Thursday, May 19, 2005

Mini-FAQ on Win Value

The most frequently asked question on Win Value so far is a variation of this: "Player hits a go-ahead home run in the 6th. Later his team scores 6 more runs in the ninth, and wins by 7. Do these later runs affect the Win Value for the home run?"

The answer is no. Win Value is determined at the time of the event, and is not affected at all by subsequent events. When that player hit that go-ahead home run, if the game played out according to the averages, the home run would have a certain value in that context. That is Win Value. The fact that the game turned out differently than the average shouldn't change the value of the home run. That is the only way to provide a reliable benchmark against which different events can be measured and compared.

For more detail, please see this post

2 Comments:

At 10:42 AM, Blogger El Gran Color Naranja said...

The assumption you have to make for WV to make sense, that these situations we are dealing with are nearly identical, is a large one. If the 6 inning home-run actually set the chance of winning to a strict percentage, I'd be more comfortable with WV, but it really doesn't. For an individual game, the conditions, batting order, pitchers involved, in-game decisions, will all effect the actual percentage chance of winning wildly.

I'm wary of using aggregate data to assess the individual situation, when the data is available for the individual situation.

 
At 10:56 AM, Blogger DM said...

I'm not sure I understand your point completely, but I'll think some more about it. Keep in mind that WV (and RV) are not designed as an absolute measure of value, but a relative one -- value relative to what the average team does. It's primary use is to compare past performance with other past performance in a more meaningful way.

I'm also not sure the "data is available for the individual situation". If you have access to it, take a look at Baseball Prospectus's "Win Percentage" chart, which indicates the chance of winning from any point in the game (inning, outs, score difference). The number of instances for many of the situations is way too small to be useful -- you come up with odd conclusions like it is better to make an out than to walk, because it happens that the "out" situation resulted in more wins than the "walk" situation, just because they were so few examples to work from. Maybe someone has compiled all of these situations for the past 40 years, but I don't know.

 

Post a Comment

<< Home