clock menu more-arrow no yes mobile

Filed under:

springing forward

in case you missed the news, st louis will host the 2009 all-star game.

the PECOTA projections have posted at Baseball Prospectus. they're subscriber-only figures, a proprietary product, so i'm not comfortable publishing the entire set of stl projections; that'd essentially be stealing. i see nothing wrong with providing some sneak peeks, however. at an initial glance, PECOTA seems to like the st louis hitters slightly more than ZIPS does while being about the same on the pitchers. carpenter projects to a 3.39 era (30 points higher than ZIPS), but reyes and wainwright are at about the same level that ZIPS has them (the wainwright projection has him as a starter). braden looper, projected mainly as a starter (20 starts, 12 relief appearances, 123 total innings), checks in at a 4.99 era. brad thompson is projected as a relief pitcher and falls in the same range era-wise as he did with ZIPS. as far as the hitters, PECOTA likes adam kennedy considerably better than ZIPS does (.728 ops); duncan has a solid projection (.844 ops), edmonds is in the same range, j-rod shines, and encarnacion plods along at his accustomed level of middleness. we should not expect much from eckstein or molina, PECOTA says.

i'll compile team aggregations for the pitchers and hitters and put up the results thursday, but i wouldn't expect the final wins projection to deviate much from the ZIPS-derived total.

while they're still fresh, i'll use the PECOTAs to conduct another little experiment. you may have seen the recent article at Beyond the Boxscore rating chris duncan as one of the best #2 hitters in baseball during 2006. even if his ops drops into the mid-.800 range this coming season, he'll still be an outstanding #2 hitter. so here's the experiment: since duncan's so good, what happens if you keep the batting order the same, but begin each game with the #2 slot in the order? like this:

1. eckstein 1. duncan
2. duncan 2. pujols
3. pujols 3. rolen
4. rolen 4. edmonds
5. edmonds 5. encarnacion
6. encarnacion 6. kennedy
7. kennedy 7. molina
8. molina 8. pitcher
9. pitcher 9. eckstein

same lineup; same sequence of hitters. you simply offset the initiation of the sequence by one slot. i liken this to daylight savings time, wherein the hours proceed in the same sequence but we "spring forward" by one hour, arbitrarily renumbering the slots in order to make more efficient use of the sun. by springing forward in the lineup and renumbering the slots, might we put the heat and light in the batting order to more efficient use?

emphatically, we might. here's the basic tradeoff we're making. we gain offense by moving the better hitters up one slot in the order, because doing so gets them all more plate appearances over the course of the season. a #1 hitter gets slightly more plate appearances than a #2, because there will always be a certain number of games where the #1 makes the last out, and the game ends with the #2 hitter on deck. a slightly smaller number of games will end with the #3 hitter on deck, and so on down the line. so by shuffling the lineup in this way, we're basically taking 15 or 20 plate appearances a year away from david eckstein and giving them to chris duncan. that seems like a sensible thing to do, insofar as duncan projects to have an ops 100 to 150 points higher than david's; kind of a no-brainer, actually. pujols would get nearly as many extra plate appearances a year; that can only help the offense too, no?

now, here's the cost of that benefit: we're diminishing the table-setting capacity in front of albert the 1st time through the order. he's more likely to bat with the bases empty in that at-bat, and of course duncan will always bat with the bases empty 1st time through. for subsequent times through the order, albert has the same two guys batting in front of him (eckstein and duncan) as always, but he will take less impactful at-bats the 1st time throught the order. on the other hand, because duncan and pujols get on base with such regularity, edmonds and rolen should have more impactful at-bats first time through the order, and that mitigates some of the cost.

to get a crude read on this, i plugged the PECOTAs for both lineups into david pinto's lineup toy to see how many extra runs --- if any --- such a lineup change might produce. here are the results:

runs / game 4.929 5.060
runs / 162 798 820

an extra 22 runs; two wins in the standings. lot cheaper and easier to pick up two marginal wins this way than by paying, say, $42m to jeff suppan or $100m to carlos lee, eh? in truth, the increase would probably come out to more like 15 runs / 1.5 wins, because as we know teams don't use the exact same lineup for 162 straight games; 'specially not la russa-managed teams. and since duncan is basically a platoon player at this point in his career, you wouldn't see this lineup vs lhp. so maybe we're only talking about one extra win in the standings. what the hell; it's a free win. why not take it?

i ran one more test, for this lineup: duncan / edmonds / pujols / rolen / encarnacion / kennedy / molina / pitcher / eckstein. here's that result:

lineup #2
runs / game 4.929 5.060 5.018
runs / 162 798 820 813

obviously you wouldn't bat duncan and edmonds 1-2 vs a left-handed pitcher, but vs right-handers you might reap some extra runs / wins--- and for free.

la russa got flamed when he started batting the pitcher 8th in the second half of 1998, but he had the right idea. it wasn't a matter of moving mcgwire forward in the lineup; he batted 3d no matter where the pitcher batted. but by putting a position player 9th (most often eli marrero), la russa increased the table-setting capacity that came up in front of mac --- and the offense improved. in 79 games with the pitcher batting 9th in 1998, the cardinals scored 391 runs, or 4.949 per game. with the pitcher batting 8th, they scored 419 runs in 83 games, or 5.048 runs per game. with the pitcher hitting 9th, the team was 39-40; hitting 8th, they went 44-39. that's a small sample size, so it doesn't prove anything; but it at least provides some real-world data that are consistent with the predictions of the theoretical models.

this will never happen in a million years; it's merely a discussion for a january morning. but when you see teams going to such desperate lengths, stretching for that extra marginal win that might put them into the playoffs, you wonder why some darkhorse contender --- say, the 2007 giants, or the cleveland indians --- doesn't try an experimental lineup over a full season and see if they like the results.

of interest elsewhere: