Among the persistent themes of Dusty Baker's tenure with the Reds has been his rather inflexible ideas about lineup construction. The CF has to lead off, the SS has to hit second, etc. And as he has said many times, what he uses is the best lineup he can come up with.
Think you could do better? Here's your chance! We're going to have a Lineup Smackdown competition over the coming week. What I'd like you to do is to submit what YOU think would be the Reds' best lineup(s) for opening day. We'll then compare how your lineups do against Baker's in the best publicly available tool I've seen to evaluate lineup construction: John Beamer's Markov chains model that was released with the Hardball Times 2008 Annual.
Here are the details. In the comments, please submit lineups in three categories:
Your "best" lineup using Baker's presumed Starting Nine vs. RHP
We will use C Hernandez, 1B Votto, 2B Phillips, 3B Rolen, SS Cabrera, LF Dickerson, CF Stubbs, RF Bruce, P Harang
Your "best" lineup using Baker's presumed Starting Nine vs. LHP
We will use C Hernandez, 1B Votto, 2B Phillips, 3B Rolen, SS Cabrera, LF Gomes, CF Stubbs, RF Bruce, P Harang
Your "best" lineup using your choice of Reds players: You choose whoever you'd like within the Reds organization to play your positions. Please be realistic about position, though (i.e. don't stick Hanigan at SS just to get his bat in the lineup). And let's keep using Harang as P in these--yes, I'm sure the lineup would be better with Owings hitting 9th.
Some other details:
- If you want to submit a couple of lineups in each category, that's fine. Let's say no more than 6 total lineups per person? I have to run these by hand. :)
- We'll use CHONE projections for all players, so please use them as the basis for your decisions.
- The Markov does account for baserunning, at least in terms of how often and how well a runner will steal bases when he's on base. Therefore, speed does matter.
- The Markov does not account for batter handedness or splits. Feel free to account for this in your lineups, but realize that the model doesn't know about them. It assumes an average pitcher, which I guess is ~3/4 right-handed and ~1/4 left-handed.
Update: regarding the above point, I will probably run lineups instituting a generic split for RHB and LHB. I have to test some things to make sure it seems like a valid approach, but this should enhance the realism of a vs. LHB or vs. RHB lineup. I will try to run them with and without these splits, since this is a new development and some folks submitted lineups before I decided to do this.
- I will include Baker's actual opening day lineup in the pool of lineups. Might use a few others from spring training too, just for kicks.
- Winners will be determined based on runs per game according to the model. This tool can be thought of as a simulation that will let us test a variety of lineups under fairly realistic conditions. It has advantages over other more simplified lineup tools, like David Pinto's, in that it can explicitly account for interactions between players within a lineup. The player that you put in the leadoff slot, for example, effects the opportunities that the guys hitting 2nd, 3rd, and 4th have to drive in runners and thus produce runs. The Markov does consider these kinds of interactions, while regression-based tools like Pinto's do not.
- Readers may remember that I did something like this in 2008 at my old blog...and that it was kind of a catastrophe because it turned out that I was not inputting the data properly (freaking embarrassing!). I know how to do it now, and so this will not be a problem again. The model generally seems to agree to other high quality lineup models, like the simulator used in The Book by Tango, MGL, and Dolphin.