Friday, February 25, 2005

larry leads off? yes!

re the advisability of batting larry walker leadoff: the results are in.

james click of baseball prospectus generously ran a quick computer-simulation study on the cardinal batting order at my request. the data suggest that the cards might indeed score a few more runs with larry leading off — but only if the other big bats (pujols edmonds and rolen) bat immediately behind him in the two thru four slots. that’s the lineup josh schulz proposed at birdwatch on wednesday.

the advantage derived from this unorthodox sequence is not huge — a margin of 9 runs, on average, over the default setting (ie, eckstein, walkerpujolsrolenedmonds, grud’k/sanders or sanders/grud’k, molina, pitcher). that finding is consistent with what stewie13 has pointed out in comments here and at birdwatch, and what tangotiger said in a recent post at baseball musings: lineup manipulations can only net you, at best, an extra 5 to 15 runs. but tangotiger adds that 5 to 15 extra runs equals one extra win, and teams throw millions at ballplayers whose marginal value equals one win. if you can net a win for free just by moving guys around in the order, why wouldn’t you?

before we move on to the complete results of click’s quick study, i am compelled to advise you, per the blogger’s oath and the published regulations of the fcc and dept of homeland security, that the following exercise is completely theoretical and that the predicted 9-run gain may be as ephemeral as iraq’s weapons of mass destruction. moreover, click offers a caveat of his own: “Keep in mind that this isn’t meant to map to actual runs scored, but is just for comparative purposes between the various lineups. Finally, the standard deviation is around 40-45 runs, so while lineup C certainly stands out from the rest, the confidence level is low.”

ev’yone get it? this is for entertainment purposes only.

i asked click to run comparisons on five diff’nt batting orders. he ran each one for a thousand seasons, then sent me the low, mean, and high run totals (per 162 games) for each order. in reverse order of potency, with mean averages, the lineups were:

LINEUP D: 964 runs
1 eck’n / 2 grud’k / 3 pujols / 4 edmonds / 5 rolen / 6 walker / 7 sanders / 8 molina / 9 pitcher (low 826, high 1133)

LINEUP A: 970 runs
1 eck’n / 2 walker / 3 pujols / 4 edmonds / 5 rolen / 6 grud’k / 7 sanders / 8 molina / 9 pitcher (low 830, high 1120)

LINEUP B: 970 runs
1 walker / 2 eck’n / 3 pujols / 4 edmonds / 5 rolen / 6 grud’k / 7 sanders / 8 molina / 9 pitcher (low 827, high 1117)

LINEUP E: 975 runs
1 walker / 2 pujols / 3 edmonds / 4 rolen / 5 sanders / 6 molina / 7 pitcher / 8 grud’k / 9 eck’n (low 847, high 1101)

LINEUP C: 979 runs
1 walker / 2 pujols / 3 edmonds / 4 rolen / 5 grud’k / 6 sanders / 7 molina / 8 pitcher / 9 eck’n (low 840, high 1143)

i note, first of all, that the question that launched this inquiry — would the cards score more by flip-flopping walker with eckstein in the top two slots in the order? — is completely moot according to click’s study: both variations averaged 970 runs (lineups A and B). i note further that the “best” lineup, C, is in fact merely a renumbered carbon-copy of the “default” lineup, A. think daylight savings time — we follow the same cycle but arbitrarily displace the starting point by one unit. the “day” isn’t any longer, but by “springing forward” from david to larry we allocate the sunshine more advantageously . . . . . which in this case means (i think) that we allocate about 60 - 80 of eckstein’s at bats to walker pujols edmonds and/or rolen.

does that not sound like a complete no-brainer thing to do? even if it didn’t net you a single run, wouldn’t any baseball fan at a given ballgame rather see one of those four guys get an extra at-bat in the 9th inning, as opposed to eckstein?

add’l thoughts: pujols does not fit the ideal profile of a #2 hitter, as defined by tangotiger’s studies; he hits too many hrs and doesn’t draw enough walks. jimmy edmonds more closely matches the thumbnail, as does walker himself. but if you line walker and edmonds up 1-2, it leaves the cards vulnerable to lefty starters and left-left relief specialists (this pair of batters alone could keep jesse orosco and john franco on mlb payrolls until their iras mature.) might it be permissible to sneak, ahem, eck’n or grud’k into the leadoff hole vs a starting lhp? . . . . here is where we begin to run into the limitations of our study, which click himself would be the first to concede. the simulation program omits dozens of variables that might change the “optimal” lineup on a given day — platoon data, batter v pitcher data, ballpark considerations, left/right alternation, etc etc. it also does not account for pinch hitters, the double-switch (a la russa perennial), or even such quaint phenomena as “hot streaks” and “slumps” — which used to be very legitimate lineup-selection criteria back before the sabrevolution.

in any case, whatever lineup la russa employs today, we know he’s going to use a diff’nt one tomorrow, and yet another one the day after; that’s just tony. do i think he should use lineup C above as often as possible? hell yes. but if he does it even once, the cards damn well better win that day — ‘coz if they lose he’ll never hear the end of it.

and we’ll never see that lineup again.