from Amazon
Wednesday, January 24, 2007
How Did Ted Williams Do Against Lefties?
There is a thread going on at Baseball Fever that discusses Ted Williams. In the very first post, there is some fascinating data. Below is my analysis of that data, and I encourage others to take on a more exhaustive study, especially focusing in on lefty hitters at Fenway.
The first post in that Fever thread is the most important. I don’t know how I missed it, but it’s worthy of further analysis.
The first part of that post is his % of HR against LHP. It comprises all his HR, and therefore, we have no worry about bias. Only 12.3% of his HR were against LHP.
Based on data that I published at Google Docs:
http://spreadsheets.google.com/pub?key=pkimQBCeCjbhd5SG54YpzRg
We see that LHP make up about 25% of the population in 1940s and 50s.
However, Fenway is notorious for LHP, and teams (including the Sox) stack up their RHP in Fenway. This is what we have for 1957:
http://www.retrosheet.org/boxesetc/WBOS01957.htm
Only 20% of the PA of Boston hitters were against LHP. That number drops down to 15% in 1958, and 18% in 1959.
It’s possible therefore that Ted Williams faced only about 70% of the LHP that other great LHH faced. And, it’s also likely that the LHP that Ted faced were better than what other hitters faced.
What we really care about is HR per opportunity, and that first chart doesn’t give it to us.
If you focus on Yaz, you will see that he only got 17% of his HR against LHP (and Fred Lynn, who spent part of his time in Fenway, was also a low 20%). We have Yaz’s career totals, and 24% of his PA were against LHP. Yaz had 2.3% of his LHP PA as HR, and 3.5% of his RHP PA as HR. (Fred Lynn is 2.8% and 4.3%). Lynn and Yaz got around 53% more HR, per opportunity, against RHP than LHP.
If we presume that Ted Williams faced only 17.5% of his pitchers as LHP, then his HR/PA rate would be 3.7% against LHP and 5.7% against RHP, or about 51% more against RHP.
Now, let’s go on to the second chart in that first entry.
Based on that data, 22% of the partial data is Ted against LHP. That number if probably a bit too high. (As stated, we should expect to see something closer to 15%-20% for the 1940s and 50s). So, the data is not random, and may be biased. It may exclude any bad LHP and alot of RHP data.
Continuing with the partial data, his HR/PA was 2.9% against LHP and 5.7% against RHP. My guess was that it should have been 3.7% and 5.7%, respectively.
So, it’s likely that the RHP data was fairly random (or at least representative), while the LHP data is biased.