But today I wanted to look at different aspect: its predictability. That is, if a catcher is good at framing in one year, can we reasonably assume that he'll be good at it the next year? One way to look at this is to look at all catchers and how they did from year to year. If they did well one year, did they do well the next and vice versa?
(I wondered about this because I was looking up Kurt Suzuki's framing numbers. They're usually been negative, but there have been some positive ones sprinkled in. I wondered how common that is.)
There is a neat little statistical gizmo to do this called a correlation coefficient. A correlation coefficient examines two sets of numbers and gives back a number between -1 and 1.
- 1 means there is a perfect correlation, like between the temperature in Celsius and the temperature in Fahrenheit.
- -1 means there is a perfectly negative correlation, like the amount you spend in a month and your checking balance.
- And if it's 0, that means the numbers have no correlation, like Joe Mauer's batting average and the migratory penguin population.
You can find the results of my study here.
The bottom line: there is a lot of predictability. The runs per season had a correlation of .76, which is high. But the correlation on pitches per game is even higher .82.
So catchers who have had a large positive effect end up continuing to do so. Unfortunately, most of the Twins who will play catcher this year didn't have a large positive effect last year. In fact, none of them did:
Kurt Suzuki: -9.1
Josmil Pinto: -4.3
Eric Fryer: -0.8
I didn't choose that order to emphasize the negative. I chose to list them in my predicted order of innings caught. It's almost as if the worse they are at pitch framing, the more likely they are to play catcher. And this is where John starts rubbing his temples.
And yet, that still might be better than last year. Because last year Joe Mauer was average (0.4) and Ryan Doumit was horrendous (-15.9). Still, it appears the new catching corps may not do the Twins revamped pitching staff any favors this year.
Since I'm sure you might want to do something like this yourself (and really, why wouldn't you - YEAY MATH), I thought I'd spell out the steps.
1. I pulled all the data I could from this great site and pasted it into a spreadsheet. It has all the catcher framing data from 2013 through 2007.
1b. I forgot to mention - I also limited the study to catchers with at least 3000 "samples" in a season which I assume are pitches.
2. I added one column to that data: "Prev Yr." You'll see why in a minute.
3. I imported that spreadsheet into an Access DB twice, once as a table called "Following" and another as "Previous".
4. I created a query joining those two tables, joining fields First Name, Last Name and "Prev Yr" from following to "Year" from the Previous field. I pulled the Names, Years, Per Game and RAA fields from each table.
5. Copy and paste the results from the query back into an Excel spreadsheet.
6. Use the "Correl" function to compare the values in the two "Per Game" and two RAA" fields.