The Optimal Blue Jays Lineups

As much as we’d all like to imagine that Spring Training performances are the least bit meaningful, the truth is that 35 at bats against inconsistent pitching should carry about as much weight in lineup formation as heart, grit and other intangible qualities.  Managers should already have a general idea of their teams’ batting order during the first week of exhibition games based on past performance beyond the small sample size that Spring Training offers.

For years, the thinking has been that a good lineup is put together like this: a speedster who won’t clog the bases leads off, a player who rarely swings and misses bats second, the team’s best hitter fills the third spot, a power hitter goes in the cleanup spot, someone who can knock in runs hits fifth, and the rest of the lineup is filled with the best hitters that the team has placed in descending order.

But as Jason Hanselman points out in his post on regressed platoon splits at Dock Of The Rays, “conventional wisdom [is] often cliché and muttered without thought.”  Hanselman goes on to mention sabermetrician Tom Tango who suggests that the actual differences between a team’s worst possible lineup order and the best possible isn’t really that great.  But for a team in the American League East that will have to eke out as many runs as possible in order to compete against the Red Sox, Yankees and Rays, the Blue Jays should be looking for an optimal lineup each and every time out.

But what does that mean?  Constructing an optimum lineup isn’t as simple as merely plugging in the players with the best possible platoon splits depending on the handedness of the opposing starting pitcher.  As Matt Klassen points out in his excellent post at FanGraphs on estimating hitter platoon skill,

One bad habit many of us might get into is looking at the platoon splits of two players at the same position, one with a career wOBA of .390 vs. RHP, the other with a career wOBA of .400 vs. LHP, and thinking, “Wow, that platoon would be almost as good as Ryan Braun.” It isn’t that simple. As in most other things, regression shows us that the distance from average is closer than it appears.

Klassen goes on to use more of Tango’s research to come up with formulas for projecting platoon splits that take into account the distinction between observed performance and true talent.  I spent some time inputting the data from the careers of potential Blue Jays lineup candidates as well as some Marcel projections into the formulas he developed to find the following platoon projections:

Player Pos OBP v. LH SLG v. LH wOBA v. LH OBP v. RH SLG v. RH wOBA v. RH
Aaron Hill 2B 0.316 0.449 0.328 0.301 0.418 0.317
Adam Lind 1B 0.293 0.409 0.301 0.335 0.483 0.354
C. Patterson OF 0.269 0.348 0.273 0.297 0.383 0.303
E.Encarnacion DH 0.330 0.470 0.341 0.318 0.435 0.332
J.P. Arencibia C 0.335 0.457 0.341 0.312 0.405 0.318
Jose Bautista 3B 0.359 0.508 0.370 0.341 0.477 0.359
Juan Rivera RF 0.323 0.438 0.325 0.305 0.420 0.318
Rajai Davis CF 0.336 0.406 0.331 0.317 0.374 0.319
S. Podsednik OF 0.319 0.357 0.300 0.338 0.389 0.324
Travis Snider LF 0.311 0.418 0.314 0.333 0.450 0.344
Y. Escobar SS 0.367 0.407 0.339 0.343 0.366 0.321

Most of the data entered was found at FanGraphs.  However, for J.P. Arencibia, I used his Minor League numbers to calculate OBP and SLG, and the MLB equivalent numbers for his wOBA calculations.

As Hanselman recommends in his post, I then took these projections and placed them in Baseball Musings’ Lineup Analysis program which measures the projected run value per game of different lineup options.  According to this program, these are the optimal lineups:

Versus LHP:

Runs per Game 1 2 3 4 5 6 7 8 9
5.011 Bautista Encarnacion Davis Hill Arencibia Snider Rivera Lind Escobar

Versus RHP:

Runs per Game 1 2 3 4 5 6 7 8 9
4.829 Bautista Snider Podsednik Lind Encarnacion Hill Arencibia Davis Escobar

The idea of batting Bautista first in any lineup seems ridiculous considering the power that we’ve seen him utilize in the past.  However, as the player most likely not to get out, there’s something to be said for having him see the most at bats on your team, especially when you consider the OBP of the rest of the Blue Jays last season.  Still, these are the most optimal lineups to use when Bautista is placed in the more traditional third or fourth spot:

Versus LHP:

Runs per Game 1 2 3 4 5 6 7 8 9
4.997 Escobar Hill Bautista Arencibia Encarnacion Lind Snider Rivera Davis

Versus RHP:

Runs per Game 1 2 3 4 5 6 7 8 9
4.820 Snider Lind Hill Bautista Encarnacion Podsednik Davis Arencibia Escobar

Snider leading off?  Maybe Cito Gaston was on to something last season.

Another way of looking at lineup construction is to consider that the first, second and fourth spot are the most important because they’re the most likely hitters to come up to bat with less than two outs, and therefore with the highest run expectancy.  Looking at the numbers, using my own judgment, and leaning more heavily on recent numbers (or lack thereof in Arencibia’s case) instead of an entire career, I would construct the following lineups:

Versus LHP:

Runs per Game 1 2 3 4 5 6 7 8 9
4.920 Escobar Davis Encarnacion Bautista Hill Snider Lind Arencibia Rivera

Versus RHP:

Runs per Game 1 2 3 4 5 6 7 8 9
4.766 Podsednik Snider Lind Bautista Encarnacion Hill Escobar Davis Arencibia

Comments (37)

  1. Do you think theres much value, or any at all, for a player to have the consistency of a set spot in the lineup nearly every game?? Like always hitting cleanup, or always being leadoff. I doubt it, and doubt you do – but you hear people in baseball say things like that. You also hear of players who get angry when they lose their spot in the order. Can’t see any guys on this team being the type to complain though.

  2. I think it would be really interesting if someone wrote a computer program that would manage a major league team using advanced metrics.

  3. Fantastic post, dude. I really like your conclusions, and I think we can obviously expected Lind/JBau going 3/4 vs RHP. And as much as E5 being being around the middle of the line-up scares me, it’s realistic and might be rewarding.

    It sure is strange, especially after so many years of the Delgados, Wellses, Rioses… hell, Cansecos and Thomases, to look at line-up full of Escobars, Davises, and Podsedniks… regardless, there’s a lot of interesting and exciting options for scoring runs, this year, which should be a lot of fun. It also should come in handy when 1) We’re not putting together dingers as per usual 2) We’re struggling to get guys on base, and can find ways to get them across instead of relying on the magical solo homer of yesteryear.

  4. You’re projections for Arencibia are awfully optimistic. Projecting the second highest wOBA against LHP on the team for a player with little major league experience and not much success over that stint seems shaky. I believe fangraphs publishes one set of projections that takes into account minor league performance but regresses it to more realistic levels than the method you used.

  5. It’s unfortunate that prestige is placed on certain spots in the order. There’s no reasonable explanation for a hitter being more comfortable in one spot over the other, and we’ve been over it before how lineup protection is a myth, so the only thing I like about consistency if you use an optimum lineup is that you’re likely going by larger sample sizes than the last seven games, which in my opinion is better.

  6. I used the Marcel projection for Arencibia. I didn’t want to mix and match.

  7. Also, “regressing” Arencibia actually increases his projection against LHP.

  8. Did you get this idea from roto hardball?

    I have never understood batting hill anywhere in the top 5 with his career walk percentage of 6.7%. I also think putting davis in the top 3 is a mistake since he walks even less than Hill. One interesting thing i think you uncovered is the possibilty of Encarnacion hitting high up in the order as he ranked second in wOBA on the team last year albeit in only 96 games. I feel many fans underate his bat and only focus on his poor defense at third.

    • Quite honestly, I don’t think his defense is that bad at all, just his godawful throwing.

      I hadn’t seen that link. The first link in the post is the one that gave me the idea, as I basically copied what he did for the Rays.

  9. It should be “eke out,” yes? Although if Snider isn’t what he think he is and Lind and Hill don’t bounce back, “eek” indeed. What these numbers don’t seem to take into account (correct me if I’m wrong) is how guys hit with men on base, which is part of the rationale for the traditional lineup I’d guess. Some guys press, some guys keep the line moving and get on base, and some guys hit better when a run can score. I’d have to look it up, but E3 (Mr. Solo Homer) in the top four of any lineup seems nuts to me on that basis.

  10. That John Farrell sure is lantern-jawed. I wish I was lantern-jawed.

  11. I wouldn’t put the only 2 lefties in the lineup together.

  12. Eke indeed.

    Hitting with RISP is like “clutch” hitting. It varies widely from year to year, and is completely dependent on luck. Guys with large enough sample sizes will almost always hit the same whether runners are on or off the bases.

  13. @Kevin – Parkes pretty much nails it. Good hitters are good with runners on, bad hitters aren’t. Science!

  14. It seems that the one thing that may affect a player’s performance based on their spot in the batting order is the type of pitches they see, which is arguably dependent both upon who follows them and who is on base – it probably matters most where the splits for the current batter and the next batter are significantly different. I have no idea if this is a large effect across the board (the best examples would be those where a change in “protection” in the order has a significant effect for a given hitter, though most players see marginally different levels of protection), but it is possible that you could take any one of those players, change their spot in the order or who bats around them, and the wOBA changes. If it makes a difference of +-5 points or more, then the whole calculation changes.

  15. Does an individual players stats change depending on where in the order they hit? If yes, then projecting who would be best in a certain spot in the batting order won’t work because the very action of moving them to a different spot in the order alters the very thing that caused to put in that spot in the first place. If you had stats for each player for each spot in the order then you could accurately make a lineup projection assuming that you had a large enough sample size which of course you wouldn’t because players to bat in every spot of the order.

    It certainly is an interesting take on it. I am all for trying a different approach and assuming that the changing of their spot in the order does not change the way they hit then I think it would be good. I especially like the lineups you constructed and the end of the post. Call Farrell and tell him to go with that.

    • Normally, a batter doesn’t have large enough sample sizes at all nine positions in the order, but because evidence suggests that a pitcher’s approach doesn’t change no matter who is batting before or after you, it shouldn’t matter.

  16. Regarding protection, I agree that generally the stats don’t work with it, but is it unreasonable to speculate that Bautista’s numbers last year had something to do with Vernon Wells batting after him? Bautista was only intentionally walked twice. Then again, he was hit by pitches 10 times, though that could easily be attributed to how he crowds the plate. It’s too bad we can’t rerun last season and put someone like Aaron Hill in the number 4 slot to see if Jose would have been pitched to differently.

    • I think that speaks more to the fact that Bautista went through a large part of the season without seeming like a threat. This was probably the quietest 50 home run season in history. If teams were avoiding walking Bautista out of fear of Vernon Wells, they clearly aren’t aware of how lacking Wells is, let alone Bautista’s brilliance.

  17. Sorry, I’ll just add that in 2006, when David Ortiz hit 54 home runs, he was intentionally walked 23 times. Are there any ideas (or ways to find out) what Boston’s batting order looked like in 2006?

  18. Batting Bautista third and Davis/Escobar 8-9 is EXACTLY the same lineup arrangement as the one we’re likely to see (1-Davis, 2-Escobar, 3-Bautista) with the exception of the fact that Bautista will get about 36 extra PAs. So it makes perfect sense that any lineup simulator will like it.

  19. Also, Ortiz hit in front of Manny, probably the best hitter in baseball, for 128 games in 2006.

  20. ugh. Batting Bautista FIRST and Davis/Escobar 8-9

  21. In econometrics (statistics applied to economics) generally if your model produces results contrary to conventional wisdom, you need to come up with an explanation of either why this conventional wisdom is incorrect in this instance, or how does this result fit with conventional wisdom. And then you need to translate it to layman-speak: we call this the economic intuition of what is going on.

    Your first two lineups appear to produce weird results: power hitters in spots 1 and 2 (Bautista, E5 and Snider) and slap and run guys in spot 3 (Davis and Pods). Bautista seems obvious to me: his high OBP is most valuable up front (reinforced by the fact that when he is restricted to #3, Escobar moves to #1 against LHP)and his probability of getting into scoring position through an extrabase hit is very high. However, I am having a hard time coming up with a justification to bat Davis and Pods third. Any ideas? The best of mine is that they are fast so they can beat out the double play more often.

  22. The third spot in the order is one of the most likely to get up to bat with two outs, meaning that run expectancy is at a minimum. It’s a waste to bat a power guy in that spot, but #3 is still going to get up to bat a lot so you want someone who isn’t terrible at getting on base.

  23. @JRock – Pods and Davis have the highest respectives OBPs per-handedness

  24. There you guys go, bringing FACTs into the argument again. I’m reluctantly converting to the numbers game, though I do think it ignores the team side of baseball, which metrics suggest is overrated, or perhaps entirely subjective. Does Johnny Mac always make one of those head-shaking defesive plays with the game on the line, or is that just the play we notice?

    There are the flukes like Pat Tabler — decent hitter most of his career, lights out with the bases loaded. Of course, he’s also the guy who still can’t figure out there’s no “t” in “across.”

    Best argument for Bautista hitting first (given the good OBP) might simply be more plate appearances per week.

  25. As far as your final optimal lineups – I don’t mind Encarnacion at 5 or 6, but I have a hard time believing that the situational difference between third and fourth could ever justify giving EE more PAs than Bautista.

  26. @Parkes. Thanks, that makes complete sense.

  27. Lineups may not make a huge difference, however, all iof the optimal lineups Parkes has created are above the 4.66 runs a game average the Jays had last season.

  28. Don’t you still have to split your lefties/righties? Otherwise your “optimal lineup” goes out the window as soon as the starter is gone…

    • There’s definitely an argument to be made for that, but I still think that maximizing your platoon split lineup against a starter is more important than worrying about one inning of potential unfavourable match ups.

  29. I wondered about the protection thing, but had not read those. Thanks.
    I honestly have no hang-ups with doing something unconventional if there is statistical evidence that it should work better. Bautista/Snider as 1/2 would be kind of fun to watch.

  30. Parkes, do the numbers show any benefit to batting behind a guy with a high OBP and good speed? I ask because in this case the hitter would face a lot more pitches from pitchers pitching out of the stretch.

    • That’s interesting. I haven’t looked at that specifically, but generally speaking, there’s no evidence to suggest that hitters do better or worse no matter who is batting ahead or behind them. Beyond the Boxscore has put up some interesting data on how the umpire unconsciously changes the strike zone in different run expectancy situations which you’d think would have a bigger impact on lineup protection than it seemingly does.

Leave a Reply

Your email address will not be published. Required fields are marked *