My first experiment with "Win Expectancy"

by krushcuts on Apr 10, 2006 6:11 PM CDT

After checking it out a while back on Lookout Landing and http://www.walkoffbalk.com/
I figured, I would make my first attempt at it. Someone, please tell me if I did it wrong. (I know the end result was right since it finished at 100%)
Best performers:
Crede, 0.303
Garcia, 0.284
Konerko, 0.2
McCarthy, 0.182
Thome, 0.119
Jenks, 0.076
Worst performers: (As if any surprise)
Podsednik, -0.114
Mackowiak, -0.099
Dye, -0.096
Pierzynski, -0.091

SouthSideSox is a community driven site. As such, users are able to express their thoughts and opinions in a FanPost, such as this one, which represents the views of this particular fan, but not necessarily the entire community or SouthSideSox editors.

Tweet Comment 13 comments | 0 recs |

Do you like this story?

Comments

Awesome

I actually did one this weekend, but I couldn't remember how to get it to graph right. I kept getting it backwards, (reverse chronologically) so I never posted it. -- (Cliff Politte and Joe Crede were bad if you were wondering.)

I'm assuming you used Studes WPA spreadsheet, not the walkoffbalk stuff manually.

AIM: SouthSideCheat

by The Cheat on Apr 10, 2006 6:22 PM CDT reply actions

Ummm...

Unfortunately, no. I was bored at work and just started messing with the walkoffbalk site. I figured I would try it out... now I am going to have to check out the predesigned stuff.

by krushcuts on Apr 10, 2006 6:29 PM CDT up reply actions

here

It will take some time to figure it out (maybe a few games) but This is much easier (I, think)

ftp://ftp.baseballgraphs.com/wpa

AIM: SouthSideCheat

by The Cheat on Apr 10, 2006 6:50 PM CDT up reply actions

Those are sweet

Nice job. The Studes spreadsheet is easy once you get the hang of it.

If you send me the data I can tweak the graphs a bit with SPSS. See this:
http://www.tufts.edu/~dtybor01/images/USvJapan0312060.JPG

by Tybor on Apr 10, 2006 7:42 PM CDT up reply actions

Cool

I got the spreadsheets from the site and was messing with the macros. Very cool setup.

by krushcuts on Apr 10, 2006 7:44 PM CDT up reply actions

Win Expectancy?

Is that some new-fangled stat-head thing? Next, you'll be telling me that Pods isn't that valuable, that Cliff Politte was lucky last year, and that Javier Vasquez will struggle with the long-ball at New Comiskey...

by madvillian on Apr 10, 2006 8:06 PM CDT reply actions

info on win expectancy

If you're interested:

http://www.hardballtimes.com/main/article/the-one-about-win-probability/

Section Six

by Akula Wolf on Apr 10, 2006 9:53 PM CDT up reply actions

Cool idea

I have always wondered about the odds of a comeback in late innings with certain situations. Is there any way to include the player's stats into it? for instance, if we know that X% of the time you will win a tie game if in the top of the ninth you have a runner on second with no outs, is there a method available that takes into account the odds of the first batter in the top of the ninth getting on base via single or walk and then a steal, or a double?

that was a convoluted sentence, but hopefully someone gets my point.

sue me, i like joe crede
Like pizza? Me too.

by FAQ on Apr 11, 2006 9:27 AM CDT reply actions

Yeah,

If you go to http://www.walkoffbalk.com

You are basically selecting the scenario.

For example, the visitors are upto bat in the 9th, no outs, are down by 2 (-2), and the result of the play is a single (runner on 1st). To see the resulting change in win expectancy, you change the runner position to second.

Bases empty: 5.6% chance to win.
Runner on first: 12% chance. (Player is credited with a 6.4% share)
Runner on second after the steal: 17.3% chance. (Same player gets credited for a 5.3% share)
If the next batter hits a single to put runners on first and third: 29.3% chance. (The new batter gets a 10% share added)

You just change the outs, bases and score during the inning to see the change.

by krushcuts on Apr 11, 2006 9:40 AM CDT up reply actions

ok, but

how does it change with the guy at bat? i mean, the specific guy at bat. so if the bases are loaded and ozuna is batting vs. bonds. bonds has a higher likelihood of walking or getting a hit. ozuna will strike out. has a metric been created to account for the abilities of the player at bat within the situation? it's cool that you can get a good idea by looking at righty lefty matchups and all that, but to actually score the run (or avoid it) you have to look at the batter and pitcher, right?

sue me, i like joe crede
Like pizza? Me too.

by FAQ on Apr 11, 2006 10:28 AM CDT up reply actions

To tell you the truth,

I have no idea where you can find something to predict that. I am sure there are databases and spreadsheets designed, but I would just look at normal statistical indicators like RISP, OBP etc for stuff that can give you a case specific sample-size.

This is my first experience trying win expectancy out since. Jeff at lookoutlanding has been doing it since last year. But the real point of the diagram is not to predict what player A will do when he is up in a specific situation (that is what his season splits will typically tell you). Instead, it shows what contribution player A made towards the outcome.

by krushcuts on Apr 11, 2006 12:25 PM CDT up reply actions