Lining-up your 2010 Diamondbacks

With the acquisition of Adam LaRoche, the roster now seems fairly solid, and the starting position players in particular appear largely set. There's already been some discussion over what the best batting order would be for the Diamondbacks in 2010, including some use of of an automated line-up analysis tool. This was the same one used to investigate Arizona before the start of this season, and there were some surprising, non-traditional line-ups suggested as a result.

This year, I'm refining the approach a little bit, by separating the analysis into two line-ups: one for when Arizona faces a left-handed starter, one for when we face a right-handed starter.

Here's the starting line-up I used:

• C Miguel Montero
• 1B Adam LaRoche vs RHP, Conor Jackson vs. LHP
• 2B Kelly Johnson
  
• SS Stephen Drew
• 3B Mark Reynolds
• LF Jackson vs. RHP, Ryan Roberts vs. LHP
• CF Chris Young
• RF Justin Upton

So, in terms of offense, the only change is Roberts replacing LaRoche against left-handed starters. I'm not sure if we actually will do that, but I did want to inject a variation into the line-up, to see what difference this might make. The next question is what numbers do we use? While we have projections from various sources, these are not broken down based upon the handedness of the pitcher. This is particularly important for someone like Justin Upton, who has murdered lefties to a tune of 210 OPS points more than righties so far in his career. This is not reflected in the amorphous blob of an overall projection.

For ease, I'll simply use a player's career splits to date. Sample size in some of these vs. LHP is smaller than I'd like e.g. Montero has only 148 PAs, so take that into consideration. The only case where I've balked is Roberts - his career total is 140 PAs, and I don't think  we can expect a .932 OPS going forward. So I've gone with the straight CHONE 2010 projections for him [the resulting OPS of .756 is also close to LaRoche vs. LHP, of .751] The table below shows the appropriate stats for the nine players involved in our hypothetical 2010 Arizona Diamondbacks line-up, along with the PAs, to give some degree of reliability for the numbers. Apart from LaRoche and Johnson, the players have spent virtually all their major-league careers in Arizona, so that helps.

We then take these numbers and plug first the vs. RHP numbers, then the vs. LHP numbers into the Lineup Analysis tool. This uses straight OBP and SLG to figure out the 'optimal' line-up for a team, and the expected runs [There are links on the page if you are interested in learning more about the math behind the line-ups it creates. However, that will not be on the test at the end of the year, so I would not be inclined to worry about it too much...] For both sets, I've used the 2009 team total for pitchers, which was .169/.194/.217. Here are the best-scoring results for both sets of stats, based on the 2002-04 model

One shortcoming of the tool is that it doesn't care if it aligns all the left- or right-handed batters together, and we can make some minor tweaks to do better in this area. We could swap Jackson and Johnson to help in this area - all the top line-ups have Johnson or Jackson leading off, with virtually no impact on production. Same for Montero and Drew: a lineup of Jackson(R), LaRoche(L), Upton(R), Reynolds(R), Montero(L), Young(R), Drew(L), Pitcher(R), Johnson(L). is projected at 4.718, only about 1.3 runs less than the best line-up over 162 games. It doesn't seem too far off what we might see.

Against left-handed pitchers, things are radically different. All the best line-ups agree that Jackson needs to hit lead-off,  with Upton in the clean-up spot, Drew sixth and Roberts ninth, after the pitcher. But the other spots are almost interchangeable. Again, aiming to alternate LHB and RHB, we can make some tweaks without derailing things too much. Going Jackson(R), Reynolds (R), Montero (L), Upton (R), Johnson (L), Young (R), Drew (L), Pitcher (R), Roberts (R), is a line-up that is still estimated to be good for 5.482 runs per game.

To put that into context, if we faced LHP every day, that'd mean more runs in a season than any team in the National League over the past couple of years. Powered by Upton's and Reynolds's destruction of lefties, we should kill left-handed pitching. Of course, much the same players conspicuously failed to kill left-handed pitching this year: we went 17-30 when facing a southpaw starter, more losses than any other team in the NL. Seeing what happens next season, might provide an early metric on which to gauge the team's progress in 2010.

Assuming a) the projections are correct, b) we use the modified "best" lineups (4.718  and 5.492 runs respectively) and, c) we face left-handers in 51 games, as in 2009, the expectation would be for Arizona to score a total of 804 runs next season. That'd be nice: an increase of 84 runs over this year, and hasn't been reached by the Diamondbacks since 2002. That team scored 816, but had a OPS+ of 93, so our offense would still be below league average, after adjusting for park factors. However, the projections here may actually be a little low, as in most cases, our starters are not on the wrong side of the aging curve, so some improvement may be expected.

I think we'd probably be happy to settle for five runs per game, and see how far that and our pitching staff might take us. Mind you, after the disaster which was 2009, I am inclined to believe that when I see it. That would be missing the main point, which is that, even with almost the same players in the line-up, the optimum order for placing them appears to be significantly different, depending on the handedness of the starter. But have at it: what do you think should be our 2010 line-up? And does it differ based on who we're facing?