Last week I posted a quick preview of an analysis of the Texas Rangers lineup. I used a Markov chain program I wrote to simulate millions of games very quickly. A simple overview would be imagining all the various outcomes of an event, and rolling a die to determine the outcome.
Baseball Musings has a often used Lineup Analyzer, however it does not incorporate any elements of base running. Base advancement is always a single base at a time, giving no value to base running ability.
The program I wrote also incorprates how often a player attempts to steal a base, their success rate of stealing a base, getting picked off, going from 1st to 3rd on a single, scoring from 1st on a double and scoring from 2nd on a single. There are also out values for their unsuccessful attempts to advance.
Some of the limitations of the program are:
- Stolen Base opportunities and attempts are only from 1st to 2nd. 3rd and Home are never stolen.
- There is no advancement on an out, only a hit (with the exception of a double play).
- There are no sacrifice bunts or sacrifice flys.
- There is no fielder's choice.
- There is no advancement on balks, wild pitches or passed balls.
So, the simulation is not perfect, but I believe to be a more reasonable approximation of a player's impact in the batting order than no base running at all.
Last week, I selected 25 different lineups using 2010 data and compared their average runs scored after 1,000,000 simulations. Since then I have run full sets of simulations at 10,000 games per lineup for all 362,880 lineup permutations using both 2010 data and Bill James' 2011 projections for hitting. I used 2010 base running values for both simulations, with the exception of Mitch Moreland due to his small sample size.
After running all possible lineups, I took the top and bottom 0.1% of each result and then determined how often a player appeared in a particular position within the top and bottom lineups.
This highlights the most common position of a player and visually makes it easy to recognize the lineups with the best and worst results. Interestingly, Mitch Moreland holds strong with his top of the lineup position with both his 2010 data with limited (but not tiny) plate appearances and his 2011 projection by Bill James. His ability to get on base, hit, hit with a fair amount of power and average base running makes him a prime candidate to hit in front of Hamilton / Cruz / Beltre power producers.
Another interesting view is the runs added or subtracted within all lineups by a player being in a particular position. The following charts show the run variance for a player for all lineups that have that player in a respective position compared to the average of all 362,880 lineups. For example, Ian Kinsler's +0.16% for the 2010 data shows that for every possible lineup that featured Kins batting leadoff scored 0.16% higher than the 4.738 average runs for all possible lineups.
The thing you immediately notice is that the absolute worst thing you can do is bat Josh Hamilton ninth. I'll file that in the "water is wet" category. Additionally, an interesting point is that using the 2011 Bill James projections, it doesn't really matter a tremendous amount where you bat Adrian Beltre. His particular combination of low walk rate, decent power and double plays offset each other when you bring down his power numbers compared to 2010.
I think there are three main questions the Rangers face when trying to determine a lineup order using these nine players (the RHP lineup).
1) Who should bat leadoff between Ian Kinsler and Elvis Andrus?
I think the answer is extremely obvious if you believe the offensive output to resemble anything like their 2010 or 2011 projections. Elvis has improved his walk rate and has great speed, but he also makes more outs on the base paths than Ian. Additionaly, Ian hits better at the plate almost every category. Unless Elvis makes a significant jump in 2011, the lineup is likely to be more productive with Ian at the top and Elvis at or near the bottom.
2) Who should bat cleanup between Nelson Cruz and Adrian Beltre?
This one is not as clear cut. From a pure production stand point, Nellie has a greater impact in the 4th position than Adrian... but both are middle of the lineup bats. And this doesn't consider the pressure or protetection someone else may offer. The fear of facing Cruz may provide better pitches for Beltre to hit, but ultimately I don't expect that it makes a tremendous difference between the two unless one struggles significantly.
3) Should Michael Young still be batting 2nd for the Texas Rangers?
Well... from a purely statistical view point, probably not. However that's predicated on a very young player in Mitch Moreland performing the way we hope he can and similar to what he showed us at the end of last year. I think there's a very promising future for Mitch based on his approach and tenacity at the plate and his success as a minor league player. But I wouldn't start 2011 with such a young player in a key position like that. Additionally, Young could experience a bounce back year from a disappointing 2010. And while Young can still hit, much like Beltre... his walk rate, double plays and limited power makes him a better candidate for the back of the lineup rather than the front.
I think most of us felt these were the answers already, but it's nice to see unbiased visual information supporting it.
I only used Bill James data because ZiPs hasn't released Red Sox projections for Beltre, 2011 PECOTA isn't out yet and CAIRO is still a little premature.