Wednesday, October 12, 2005

ALCS Game 1

For all my blather about "advantages", LAnaheim showed little ill effects of their magical baseball tour. Last night's game confirmed that these teams are just as evenly matched as suspected. Both starting pitchers were solid, giving up only a couple of solid hits each. Anaheim's "big inning" consisted of two singles and three balls that never made it out of the infield. Credit to Orlando Cabrera for busting up a potentially inning-ending double play. White Sox starter Jose Contreras deserved a better fate, but then again, he should have gotten a few more strikeouts with men on base.

Paul Byrd pitched well for the Angels, keeping Chicago off-balance by changing speeds, but the White Sox may well have small-balled themselves into oblivion. First of all, Scott Podsednik continued his sorry streak of basestealing in the playoffs, getting thrown out by a good margin. There was a pitchout on, but frankly, that almost makes it worse. Either the Angels know the White Sox steal sign, or Podzilla has become predictable in his stealing patterns. The green light needs to be changed to bridge out, dead end, or some other more effective signage. When Podsednik gets on base, he needs to stay there. Chicago's best chance to score is multi-run homers from Dye and Konerko while Podsednik and/or Iguchi are on base. Last I checked, you don't need to be on second base to score on a home run.

The White Sox also fell victim to bunt mania throughout the game. Jermaine Dye, who hit 31 homers in the regular season, tried to bunt his way onto first in the 6th and failed. Podsednik failed at a sacrifice bunt attempt in the 8th. Most costly was Aaron Rowand's failed sacrifice in the 9th. Yes, the 9th inning is the best time for a sacrifice bunt to tie the game, but it shouldn't be automatic, especially with K-Rod on the mound throwing hard-to-bunt filth. Wouldn't it be better to take some pitches and try to work a walk, thus achieving the goal of moving the runner up while at the same time putting the winning run on base? If a hitter falls behind quickly, then revert to the sacrifice. But if he gets ahead in the count 2-0, then he is much more likely to walk, or even better, get a hit! It seems to make more sense to wait for a bunt attempt until a hitter falls behind 0-1 or even 0-2 or 1-2. The conventional wisdom is that you shouldn't bunt with 2 strikes, because a foul would produce--gasp--a strikeout! But a player's batting average when behind in the count 0-2 and 1-2 is usually about 150 or 200 points lower than when he is ahead. That's the perfect time to sacrifice!

It doesn't pay to play in fear of the strikeout, or the double play, especially with a fast man at bat like Rowand. You play to win the game, right? The Yankees showed on Monday that K-Rod was vulnerable. Sooner or later he's going to throw a slider that hangs a bit too much. Why not let Rowand wait and see if that pitch comes rather than throwing away his at bat from the outset? I concede that I probably wouldn't be bringing this up if the bunt had worked and the White Sox had scored, but last night proved that even the "safe" play isn't a sure thing. I hope that sometime I'll see some decision made by a manager who plays out of desire to win rather than fear of losing.

I'd also like to mention that Thursday night, I'll be attending a Baseball Prospectus event at the Giordano's 3 blocks from my apartment. Thank you, serendipity! A few BP writers will be there to talk baseball and the NLCS will be on. Combined with the pizza, it's sure to be a good night. I'll be sure to bring up the whole sac bunt when down 0-2 and 1-2 thing, and yes Nate, I will bring up the EqA issue about a particle of light stealing 3rd base. I'll include all their relevant answers in my writeup of NLCS Game 2.

1 Comments:

Anonymous Anonymous said...

re: EqA:

First, it doesn't have to be a particle of light. Stealing is dangerous, yes, but that danger can be captured in CS and shouldn't need to be squeezed out of SB itself. The only other negative impact an SB has on your offensive value is that it means you didn't hit a home run your last time at the plate, and again, this is captured in other stats (H, TB).


The problem I have with the raw EqA formula is two-tiered:

1. For REqAs above 1.5, each additional stolen base lowers REqA. This is clearly wrong. In practice, it's not likely to come up, but I see it as indicative of a real problem with the formula. (For illustrative purposes, Carlos Lee's REqA this year was between 1 and 1.1.)

2. Even in practical cases, REqA accounts a stolen base from a good hitter less valuable than a stolen base from a bad hitter. This might be OK--a stolen base from a poor hitter batting 9th in front of a good hitter batting 1st in the lineup might be more valuable than a stolen base from the excellent power-hitting cleanup man with a poor hitter in the 5th spot--except the diminishing returns with increasing REqA seem too severe, as evidenced by the fact that a steal has no effect whatsoever on a REqA of 1.5. Also, it seems to me that the true value of a stolen base is determined much more by who is hitting behind you than directly by how well you hit. If the current formula really does capture some legitimate diminishing returns from stolen bases with increasing REqA, I suspect it is more a coincidence of the way lineups are traditionally set up than a real effect.

Questions:

-Is the diminishing value of an SB with increasing REqA real, or is the way stolen bases are incorporated into the formula more of a mathematical correction factor? To put it another way, does REqA's use of SBs improve correlation with run production because it's correct or just because it's better than nothing?
-Might it be better to remove SBs from the denominator and change the weight of SB in the numerator (along with the weight of CS in the denominator)? Such a formula has the unfortunate property that a pinch runner who has no plate appearances and has never been caught stealing but who has a positive number of stolen bases has an undefined REqA, but it is at least monotonic in every stat.
-Alternately, might it be better to separate hitting and baserunning into two separate (appropriately weighted) terms? e.g. (H + TB + 1.5(BB + HBP) + SH + SF)/PA + (w1*SB - w2*CS)/(H + BB + HBP - HR + SB + CS), where w1 and w2 are some appropriate weights. (In this example formula, what we really want in the denominator is appearances as a baserunner, but pinch running complicates things.)


-Nate

7:14 PM  

Post a Comment

<< Home