Platoon Differential

[Phantom]

10 minutes ago, Tony-OH said:

This is silly, because I'm not talking about time, I'm talking about a particular variable. I totally understand SSS or cherry picking stats to make a point, but their is a huge difference to saying saying Jones OPSd .550 on Tuesdays in April, then saying Dan Pasqua OPS'd .192./275/.321/.596 against LHP vs 252/.340/.458/.798 against RHP. 

I would agree that most everyday players don't have much of a split, but you can't take data over years and years and say a platoon difference doesn't occur. It definitely occurs with some players. That is an undeniable fact and I can easily keep posting platoon players who had long careers being just that.

You keep posting about "time" as if a SSS over time is better than a small sample size. And batting on Tuesday over a career is the same "time" as hitting against lefties over a career. You are taking a small sample size, over a career, and applying cause to it. And I am asking if the same would be true if the variable in question were more absurd. Isn't it possible that Adam Jones has trouble sleeping on Monday nights for some reason (maybe because they are off nights) or that he has custody of his kids that day or that he has by chance faces awesome pitching for 150 of the 300 PAs in question on Tuesdays? But you wouldn't immediately assume that, right?

So the question remains, when we are presented with a player who has a SSS of bad same hand platoon, why do we assume skill as opposed to day of the week like fluke? I think you would say the answer is, you can see it, but if we can see it, we can predict it, right? And to my knowledge no one can actually predict same hand platoon better than machines that are assuming randomness.

[Since'54]

2 hours ago, DrungoHazewood said:

Doesn't it really defy preconceived notions that help you define what you consider logical?

Should this be a matter of notions -- preconceived or otherwise?  Its certainly a matter of statistical fact and I am particularly surprised by the assertion that variation doesn't apply for individual batters. And then the immediate citation of an exception.

[Phantom]

21 minutes ago, Frobby said:

I mostly agree with Tony on this, with one caveat.    I'm sure there are some guys who get pigeonholed early in their career based on a small sample of data, who in a larger sample would have "normal" platoon splits.    That's why I think it's important to observe the player and not just look at the numbers.  

Can you give me some modern examples of players that pass or fail the eye test when it comes to same side splits?

[Tony-OH]

17 minutes ago, Phantom said:

You keep posting about "time" as if a SSS over time is better than a small sample size. And batting on Tuesday over a career is the same "time" as hitting against lefties over a career. You are taking a small sample size, over a career, and applying cause to it. And I am asking if the same would be true if the variable in question were more absurd. Isn't it possible that Adam Jones has trouble sleeping on Monday nights for some reason (maybe because they are off nights) or that he has custody of his kids that day or that he has by chance faces awesome pitching for 150 of the 300 PAs in question on Tuesdays? But you wouldn't immediately assume that, right?

So the question remains, when we are presented with a player who has a SSS of bad same hand platoon, why do we assume skill as opposed to day of the week like fluke? I think you would say the answer is, you can see it, but if we can see it, we can predict it, right? And to my knowledge no one can actually predict same hand platoon better than machines that are assuming randomness.

Yeah, we'll have to agree too disagree. 

[Tony-OH]

25 minutes ago, Frobby said:

I mostly agree with Tony on this, with one caveat.    I'm sure there are some guys who get pigeonholed early in their career based on a small sample of data, who in a larger sample would have "normal" platoon splits.    That's why I think it's important to observe the player and not just look at the numbers.  

I think that's mostly because their minor league numbers already had major platoon differences. Now roster composition at the time of their arrival in the big leagues could certainly cause for player to be initially platooned, but I bet there are not too many guys who were pigeon holed as a platoon player without significant results, most likely from the minors. I also think scouting is also used here and although that stat guys will surely pop-poo the human element, guys who just look awful against one side will get that reputation.

Very few players, especially prospects are platooned in the minors because teams are trying to see if they can hit both hands. Once a track record is established that matches the scouting reports, that player will mostly likely have that platoon tag on him.

[Aristotelian]

Seems crazy to me. Might as well just throw out AVG, OBP, and OPS because they are based on only slightly larger (not orders of magnitude larger) samples than individual career platoon splits. 

[CaptainRedbeard]

I will agree that the significance of players who are "lefty mashers" and are utterly hopeless vs RHP are overstated.  I don't think there's such thing as a player having extremely large platoon splits to the point that they're an all star vs one hand but a AAA player vs another.  But to say that there aren't players who justifiably have a smaller platoon split than average than others is completely inane.

Btw, another good example of a "reverse" platoon split:  Ichiro Suzuki. 114 wRC+ vs LHP (2804 PA), 99 vs RHP (6943 PA).  If somebody tries to tell me that Ichiro was significantly worse vs. LHP than RHP my head might explode.

 

[Frobby]

41 minutes ago, Phantom said:

Can you give me some modern examples of players that pass or fail the eye test when it comes to same side splits?

Adam Lind comes to mind.  

[Drifter]

4 hours ago, DrungoHazewood said:

Because they grow up with that understanding, and because that understanding is re-enforced every day by selective observations

Selective observations is key.

Many stats can be tweaked in many ways to prove out a theory..  after awhile, I Iose interest and look at the main stats that I grew up following. I do like a lot of the newer stats. SLG and OPS for hitters, WHIP for pitchers.

Back to the main point.. I'm 50/50 on platooning. I understand the stats side, but I also lament over the fact that a hot hitter sits out a game or 2 because the numbers suggest the next pitcher(s) will win the battle versus him. Momentum is lost.

[Aristotelian]

I would agree with James that if you had a same-side reliever with extreme splits, like .250 OPS differential, I would expect his differential to win out over a hitter with a small to moderate reverse split. And I would agree that it is more common for pitchers to have extreme splits than hitters, because pitchers have such different repertoires while hitting is pretty much hitting. Still, even if I had an Adam Jones or Schoop type of righty, I would not hesitate to bring in an O'Day type of specialist, and vice versa.

 

[BohKnowsBmore]

1 hour ago, Phantom said:

You keep posting about "time" as if a SSS over time is better than a small sample size. And batting on Tuesday over a career is the same "time" as hitting against lefties over a career. You are taking a small sample size, over a career, and applying cause to it. And I am asking if the same would be true if the variable in question were more absurd. Isn't it possible that Adam Jones has trouble sleeping on Monday nights for some reason (maybe because they are off nights) or that he has custody of his kids that day or that he has by chance faces awesome pitching for 150 of the 300 PAs in question on Tuesdays? But you wouldn't immediately assume that, right?

So the question remains, when we are presented with a player who has a SSS of bad same hand platoon, why do we assume skill as opposed to day of the week like fluke? I think you would say the answer is, you can see it, but if we can see it, we can predict it, right? And to my knowledge no one can actually predict same hand platoon better than machines that are assuming randomness.

You're ignoring a key tenet of statistical analysis here, though.  Just because we have a statistical relationship, we don't stop asking ourselves from a logical standpoint whether a causal relationship makes sense.  Handedness and platoon splits certainly do.  Days of the week do not, unless Tuesdays are the only time day games are played, or some other factor that would make there be a compelling case for why we're not just seeing randomness that manifested itself in one particular fashion.

[El Gordo]

2 hours ago, DrungoHazewood said:

Lowenstein had years where he wasn't platooned and had normal splits, like 1974. But he just wasn't very good. So to minimize the impact the Indians started strictly platooning him. The O's continued that even as his performance against right handers took off. It worked, but he might have hit lefties far better as if given more than a dozen PAs a year.

I think I'm understanding this. If Lowenstein Had been given closer to the same 3600 PA v LHP, his splits wouldn't be so extreme ,rather much closer to the average? That platoon splits are to some extent a self fufilling prophesy?

[SteveA]

4 hours ago, NCRaven said:

My only problem is that Bill James did not explain how is understanding of platoon differential can be used to change the way we play the game.  Okay, give James theory, how should managers change the way that they manage the game?  Most managers try to optimize their lineups by playing lefty hitters against righty pitchers as often as possible.  And vice versa.  Should that change?  I guess I'm asking, what's his point?

Well, one implication would be that a guy like Seth Smith who has exhibited a really strong platoon differential should not be treated differently than a guy like Chris Davis who hasn't had as strong a differential.   You should not "hide" Smith from facing lefthanders any more than any other guy.   If you have a righty to platoon with him, I guess go ahead, but nowadays with short benches you can rarely afford to platoon at more than one position.

And if you did that, and Smith went out and matched his atrocious career numbers vs LH... you'd probably be fired.

[schittenden]

I'm just curious.  Would you apply the same logic to LOOGYs?  I mean, after all, if they only had faced more righties, perhaps their differentials wouldn't be so bad.  The logic seems the same.

[SteveA]

7 minutes ago, schittenden said:

I'm just curious.  Would you apply the same logic to LOOGYs?  I mean, after all, if they only had faced more righties, perhaps their differentials wouldn't be so bad.  The logic seems the same.

There seems to be a few "asymmetrical" type things in sabermetrics that has always bothered me a bit.   The one that always jumped out at me the most was how a high emphasis was put on pitchers' stirkeout rates (more = good, obviously) and yet you aren't supposed to care if your hitters strike out a lot.   If more strikeouts is good for a pitcher, it seems to me it should be bad for a hitter.

 --