1999 to 2007 - Part I

By Carl in the Bay State

Last year, I tried to answer the question ‘Does Coors have an effect on the pitcher, enough to affect their next performance?’ Little did I know it was going to be so excruciating just to get the correct data for 2006, the season upon which my exercise was based.

And, after obtaining the data, I had to run analysis on it, decide on acceptable formulae for the analysis, graph it, and write it up. This left me no time to prepare for my fantasy leagues, as I had to go right into my drafts for 2007 totally unprepared for anything other than knowing which pitchers look the best at Coors!

So, what do I do this year? I tell Bob in Cleveland and GP that I will update and expand it using more years. They did not put a gun to my head; promise me the first pick in my draft, nothing. Was I nuts to attempt to do this again?

The answer is ‘yes’!!

More years and Better Data - Thanks to Cory Schwartz and MLB.com

So, here I am, doing it again, but with eight times the data I had before. Still, I’ve done it once, so it must have been a little easier for me to do the second time, right?

Well, it was, because I had better data. You see, Cory Schwartz, Director of Stats for MLB.com, made it a great deal easier this time, by providing me with the MLB data for all starters from 1999-2006; he, then, went even further in the fall by sending me the 2007 stats. Boom goes the dynamite!!

The thing I lacked last year was sufficient data points - with nine years of starts that would no longer be a problem, plus the data also depicts pitching trends. For instance, it enabled me to assess the impact of the Humidor, which was brought in before the 2002 season. Take a look at the trends in the graphs as you read this - the information is there.

Keep in mind, this does not look at any relievers or any relief performances by these starters. This is all STARTS, and nothing but starts.

The Data

As noted above, I used MLB Game Logs. This source is wonderful because the data is consistently structured and I did not have to deal with corrupted data downloads as I did last year. I started by identifying the Coors field games. After that, I had to isolate/identify the games immediately following Coors. This pretty much meant doing it by hand, which took a while.

Then, I removed all of the pitchers who did not pitch in Coors for that year because I wanted only the logs for pitchers who pitched in Coors that year, for comparative analysis. With this, I had the Coors games, the starters’ next starts, and the rest of the games for that pitcher in that season.

Effectively, there were 1,458 Coors starters used as the base starting point for this data table - 729 starters for Colorado and 729 starters for their opponents. This also meant nearly as many ‘Games After Coors’ (Next Starts) that had to be analyzed.

Analyzing the Data

To properly analyze this, I had to use stats that summarize a pitcher’s performance. I used three commonly accepted ones - K/BB, ERA, and WHIP. I also tried two newer ones that I had used last year - GmScr and ABA, so I was able to compare 2006 for each as a check.

For those of you who aren’t familiar with those last two, I’ll provide the background at the time - it’ll save you having to refer back to this article when we come to it.

The Results

Over the coming couple of days, the results of this analysis will be published in simple, easy-to-manage chunks, looking at each statistic in isolation.

And some of the results will come as a surprise - I promise you!

CITBS