THE BOOK--Playing The Percentages In Baseball

I don’t understand this.  Outside of run estimators, can you explain why this is?

http://www.tangotiger.net/ops2.html

"Essentially, as the walks and HR go up, I decrease the hits.”

AHHH, because they have the same SLG.  With less BA, you need more power.  Obvious, but somehow I missed it.  Good explanation.

The only reason is because OPS undervalues walks and players with lower BA and the same OPS have more walks (on the average).  That’s it.  It has nothing do with SLG or power.

OPS undervalues walks, got it.  But that’s because it undervalues OBP.  I’m with you to the extent you mean that when choosing between guys with the same OPS, you take the guy with a higher OBP.

But here (in the example) the players have about the same OBP.  Their H+BB would be about the same.  Low-BA guy is going to have fewer H.  But he will have a higher ISO.  That means he has more XBH.  That is what makes him more valuable.

Seems like this is the tradeoff to me: low BA-guy is trading away some 1B and replacing them with BB and XBH.  He loses value when he loses 1B for BB, but he more than gains it back with his few extra XBH.

Bottom line is that OPS is not a very good stat for measuring value.  It is only an approximation.  Something like lwts is. If you take 2 guys with the same OPS and different BA, the one with the lower BA will have the better lwts, on the average.  That was all Patriot was trying to say.  To tell you the truth, I haven’t thought about it that much.

I never really gave this topic much thought either. A couple weeks ago, I would of thought that the player with the higher Batting Average would have the highest linear weights if you compared him to a player with lesser Batting Average and equal OBP and SLG.

What little respect I had for Batting Average is all gone now. It is the most misleading stat out of all the stats that MLB tablulates. It is even more misleading than RBI’s. You do need at least a minimal amount of power to drive in 100 runs. Yes there are exceptions, like Tommy Herr in 1985.

I used to believe that Rob Deer was an awful player. When I was a kid, I remember glancing over the Baltimore Sun’s final statistics for the 1991 season, and thinking that Deer’s .179 Batting Average was a mis-print.

The one thing one can say in batting average’s favor is that, together with OBP and SLG, it can be useful in describing the “shape” of someone’s season.  .275/.326/.400 may not be any different in value from .225/.319/.400 (except for the OBP difference), but those two players produced the value that they did in very different ways, and giving the triple slash stats is a nice and compact way of conveying the manner in which they differed.

I’m pretty sure that MGL is right here. I took a look at player pairs who produced the same OPS in the same number of PAs, with PAs over 35, 1993-2007. There were 468 player pairs.

I then figured wOBA for each player pair, and looked at the difference between wOBA for each player. I then took a look at the difference in walk rate, taking player A’s walks minus player B’s walks, then dividing by PAs. I also did the same with home runs, total bases and hits as a control.

Correlation between difference in wOBA and difference in walk rates: 0.583
Correlation between difference in wOBA and difference in home run rates: -0.206
Correlation between difference in wOBA and difference in total bases rates: -0.496
Correlation between difference in wOBA and difference in hit run rates: -0.224

Colin,

So what does this prove? You’ll have to simplify your explanation a bit.

I don’t know that I’ve proved anything - just tested a few things. This isn’t a full-on study, just a thumbnail sketch of one.

But to lay things out a bit (and keep in mind I don’t have the spreadsheet in front of me right now):

* Players who have more walks will have a better wOBA than a player with an identical OPS in the same number of PAs.
* Conversely, players with more total bases will have a worse wOBA than another player with an identical OPS in the same number of PAs.

Thinking it over, I really wish I had done TB-H as well, as a measure of extra base hits. You can get an idea of it looking at the numbers from HR, TB and H, but it’d be nice to see the actual numbers.