THE BOOK--Playing The Percentages In Baseball

I’m confused.

“The reality is that you don’t even have to start the game from scratch, because just keeping a running total of runs scored and allowed will give you the same answer.”

Is this a suggestion of a change MLB could make, look at run differential instead of W-L?  The concept of leverage would go out the window.  You have just as much reason to use Mariano Rivera in the 4th inning of a game you’re losing 18-2 as the 9th inning of a 2-1 game.

Or just the observation that for this year, R-RA closely matches W-L?  It’s not always the case.

Right.  I’m not suggesting that they do it.  But, that if they had done it for 2011, no one would have noticed the difference.

That’s quite jarring for a fan, isn’t it?  That the idea of clutch, of winning close ones, the 16-inning games, etc.  All that, essentially, meant nothing.  That they should simply have a running total of their 1200 innings so far, by just counting runs.

Isn’t this obvious? I’m not saying that to be flip or smart assy or anything of the sort.

But, since most wins and losses are not blow outs, wouldn’t wins and losses be expected to have a very strong correlation?

For run differential and winning percentage to not “strongly correlate” a team would have to have a bunch of lopsided wins and/or losses, which over the course of a season is very unlikely.

Maybe I am missing the point being made. If that’s the case, I apologize.

That is interesting to think about it that way. Do you think that a manager could manage so as to break this?

I.e., by manipulating his bullpen mostly, bringing in Rivera once every other day in the 9th inning regardless of the score. If what really mattered was a running total of runs scored and allowed (but no one else knew...), could he win the overall season total but lose more individual games.

Anyways, my point really is that if things worked that way (9 innings every day, no wins or losses, just sum up the runs), then managers would do things differently.

It would certainly be a more boring sport (yes I know you’re not arguing they should actually do it, I’m just commenting).

If you thought cricket matches were long at three days, you ain’t seen nothing yet! We’ve got 6 months for one game (against 15 other teams all at the same time)!

The practical implications of division standings being based on run differential could be fun since it would change how team’s respond to blowouts. For example, you could see a team bring in its closer to preserve the differential of the current game even though they’re down by 10. Likewise, hitters may remain more focused late in blowouts if they know tacking on a couple runs would help them pick up ground in the standings.

CC: It is not obvious that you would get an almost perfect ordinal ranking of W/L or runs differential.  (Of course, you are limited to 4-6 teams per division.)

We’re not talking about “strong” correlation.  We’re talking about almost perfect correlation.

Angels outperform their pythagorean every year.  They’re doing it again this year.  I think it was Neyer who wrote about this and decided it was a fluke and that in 100 years of team-seasons you were bound to have anamolies like this.  Which struck me as a classic example of someone having drunk too much of their own kool-aid.

Anyway, it’s not almost perfect correlation.  it’s got a standard deviation of however many wins (2? 3?) that is about what you might guess.

In a sense baseball already does something like this in each game.  There are 9 innings (sometimes more) in a game.  They don’t keep track of how many innings are won.  I wonder if innings won/lost percentage also has the same ordering.

read on another thread that the std dev on pythag wins is 3.95.  “2/3 of teams will win within 4 games of what you would predict based on their RS and RA” puts a different spin on the issue compared to “look, you could just track RA and RS and forget about W and L”. 

Sometimes, I’m not sure if Rob is a baseball analyst or a SABR concept throat rammer-downer....

Pierre we’re talking about ordinals.

right.  I guess I assumed there was some larger point being made.  Yes, the standings are pretty spread out this year....