numbers suggest that the usual assumption of a groundball pitcher being a good fit for Coors Field is correct.

So how does a pitcher generate high groundball percentages? Well, it’s important to keep the ball down. We know the Rockies know this, and that’s why they install horizontal strings at the plate during spring training, indicating the lower part of the strike zone. But it’s also important to have effective breaking pitches.


Source: Official Instagram of the @rockies

Thankfully for us, there’s a really useful technology called PITCHf/x, which tracks the ball’s movement through a set of cameras, and has been active in all major league parks since 2007. One of its many uses, is to determine the changes in trajectory caused solely by the ball’s rotation. I wanted to see how vertical and horizontal movement in pitches, influenced a pitchers success. So I did the same calculations used for the other stats, only this time for movement values of the four main pitches: Fastball (four-seam), Slider, Curveball, and Changeup. The first four stats on the following table indicate horizontal movement, while the other four represent vertical movement.

Correlation of movement in pitches with FIP
Stat Rockies Non-Rockies
FA-X 0.22 0.14
SL-X 0.35 0.04
CU-X 0.22 0.06
CH-X 0.40 0.06
FA-Z 0.01 0.03
SL-Z 0.54 0.16
CU-Z 0.30 0.16
CH-Z 0.30 0.10

The results are pretty clear here, and the first thing that comes to mind is: movement is good… if you’re pitching for the Rockies.

As I stated before, the thin air of Denver reduces movement caused by the rotation of the ball, so achieving great movement on pitches for the guys on 20th & Blake is not an easy task. With the exception of the four-seam fastball, these numbers tell us that there is an important correlation between movement in general, and pitching success. The most interesting fact, is the huge difference in the results obtained for Non-Rockies starting pitching. The data suggests that the amount of movement on these pitches, is not a necessarily a determining factor to a pitcher’s success if they play for any other team.

All of this confirms the profile that the Rockies’ Front Office looks for when it comes to acquiring starting pitching recently. Instead of going for the big strike out guys, they’re going for the groundball type, with good breaking stuff. This is what got them interested in Brett Anderson, it’s what they liked about Jordan Lyles, and it’s also part of why everyone is so excited about Eddie Butler.

About Juan Pablo Zubillaga

Venezuelan die-hard Colorado Rockies fan. In addition to watching and analyzing as much baseball as I can, I'm also a Chemical Engineer // Venezolano, fanático de los Colorado Rockies. Además de ver y analizar la mayor cantidad de béisbol posible, también soy Ingeniero Químico.
Add Comment Register

Leave a Reply

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>