Sabremetrics for Dummies – Part 1 of ?

Yes there is a double-entendre in the title.

  • Are you a dummy if you take them to seriously?
  • Do you need to know more about them before you quote them?

Both are a little bit true IMO.

I am not sure how many parts I will end up writing. It depends on the comments and questions.

And let me be the first to admit that due to my background (advanced degree is statistics and a career as a consultant advising companies on how any and all of how to organize, analyze, report on and interpret their data) that I have a distinct (some might say biased) perspective.

The traditional metrics (e.g., BA, SLG, OPS, and so on) are descriptive and objective. They describe what happened. There is no judgment (except for the occasional error called a hit and vice versa) involved here. The numbers are what the numbers are. There is no concept of statistical significance here. There is an issue of what I like to refer to as practical significance – are they different enough to have an impact. For example, last year when Danny was hitting around .200 a number of folks commented that they would be happy to have him start if his BA was at least .230. My view has always been that Danny added value because of his defense and that the difference between .200 and .230 is 18 hits in 600 ABs – or a little less than 1 more hit a week. WWWWeeeellll, IMO Danny saves way more that one hit in two weeks time. And since the point of the game is to score more runs, preventing hits is pretty important too. The point of this example is that context is important. When you look at the numbers, you need the context (channeling Crash Davis here).

Most/many of the Sabremetics on the other hand are different. You can’t look at the box score, see the numbers and say yup. The one I dislike the most is WAR. It is also the most used. But now let me clarify what I mean when I say I dislike it: I mean I don’t like how it is used. It is an interesting number. But it is a prediction. Just like, for example, economic forecasts that predict what will happen, WAR is a prediction of what would have happened IF. . . . And unlike forecasts of future events that can be improved as forecasters get the future data and improve their models, you can’t do that with the WAR metric. We don’t have a parallel universe for each player where they are replaced so we can collect the data. WAR is a prediction and, as such, it has a margin of error. But the margin of error is never included. When comparing two players with a WAR of 10 and a WAR of 6, it matters whether the margin of error is 1 or 5. IMO that makes WAR a number that I won’t put much credence in.

There are a number of other issues with WAR as well:

  • exactly what is a replacement player?
  • how is the value of, for example, Ian Desmond going to the mound and settling the pitcher down measured, included and weighted? I don’t think it is.
  • how is the value of, for example, Jayson Werth have 10 pitch ABs (which helps other players) measured, included and weighted?
  • and so on . . . . .

But that’s enough for now. Your feedback and questions will determine whether and when there is a Part 2 and what it covers.

Jump to Comments Box

This entry was posted in Sabermetrics and tagged . Bookmark the permalink.