PLATOON: The 2011 Diamondbacks, Handedness, and last Wednesday's Lineup

Note: If numbers make your eyes glaze over, feel free to skip down to the conclusions.

I like the idea of platooning Parra

According to baseball reference, Parra’s career OPS+ against RHP is 110, against LHP is 62. As said above, he would still get 75% of the starts. Willy Mo, or Cowgill, get the other starts. If it’s Willy Mo, you can always put Parra in as a defensive replacement late in the game.

It seems that platooning has gone out of fashion in the big leagues. I grew up a Royals fan in Kansas, and they platooned players all the time. They took a lot of mediocre ballplayers and turned them into a decent L/R platoon team.

by Craig from Az on Jun 15, 2011 9:23 AM MST

"Platooning" generally involves switching certain players in or out of the lineup based on a particular split -- usually involving benching left-handed players against a lefty starter, or righty players against RHPs. That's what we're going to be discussing here.

Take the June 8th lineup against the Pirates. We were facing lefty Paul Maholm, so Kirk Gibson sat Gerardo Parra for Xavier Nady; sat Stephen Drew for Willie Bloomquist; sat Miguel Montero for Henry Blanco; sat Kelly Johnson for Ryan Roberts; and played Melvin Mora at 3B. Yet, strangely, he kept Juan Miranda, a lefty, in at first base. The total lineup went:

1. Willie Bloomquist, SS
2. Melvin Mora, 3B
3. Justin Upton, RF
4. Chris Young, CF
5. Juan Miranda, 1B
6. Xavier Nady, LF
7. Henry Blanco, C
8. Ryan Roberts, 2B
9. Zach Duke, P

Think that lineup caused angst among some Dbacks fans? Go back to IHateSouthBend's recap to which I linked above.

Nady, Bloomquist, Blanco, and Mora combined for a .435 OPS in the game.

When KJ, Drew, and Parra came into the game, they produced a 1.381 OPS.

KJ, Drew, Miggy, and Parra's overall OPS' this season are .721, .818, .852, and .741 respectively. Their replacements Mora, Bloomquist, Blanco, and Nady are at .522, .611, .847, and .705.

KJ (.721 on season, .565 vs. lefties) was replaced by Mora (.522 season, .650 lefties).
Drew (.818 on season, .737 vs. lefties) was replaced by Bloomquist (.611 on season, .639 lefties).
Miggy (.852 on season, .705 vs. LHP) was replaced by Blanco (.847 season, 1.282 LHP in 24 ABs).
Parra (.741, .631 LHP) was replaced by Nady (.705, .661 LHP).

Using just those numbers, and weighting each player's splits based on the handedness of each actual PA seen, did it make sense?

C (Blanco) saw 2 LHP PAs, 3 RHP PAs.
LF (Nady-Parra) saw 2 LHP PAs, 4 RHP PAs.
SS (Bloomquist-Drew) saw 3 LHP PAs, 4 RHP PAs.
2B-3B (Mora-KJ, since Ryan Roberts plays the other position) saw 3 LHP PAs, 3 RHP PAs.

Remember, of course, that some of these are very small samples, and I'm only using this season's numbers.

Weighing each player's split OPS for actual plate appearances vs. pitchers seen, we get:

C
Blanco: (.4*1.282)+(.6*.296) = .690 OPS
Miggy: (.4*.705)+(.6*.894) = .818 OPS

2B-3B
Mora: (.5*.650)+(.5*.468) = .559 OPS
KJ: (.5*.565)+(.5*.763) = .664 OPS

SS
Bloomquist: (.429*.611)+(.571*.595) = .602 OPS
Drew: (.429*.737)+(.571*.848) = .800 OPS

LF
Nady: (.333*.661)+(.667*.743) = .716 OPS
Parra: (.333*.631)+(.667*.775) = .727 OPS

So, the only platoon that even came close to making sense was putting Nady's bat in for Parra. (Ignoring Parra's defensive advantage)

But wait! The test isn't fair, you say, because it was an extra-innings game, and the Diamondbacks had a higher % of plate appearances against righties than they would in a regulation game. To Kirk Gibson's credit, he put Drew, Johnson, and Parra in late. What if we re-weigh each player's split OPS, only including the plate appearances they had in regulation? Assume that the Dbacks don't tie it? (And they wouldn't have, without Drew and KJ.)

C saw a lefty twice and righty twice.
LF saw the same.
2B-3B saw a lefty thrice and righty once.
SS saw a lefty thrice and righty twice.

Recalculating, using the LHP/RHP ratio seen in regulation, we get:

C
Blanco: (.5*1.282)+(.5*.296) = .789 OPS
Miggy: (.5*.705)+(.5*.894) = .796 OPS

2B-3B
Mora: (.75*.650)+(.25*.468) = .605 OPS
KJ: (.75*.565)+(.25*.763) = .615 OPS

SS
Bloomquist: (.6*.611)+(.4*.595) = .605 OPS
Drew: (.6*.737)+(.4*.848) = .781 OPS

LF
Nady: (.5*.661)+(.5*.743) = .702 OPS
Parra: (.5*.631)+(.5*.775) = .703 OPS

What do these numbers tell us? They tell us that, if you assume that each player's platoon split thus far this season represents their true talent, then yes, the platoons at C, LF, and 2B/3B were actually defensible individually. None of those platoons made that position a BETTER hitting spot, but it was extremely close at each position.

• One final thing these numbers tell us? Willie Bloomquist should NEVER, EVER, EVER replace a healthy Stephen Drew EVER. (We will come back to this, so don't forget that I said it.)
  

Problems with this analysis? Absolutely. Do you REALLY think Henry Blanco will continue to hammer lefties (1.282 OPS) and run away crying from righties (.296 OPS)? Probably not. That lefty split should drop significantly, and the righty split should rise by quite a bit.

Now, what about going forward? Trying to pick out the best platooned lineup in the future? Well, we can try to do some predictions. These numbers are only through 60+ games and still reflect a fairly small sample size. Additionally, while OPS weighs on base % and slugging % equally, research into team scoring has shown that they are NOT equally valuable in predicting a team's ability to score runs, nor do they reflect the differences across ballparks. Let's use something more advanced.

For the predictive part of this story, I switched from OPS to wRC+, a hitting metric from FanGraphs that values on base % more than slugging, compensates for leagues and ballparks, and uses a scale with 100 being a major league average hitter -- that is, above 100 is above average, below 100 is below average, etc.

I also made a pretty large assumption, trying to regress each player's platoon splits, by averaging their wRC+ vs. RHP or LHP this year with their career numbers. For Blanco, since his splits are ridiculous, I used his straight career numbers, and for Sean Burroughs (yes, I took EVERY hitter on the roster) I used his career split against LHPs, since he only has two PAs vs. lefties this year. (0-2, in case you were wondering)

The numbers still aren't perfect, of course. 39 year old dudes like Melvin Mora are unlikely to hit at their career averages, and much of Xavier Nady's data comes from before he had two Tommy John surgeries. Simply using the mean of Kelly Johnson's splits ends up ignoring the fact that, although he has a regular platoon split this season, his career numbers still show a reverse split -- Kelly traditionally hits lefties better. Juan Miranda is the same way. In Kelly's case, like Burroughs, I used his career split against lefties, because I felt that it was a more reliable predictor of the future than his crappy 2011 numbers. (That number is 117, with a 47 this season. Averaging those would give an 82, and averaging his career number twice as heavily would give a 94. But I digest)

That having been said, I took these revised numbers and weighed them for likely plate appearances against different pitchers. The Pittsburgh game had nobody accrue more than 5 plate appearances during regulation, with most only getting 4, but that was a 2-2 pitcher's duel after the end of 9. In weighing each player's platoon split wRC+, I used 6 plate appearances and several different LHP/RHP ratios. Below are the players ranked in descending order of expected wRC+ in that game, and the dudes who should be starters.

6 RHPA (righty starter, all righty relievers):

1. Miggy, C, 123
2. Drew, SS, 117
3. Upton, RF, 117
4. Miranda, 1B, 106
5. KJ, 2B, 106
6. RyRo, 3B, 101
7. Nady, LF, 97
8. CY, CF, 95
9. Parra, 93
10. Burroughs, 85
11. Mora, 64
12. Willie B, 59
13. Blanco, 51

Obviously, this is bat-only, and ignores Parra's awesome defensive advantage over Nady. In reality, you'd probably switch them and have Nady be the first bat off the bench.

5 RHPA with a LOOGY (These numbers represent the dudes who are better vs. lefties facing a righty 6x, and the dudes who are better vs. righties facing a righty 5x and a lefty 1x. That assumes, of course, that opposing managers have access to the data and use it accordingly -- not using a LOOGY vs. Miranda or KJ, for example. Nady has a reverse split this year, even after averaging, so he's treated like a lefty):

1. Miggy, C, 118
2. Upton, RF, 117
3. Drew, SS, 112
4. Miranda, 1B, 106
5. KJ, 2B, 106
6. RyRo, 3B, 106
7. Nady, LF, 96
8. CY, CF, 95
9. Parra, 88
10. Burroughs, 82
11. Mora, 64
12. Willie B, 59
13. Blanco, 51
    

The only change to the starters is a switch of Drew and Upton.

This next scenario is unlikely -- it reflects 4 PAs vs. a righty, and 2 PAs vs. a lefty for everyone on the roster. Either the LHP starter got pulled early, or a team is especially lefty-heavy in its bullpen. Or else it's against the Cardinals, because Tony LaRussa is, of course, a genius. ;-)

1. Upton, RF, 129
2. Miranda, 1B, 123
3. Miggy, C, 112
4. RyRo, 3B, 110
5. KJ, 2B, 110
6. CY, CF, 107
7. Drew, SS, 106
8. Nady, LF, 94
9. Parra, 82
10. Burroughs, 79
11. Mora, 72
12. Willie B, 68
13. Blanco, 65

Same group of starters, but different order. Miggy is no longer our best hitter, and even facing a lefty twice moves Drew well down the list.

3LHPA, 3RHPA: I think this is the most likely scenario when a lefty starts. Either this or the next one.

1. Upton, RF, 135
2. Miranda, 1B, 132
3. RyRo, 3B, 115
4. CY, CF, 113
5. KJ, 2B, 112
   
6. Miggy, C, 107
7. Drew, SS, 101
8. Nady, LF, 93
9. Parra, 77
10. Mora, 77
11. Burroughs, 76
12. Willie B, 73
13. Blanco, 73

Starters are still the same, but the right-handed cream is really rising to the top.

4LHPAs, 2RHPAs: Lefty starter, righty relievers.

1. Upton, RF, 140
2. Miranda, 1B, 140
3. CY, CF, 119
4. RyRo, 3B, 119
5. KJ, 2B, 113
6. Miggy, C, 102
7. Drew, SS, 96
8. Nady, LF, 92
9. Mora, 81
10. Blanco, 80
11. Willie B, 77
12. Burroughs, 73
13. Parra, 72

Wow, does Juan Miranda have a helluva reverse split. Moving right along:

5LHPAs, 1RHPA: Lefty starter, maybe 1 lefty reliever, righty setup guy and closer. Or something.

1. Miranda, 1B, 148
2. Upton, RF, 146
3. CY, CF, 125
4. RyRo, 3B, 123
5. KJ, 2B, 115
6. Miggy, C, 96
7. Drew, SS, 90
8. Nady, LF, 90
9. Blanco, 87
10. Mora, 85
11. Willie B, 81
12. Burroughs, 70
13. Parra, 66

Miranda becomes the top hitter using this process. I doubt he's likely to outhit Upton vs. lefties moving forward, but those are the numbers. Finally:

6LHPA: This sinister lefty starter pitches complete games every time. Must be Eddie Plank, HOFer who holds the career record for complete games by a lefty, (410) while pitching for the Philadelphia Athletics and St. Louis Browns in the AL and St. Louis Terriers in the Federal League.

1. Miranda, 1B, 157
2. Upton, RF, 152
3. CY, CF, 131
4. RyRo, 3B, 128
5. KJ, 2B, 117
6. Blanco, C, 94
7. Miggy, 91
8. Nady, LF, 89
9. Mora, 89
10. Willie B, SS, 86
11. Drew, 85
12. Burroughs, 67
    
13. Parra, 61

NOW we see some real movement. Six plate appearances against a lefty, none against a righty. These are, essentially, just each player's regressed platoon split -- just to reiterate, for most players, these numbers represent [(2011 wRC+ vs. LHP)+(Career wRC+ vs. LHP)]/2, except in the cases of Burroughs, Blanco, and KJ, as noted above.

IF you think these provide a decent way to predict each player's future platoon split performance, and that's a HUGE "IF", and you're trying to come up with a bat-only lineup, these are the guys you want starting. When facing ONLY lefties, benching Drew for Willie Bloomquist at SS, benching Parra for Xavier Nady (even with the defensive advantage, Parra is traditionally terrible against lefties), and benching Miguel Montero for Henry Blanco actually makes sense. What about Melvin Mora? Well, remember how we calculated KJ's regressed split differently? Mora's numbers are buoyed by having had a pretty good career, but if we followed the same rule for Kelly Johnson that we did for most other starters, (using a straight average of this season and career, rather than just career) KJ has an 82 wRC+ against 6 LHPAs, and an 86 wRC+ against 5 LHPAs and 1 RHPAs. That still puts him barely above Mora (85) when he faces a righty once, but if all PAs are against lefties, RyRo (128) moves over to 2B, and Mora (89) starts at 3B, leaving KJ's 82 on the bench. In case you're wondering, that gives us a lineup of:

1. Miranda, 1B, 157
2. Upton, RF, 152
3. CY, CF, 131
4. RyRo, 2B, 128
5. Blanco, C, 94
6. Miggy, 91
7. Nady, LF, 89
8. Mora, 3B 89
9. Willie B, SS, 86
10. Drew, 85
    
11. KJ, 82
    
12. Burroughs, 67
    
13. Parra, 61

Note, also, that Willie B, in this case, would be the better-hitting 2B. Another way of looking at this lineup is that this should be our starting lineup against a lefty starter, with Miggy, KJ, and Drew first off the bench to pinch hit.

Look familiar? That's Kirk Gibson's exact lineup on June 8th against the Pirates. Gibson even followed my advice and used KJ and Drew to pinch hit in the 8th, after Maholm left. They then replaced Mora and Bloomquist in the field, with Roberts moving over from 2B to 3B. Parra came in later and played LF, and Nady moved to 1B, upgrading both positions defensively. While Gibson didn't use Miggy to pinch hit, it's understandable why he didn't: he's the only pure catcher left on the bench, and Blanco is old and rickety.

So, now what?

This whole thing was a little funky, and the results are quite arbitrary and contrived. There are plenty of issues with it, and it should only be considered an interesting thought experiment. I like to think that, if anything, this proves that there might be method to Kirk Gibson's madness, and that he wasn't just going "by the book" with his lineup against Maholm. Let it be noted that I honestly wasn't expecting anything CLOSE to these results before I started, and I'm just as surprised as you. Maybe even moreso.

Problems? Plenty. I think my "regressed splits" gave too much weight to career numbers for some people, (Mora, Burroughs, Parra, Drew, etc) and too much weight to small samples this season for others (Roberts, Miranda, etc). One could even make the argument that both are true in the case of Nady, with his weird reverse split this season being weighed equally with his pre-injury numbers. Kelly Johnson's numbers required a tough judgment call, too, and anyone might choose to do things differently. Also, I took players' overall splits, rather than using specific splits vs. RH starters, LH relievers, etc, and I didn't really touch pinch hitting, which is, of course, an important aspect of NL baseball. Also, again, this is ONLY going by hitting -- I don't even pretend to touch baserunning or fielding.

Are there any hard and fast conclusions we can draw? First, the obvious ones:

• Justin Upton should NEVER SIT IN ANY GAME EVER.
• Sean Burroughs should NEVER START ANY GAME EVER.
• Juan Miranda should be our everyday starter at 1B, until he proves otherwise.
• I spent WAY TOO MUCH TIME on this.

Now, some conclusions that maybe aren't so obvious?

• Miguel Montero may be our best hitter against right-handed pitching.
  
• Some of our lefty hitters really DO have some drastic platoon splits in their career.
• This is especially true of Parra and Drew.
• Xavier Nady probably isn't as bad an option in LF as we've all assumed, at least with the bat.
• Platooning Miggy with "Hector" Blanco against lefties isn't the worst thing in the world.

Finally, and jinnah should be pleased to know this,

• Willie Bloomquist could MAYBE start at SS instead of a healthy Stephen Drew. Occasionally. Maybe. ;-)

Thoughts? Complaints? Not too crazy about how I went about this? Please include them in your comments.