The Baseball Graphs Blog
Friday, January 04, 2008
New Historical Win Shares File
Hello! I haven’t been by in a while, and I feel just sick about it. Hopefully, if you like my work, you’ve been reading my stuff at The Hardball Times. Keep on checking out THT, cause I’m not going to pick up things here for a while.
But I did want to let everyone know that I’ve updated the historical Win Shares file to include 2007 Win Shares. You can download this zipped file from ftp://ftp.baseballgraphs.com/winshares.
I’ve changed the format this year. There are now just two .csv files, which you can import into Access. If you want to import them into Excel, you’ll have to chop them up a bit cause they’re just too dang long. The first file contains all the detailed Win Shares info for every player from the past to the present. The second file contains “expected Win Shares” info for every player and year since 1900. You can use this second file to create your own “Win Shares Above Replacement” data.
Win Shares suffers from a lack of “Loss Shares.” This sounds like it might not be a big problem, but it is. It’s the difference between accruing 10 Win Shares in 40 games or an entire season. 40 games is a lot more impressive. The expected Win Shares data helps you get over this hurdle.
Expected Win Shares are the number of Win Shares an average player would have accrued, given that specific player’s times at bat, innings in the field and batters faced from the mound. It’s based on specific averages from each league and year. So if Hank Aaron had 29 Win Shares and 17 expected Win Shares in 1955 (he did), then you could say he was 12 Win Shares above average. Taking it a step further, you could even say he was 29-5 in 1955 (the year of my birth, by the way).
There’s a problem with the won/loss approach. Some players had such good years that they accrued more than twice as many expected Win Shares, so they would wind up with “negative” loss shares. So it’s not a perfect approach. But this info does allow you to get beyond the lack of Loss Shares.
The best approach is to calculate Win Shares Above Replacement Level, which is what we do at the Hardball Times. With this data in these files, you can now calculate your own replacement level. Here’s a link to some work I did about Win Shares replacement levels a couple of years ago. The replacement approach accomplishes the same thing as Loss Shares.
Anyway, have fun with the data. I took a lot of extra time this year to make sure I got all the IDs correct, but you may find a mistake or two. Or more. Please let me know if you do.