THE BOOK--Playing The Percentages In Baseball

I wrote this in the comments section:

“While the most optimal lineup only increases offensive output by about 5-15 runs a year over a traditional lineup, for teams with weak offenses those marginal runs are all the more important.”

I cannot think of any reason why 5-10 runs would be “more important” for a lousy lineup than for an average or good lineup. 5-10 runs is .5 to 1 win, regardless of the quality of the lineup.

Plus, the Padres do NOT have a lousy lineup, according to my projections at least. In fact, their lineup is above average. SF and PIT have a lousy lineup.

Anyway, I ran your optimal lineup through my sim, which pretty much accounts for just about everything (and uses projections for everything), and the Padres prevailing and conventional lineup (the one you printed above as the “fairly frequent” one).

Versus a RHSP (and then random RH and LH RP’s), your optimal lineup (except I batted the pitcher 9th) does around 6 runs per 162 games better than the traditional one above. And of course, there is no reason to bat 3 LHB in the first 4 slots as Black does. That is just asking for trouble late in the game.

The biggest problem with the current lineup, as you might suspect, is Eckstein batting second and getting all those extra PA. If you do nothing but switch him to 8th and bat Headly 2nd and Kouz 5th, you actually gain 10 runs per 162 games, which is a little better than your optimal lineup…

MGL, if you put Kouz 5th, you have 3 lefties in the first 4 slots don’t you?

The question is how many runs do you lose upfront by stacking your lefties (i.e. intentionally making it suboptimal) versus how many runs do you gain in the occasions where a lefty reliever comes in.

I would guess that gaining runs by lineup optimisation and other such tecniques is more valuable for smaller poorer clubs (as in financialy challenged, not neccesairly poor in terms of W-L success) simply because these clubs have less oportunity to add runs by “brute force”.

On the other hand one can argue that given how simple and cheap it is to do this kind of thing EVERY team should be doing it unless they for some strange reason are not too fond of their money.

Also it might matter where you are relative to the playoff “sweetspot”. If you are far enough away from beeing competitive other things may matter more. (Such as inflating the stats of players you want to trade and deflating the stats of players you want to keep.)

I think that prehaps he means that the 5-10 runs will be more important for a team trying to make playoffs.

For example, if you take a team thats going to win 96 games, 5-10 runs (1 win max) isnt as relevant compared to the 89 win team thats trying to get to 90 and make the playoffs.

The first part here is from my reply to MGL at Driveline:

“Thanks, MGL. I confess that I was hoping to ‘sneak’ the Padres in because after the [depressing] Royals post, I was hoping to have a guy with a projected wOBA over .360 in the top 5. I didn’t do an overall projection for the team, but looking back at it, having more than half of their hitters with an expected wOBA (CHONE projections) over .330 shoud have tipped me off.

As for the runs being “more valuable,” you are right, of course. It would have been more obvious to me had I thought about it in terms of ~10 runs per win instead of relative offensive output, although I guess I could make an argument about budget constraints.”

I need to get one of those cool simulators, or program one myself, which is what I’m guessing you did, MGL. I’ll get on that right after I get the park factors working with my SQL database, doing that aging curves project I’ve been thinking about, make room on my hard drive for Retrosheet, Pitch f/x, clean up my office, doing some actual work, learning how to program…

Thanks for the comments, all. For anyone who has time, I’d love to hear your thoughts/mockery of my idea of having Billy Butler lead off against LHP.

Tango, sure, I wasn’t even paying attention to the LH/RH issue in the lineup I tested. I just wanted to get Eckstein out of the two hole, figuring that that was one of the biggest problems with Black’s lineup.  At first I just switched Kouz and Eckstein, but that was WORSE, even though Kouz is a much better hitter.

My guess is that the L/R thing is important.  IOW, probably the first thing a manager should do is to make sure that he has hit lefties split up.  The reason I guess that is the case is that the opposing manager is obviously going to exploit that weakness mostly in high leverage situations which is going to greatly magnify the loss to the lineup in question.  I would love to see someone try and quantify that loss.  I am going to guess that the difference between having 2 lefties in a row and/or 3 out of 4, versus never more than 2 lefties for every 4 consecutive batters, is worth 1-2 wins a year.

Funny how because Bud Black had such a great reputation when he was with the Angels, for some reason he rarely gets criticized as the Padres’ manager, even by Pads fans (I think).  Yet, his team, the last 2 years, has so dismally underperformed their talent, in terms of wins/losses, you almost have to can him even if you can’t figure out anything specific that he is doing wrong.  Their underperformance last year was almost unprecedented (I would guess).  (That is one reason why almost everyone understated their w/l projection this year, like Tampa last year, and why Devil Fingers just assumed that they had a lousy lineup.  BTW, NO team that has an above average everyday lineup and Chris Young and Jake Peavy in their rotation and Heath Bell as their closer can be a bad team.) So now we have evidence that Black is in fact not a very good manager, at least in terms of lineup construction.  He has one of the worst hitters in the lineup batting 2nd because he is your “prototypical old-school #2 hitter” and he has 2 good players batting 6th and 7th.

MGL:

Could you please go into more detail about pitcher batting 9th issue. Does your program work the way it does because you disagree with the “second leadoff theory” for NL parks, or do you enter it into the program because the program doesn’t recognize how the model works?

Does that make sense? I’ll try to be more clear: I’m just asking if you disagree on the pitcher hitting 8th idea, or if you are just making an adjustment for the program.

Your question is clear, Devil.  My main sim does not allow you to put in the pitchers batting anything other than 9th, or at least it requires a little “work.”

On the other hand, I have tried several times in the past putting the pitcher in 8th and the worst hitter in 9th, and I have always gotten fewer runs and not more.  So yes, I do question the pitcher batting 8th “theory.” Maybe it is correct for certain lineups and not for others. My guess is that it is close to a wash either way.

OK, I ran your optimal lineup versus a RHP both ways - one with the pitcher batting 9th and the other with the pitcher batting 8th.  Keep in mind that I am only running 300,000 games for each, so that one standard error of any difference is around 1.25 runs per year.

With the pitcher batting 9th and your optimal lineup (Hundley batting 8th), the Pads score 4.779 rpg versus a RHP in my sim.  With the pitcher batting 8th and Hundley batting 9th, they score 4,763 rpg, which is 2.6 runs less per 162 games.

The few times I have run pitchers batting 8th, I’ve gotten fewer runs, so in general I am not on board for the pitcher batting 8th.  What I am definitely on board for is the pitcher rarely batting....