And in the mold of so many others, he will no doubt beat the S.F. Geronts[tm] at least twice in 2005.

Mohr hits 30 HRs this year and Jim Bowden signs him to a 3 year, 15 million dollar deal next year.

How bad is Brad Hawpe in the OF? I know he's a converted 1B blocked by Helton, most players like that turn in some real ugly UZR's.

Mohr might be the better all around player of the two. If he starts I can see a Jay Payton, 2003 type of season for him.

And in the mold of so many others, he will no doubt beat the S.F. Geronts[tm] at least twice in 2005.


Of course - that's par for the course.

giving him that reek of awesomeness

I don't think that's awesomeness you're smelling Dan.

I always thought the Rockies had a plan...it just changes every month. This month's plan-sign as many ex-Phillie/ex-Twins as possible. With Milton off the market, Mohr was the next logical choice.

Mohr signed for $100,000 less than the $1 million the White Sox are flushing down the toilet on Timo Perez. That does not make me happy.

or being the right side of a platoon despite the fact that he's shown a reverse platoon split during his time in the majors.

Dan, does this mean that you don't subscribe to the theory that all right handed batters have equivalent platoon splits? Or do you have evidence to contradict this theory? It wouldn't seem that Mohr has enough ABs to draw such a conclusion.

I think platoon split is real, for whatever reason. R vs. R is not as extreme as L vs. L but surely exists.

Dan, does this mean that you don't subscribe to the theory that all right handed batters have equivalent platoon splits? Or do you have evidence to contradict this theory? It wouldn't seem that Mohr has enough ABs to draw such a conclusion.

No, I'm not Dan, but let me interject...

With every rule there are exceptions. Even if "all" RHBs are 8% better against lefties and the rest is sampling variation, I think anyone would give room for freaks-of-nature and other anomolies that would not fit in the category "all". Is Mohr an anomoly? I wouldn't give anyone the moniker of having reverse platoon splits until they had 5 full big league seasons under their belt that showed a distinct pattern. Mohr's not there yet, but in over 1000 major league ABs he's shown a distinct ability to hit better against rightes than lefties. That's not enough to prove to me he has "reverse splits", but its enough to give me pause before using him as a L-R platoon player.

Its a good signing by the Rockies. Mohr's good enough to be a starting player, and would make a good 4th OF/pinch hitter on most any club.

I always thought the Rockies had a plan...it just changes every month.

John Kerry, Rockies GM?

Dan, resist the urge to jump the gun on the Randy Johnson deal

As far as RHB platoon splits are concerned:

One of the reasons we "know" that all RHB's have around the same platoon ratio, is that if we look at all batters who have shown extreme splits over any period of time, they still revert back to around the league average. So that necessarily means that no matter how many years Mohr or anyone else has anomolous splits, our best estimate of his true platoon ratio (our projection) is still around league average. A lot of people misinterpret what it means when we say that "clutch hitting does not exists" or "anything other than league average plartoon ratio does not exist for RHB's." This is an example of that misunderstanding.

When we say that (and it is true), that means that no matter what the sample size or magnitude of the splits, a RHB's true platoon ratio is somewhere around league average (1.09 for OPS).

Let me give you a good and clear example of what I mean. I say that no matter what a coin's sample heads/tails ratio is, the coin's true probabilty of landing on heads (or tails) is around .5. I don't think anyone would quarrel with that. Does that mean that there are no biased coins floating around? Probably not. There probably are. Does that mean that if I come to you and say, "Hey I just flipped this quarter I found in my pocket 1000 times and it came up heads 700 of those times," that I likely found one of those biased coins? No. It is a little more likely that if there are some biased coins, that this is one of them, but the chances went from like 1 in 10 million to 1 in 5 million, essentially meaning that our best estimate for this coin's true P of heads is now .50000001 rather than .5, or something like that.

Same thing for RHB's. And clutch hitting. And lots of other stuff like that.

Here is the semi-mathematical explanation. In general, the larger the sample size, the less we regress the sample result towards the population mean, for all things that have some variation in the population. However, if there is no variation in the population, then no matter how large the sample, the regression is always 100%. If there is little variation in the population (like RHB's platoon ratios or clutch hitting), then the regression is not that sensitive to even (reasonably) large sample sizes. For example, if a RHB's sample platoon ratio is 1.25 after 3000 PA's (5 years or so), the regression equation still tells us to regress by 83%. That gives us a true ratio of 1.12 rather than 1.09, and still far from the sample ratio of 1.25.

BTW, Timo Perez is not bad at all as a CF'er (I have him projected at around 1 win over replacement, in Slwts). Ditto for Mohr in RF. In fact, he is like 1.5 WARP (Slwts). That is a great signing for the Rox (one of their few good signings in recent memory).

(Hawpe's UZR in 04 was zero. If he is anywhere near average on defense, he has a nice overall projection.)

MGL, what you say makes a lot of sense, but it still seems a bit counter-intuitive that real variation would be so small as to always get swallowed up by the regression. Hitters vary a lot in their abilitly to hit pitches in different locations, and it seems that there would also be a lot of variation in platoon ability. Not that I'm really questioning your analysis, but it just seems like an odd result on the face of it. Thanks for the lucid explanation though.

I've always been somewhat skeptical of the platoon split, mainly because I haven't seen any of the studies first hand. Does anyone have a copy?

For me to be convinced, over a reasonable number of plate appearances (let's say 3000), only 5% of hitters should be outside two times the standard deviation away from this mean. Now the regression might work for the whole, but that doesn't mean there couldn't be a noticeable number of people who are well outside of the confindence interval for legitimate reasons.

MGL, the thing is hitters are not coins.

I understand that for predictive purposes you have found that using the same platoon ratios for everyone is most effective, but can you not even concieve of a player that had unique triats (unsymtrical musculature, funny eyes, something) that would create a true talent in that individual that was other than 1.09. Since we're dealing with a population of one (the freak) there would really be no way to tell if this was just a fluke of sampling (he was lucky to hit better against righties, for example), but maybe he has a reverse split? You can't say for sure, and I agree with that, but it seems foolish to rule it out.

can you not even concieve of a player that had unique triats (unsymtrical musculature, funny eyes, something) that would create a true talent in that individual that was other than 1.09.

I think he can concieve of it, but that we wouldn't be able to identify such a player from the data.

I think he can concieve of it, but that we wouldn't be able to identify such a player from the data.

Yes, that is correct.

And yes, there does appear to be some variation in the population. (That is indeed the difference between players and coins.)

That's why the regression coefficient is not zero (100% regression regardless of sample size). As well, when we construct these projection models using regression equations, we assume a somewhat normal distribution of talent. That means that we are assuming that there exist players who have true platoon ratios of any magnitude. However, because the range of variation in the population is narrow (most players are right around the league average), even when we see a player who has a large sample split over a large data set, it is overwhelmingly more likely that he has a true near average platoon ratio who got lucky then he is one of those rare players who has a true anomolous split.

It is Bayesian probability at its best!

None of these things are my opinion. They are mathematical facts based on what we find in the empirical data regarding platoon ratios among the population of all baseball players.

MGL's rule # 4,618 (an important one in life):

(For those of you who hate when I post one of MGL's rules, please disregard.)

It is fruitless to "disagree" with something when you either don't have all the data or you don't have the necessary background/capability to have an informed opinion. The exception is when the assertion (that you disagree with) is patently ridiculous in the face of reason, and even then, you have to be careful.

A corollary to that rule is:

When all the facts are layed on the table, most disagreements must evaporate...

It is fruitless to "disagree" with something when you either don't have all the data or you don't have the necessary background/capability to have an informed opinion.

That would preclude about 99.9% of the public from voting. Actually, that doesn't sound like a bad thing.

That would preclude about 99.9% of the public from voting. Actually, that doesn't sound like a bad thing.

Almost every disagreement in politics/world affairs comes from a lack of data (essentially the "truth") or a lack of understanding of that data.

Case in point (though not politics or world affairs) was the excellent article recently on BP on interleague play and attendance.

A related MGL's rule:

Never form an opinion on something you know little about until and unless you hear both sides of the story from a credible and informed source...

I agree completely with #19. The problem is that ambiguity is not something the human mind accepts easily. Very few people can accept a state of uncertainty while they wait for more data. If there is one thing we should be teaching in school it's that skill.

A related MGL's rule:

You should post these on the Wiki.

Very few people can accept a state of uncertainty while they wait for more data.

Not true, in fact the converse is true in many cases. People often are paralyzed in decision making when faced with uncertainty.

If there is one thing we should be teaching in school it's that skill.

It sounds like Real Time Decision making should be taught more in school.

Very few people can accept a state of uncertainty while they wait for more data.

Decisions, including very important ones, cannot always be put while off while you gather sufficient, or additional, data. You sometimes have to make your choices with the information you have, not the information you might want or wish to have at a later time.

Maybe I'm thinking about different things than you guys are. I'm not thinking about decisions as much as perceptions. How many ancient people looked up those points of light in the night sky and thought, "I don't know what they are, have to wait for more data". Lack of valid information never stops people from forming theories and sticking to them past the point of reason. Our minds find patterns where there are none rather than accept randomness or events beyond our comprehension.

Or maybe it's just me. I found the state of not knowing what will happen to me after death more painful than accepting that I will cease to exist in any form.

Decisions, including very important ones, cannot always be put while off while you gather sufficient, or additional, data. You sometimes have to make your choices with the information you have, not the information you might want or wish to have at a later time.

That's why you trust experts. Or, you elect people who you think have good judgement and values and trust them to trust the experts. That makes a hell of a lot more sense than making decisions about things you can't begin to comprehend.

Lack of valid information never stops people from forming theories and sticking to them past the point of reason.

I know, I deal with DIPS proponents, Bonds apologists and Beane worshippers all the time.

Our minds find patterns where there are none rather than accept randomness or events beyond our comprehension.

Or your mind accepts the pattern of randomness when you are unable to discern a recognizable pattern. If randomness is wrong, its no more of an inaccurate or dangerous conclusion than assuming its because of magic pixie dust.

RTMs are simple, if you don't produce a result within x time its a failure. Period. Its just as much a failure as if you take the wrong action. There are real world analogues.

ATMs are also simple, you gather as much information in the time you can and make the best decision at that point. There are real world analogues.

Most important there are situations where doing nothing produces a worse result than doing something, even if that something is not optimal. A great deal of real world analogues follow this model.

If time is known, than use all available time and all available resources, including experts to form the most optimal solution, then make a decision. If time is not known, make your best decision on time, gather all available resources and make a decision.

That's why you trust experts. Or, you elect people who you think have good judgement and values and trust them to trust the experts. That makes a hell of a lot more sense than making decisions about things you can't begin to comprehend.

Yes. Be quiet. These things are too complex for your plebian mind. Trust your betters to make the decisions you need made. Go back to sleep, little child.

Or your mind accepts the pattern of randomness when you are unable to discern a recognizable pattern.

No. That's only possible with training.

Yes. Be quiet. These things are too complex for your plebian mind. Trust your betters to make the decisions you need made. Go back to sleep, little child.

Humility is a virture my son.

Humility is a virture my son.

Sheep are slaughtered, Pops.

No. That's only possible with training.

Most decision making is a result of training. But the training could be inadequate. Saying the result is due to "randomness" is no more correct than saying its "the will of God" if that is your default condition for imperfect information. A phenomena or problem either has cause/explanation, does not have cause/explanation, or is intractable in case/explanation. Categorizing it incorrectly with a vogue term of art does not imbibe you with any more intellectual superiority than those who would attribute the condition to Zeus.

Call your deity whatever you wish. Understand ontology in any manner you can find, I begrudge you not your choice, but don't pretend your existentialism is inherently superior to someone's polytheism.

Someone tell me again how Backlasher is better to talk to than RossCW.

Never form an opinion on something you know little about until and unless you hear both sides of the story from a credible and informed source...

You can't find them on TV.

You sometimes have to make your choices with the information you have, not the information you might want or wish to have at a later time.

i was smiling sadly because this sentence reminded me of Rumsfeld's response to that soldier's question. *i know it's not same in any form* - it was just the musicality of the similar phrasology..

;-*)

i was smiling sadly because this sentence reminded me of Rumsfeld's response to that soldier's question. em]

Good point. Both were extremely reasonable statements I would say.