If he cuts down on his strikeouts he won't keep up that BABIP but it will certainly help his average. If he keeps up a 27.1% k rate it might get ugly quick, even if a few more fly balls go over the fence.

Then there's the possibilty that it isn't luck, but that he has stumbled into the secret of how to hit a baseball better, making David Wright the pioneer of a new baseball era, like Babe Ruth. Pretty unlikely, but not impossible.

You know, this is the kind of stuff that makes mainstream people hate statheads. It's baseball! It's a game!! Why is it so wrong to just enjoy David Wright? Why must everything be about projecting the future - just enjoy the present, people!

Does the rampant overuse of “luck” to explain any presently unexplainable outcome grate on anyone else?

Well, it is an achievement. In the same way that a no-hitter is an achievement.

Does the rampant overuse of “luck” to explain any presently unexplainable outcome grate on anyone else?

There are no words for how annoying I find it.

I'll make you a deal. I'll agree that Wright's BABIP has been flukey/lucky, if the announcers stop harping on Wright's strikeouts and lower home run production. Last night you would have thought that the guy was hitting .260 with a .700 OPS -- rather than .350 with a .940 OPS -- the way Steve Phillips just harped on how he needs to strike out less and how they need him to hit with more power because Delgado and now Beltran are out (never mind that he has a .502 SLG average). It took Sutcliffe to finally say, "Um, the guy is leading the league in hitting, at .350. Can we restore some sanity here, please????"

Wright is likely to return to career norms in BABIP the rest of the way, but he's also likely to return to career power norms. Which means he's likely to be the best hitting third baseman in the NL. Which means everyone should basically just STFU.

Damn, I'm ornery here. Why does listening to Steve Phillips broadcasting a Mets' game always do that to me?

Although I don't know why I think this, for some reason I have it stuck in my head that Wright's high BABIP is likely a statistical fluke, while his high K rate and lack of power are likely something to worry about. Can someone explain to me why I am wrong?

Damn, I'm ornery here. Why does listening to Steve Phillips broadcasting a Mets' game always do that to me?

Because Steve Phillips is demonstrably dumber than a sack of hammers, yet is paid who knows how many gazillions of dollars a year to disgorge his ill-considered observations on TV?

Just guessing, here.

Damn, I'm ornery here. Why does listening to Steve Phillips broadcasting a Mets' game always do that to me?

It could be worse. He could be working for the Mets again.

Can someone explain to me why I am wrong?

Because they are BOTH the product of the exact same thing: sample size flukes. Any departure from a normal pattern can happen in half (or less than half) a season. In 2006, for instance, Wright had 20 home runs in 386 PAs pre-ASB, and 6 homers in 275 PA's after the break. People looked for all sorts of explanations, like his having lost his power stroke because of the Home Run Derby. In 2007 and 2008, his power was back just fine.

Is it possible something more is behind it? Can't say it's not for 100% certain, but I'd say that until you have something more to go on, go with the guy's career norms and not a 65 game fluke as your simplest explanation. For both things.

Wright hit .280 for April and I was reading stories about how he was through and would soon join Joe Charboneau in the ranks of great fizzlers-out. Now he's hit .382 since April and the stories are about how that means he's obviously in big trouble. Jeez, it's like Sam says: Wright has been one of the best players in the league for several years, he's only 26, his season line looks a little odd but he's been ... one of the best players in the league. Big surprise.

I also notice that the Mets are in second place, the federal budget is in the red, and there is upheaval in Iran. All of which can be traced to the fault of David Wright.

Although I don't know why I think this, for some reason I have it stuck in my head that Wright's high BABIP is likely a statistical fluke, while his high K rate and lack of power are likely something to worry about. Can someone explain to me why I am wrong?

You're wrong because the player in question is David Wright, who has a well-established track record as a tremendous major league hitter.

If he were a rookie without much history behind him, it would be very reasonable to look at what he's doing and conclude that something's got to give: he can't sustain that BA with that K rate and that HR rate. No one does for any length of time; either the BA dramatically drops or the HR rate dramatically rises, or both.

But we know Wright; we know he's one of the best power hitters in the game. Yes, this so-far 2009 stat line is flukey and won't be sustained, but the reversion to his career norm rates will simply mean that he'll hit .330 with 20 HRs this year, instead of his typical .310-.320 with 30 HRs. Not exactly a problem.

EDIT: Or what Sam said in #11.

It could be worse. He could be working for the Mets again.

Assistant to the GM Would be the best job for him. Then the GM can just ignore his banalities and he can leave the rest of us alone because he would be mercifully forbidden to talk to us.

#11 and #14 are reassuring. I understand sample size and regression to mean on an intellectual level, but my gut is fooled by the crazy numbers.

I understand sample size and regression to mean on an intellectual level, but my gut is fooled by the crazy numbers.

The thing is that we "know" that the BABIP is unsustainable, because it would set records. Nobody's ever hit with that high BABIP for a whole season. But people have struck out as much as Wright is doing and people have hit as few home runs as Wright is hitting. So it seems possible that Wright is playing at his true level in those categories, while it does not seem possible that his BABIP is his true level.

That's why it seems worrisome, but Steve and Sam are correct anyway.

While I agree there are sample size issues and the BABIP will decline while the K rate will improve, K rate is far more significant for a batter over the same PA than BABIP rate.

Here are links to expected regression % to mean for both K rate and BABIP for different PA:

BABIP

K rate

Dan, how did you figure those? I'd be much obliged.

Does the rampant overuse of “luck” to explain any presently unexplainable outcome grate on anyone else?

BABIP has been one of the most poorly communicated stats in the baseball world. When BABIP came into the lexicon, it was basically introduced as" "A BABIP of .300 is newton's 1st law, anything above or below this is "luck".

Turns out hitters can have BABIP numbers that are significantly divergent from .300 and luck has little to do with it.

Dan, how did you figure those? I'd be much obliged.

I had to develop a regression model for ZiPS and for minor league translations because aging/development and regression to mean are different issues and you want to be able to treat them individually.

To put it simply, with retrosheet data, I had data for all hitters for the last 50 years for every cluster of at-bats (I applied generic aging factors in reverse where clusters go over multiple seasons).

Then, with the help of Statistica, I looked at various stats for every PA sample size possible for every player up to career length - 1000 PA. For example, with Pete Rose, that amounted to well over 100 million chunks of plate appearances.

BABIP has been one of the most poorly communicated stats in the baseball world. When BABIP came into the lexicon, it was basically introduced as" "A BABIP of .300 is newton's 1st law, anything above or below this is "luck".

Especially because of the difference (however small you may believe it to be, it's still exists) between hitter BABIP and pitcher BABIP.

BABIP has been one of the most poorly communicated stats in the baseball world. When BABIP came into the lexicon, it was basically introduced as" "A BABIP of .300 is newton's 1st law, anything above or below this is "luck".

Turns out hitters can have BABIP numbers that are significantly divergent from .300 and luck has little to do with it.

That's badly overstating. Nobody, at least nobody who thought about it for more than 5 seconds, ever thought BABIP was luck for hitters.

Heck, when BABIP was introduced into the lexicon, almost nobody but Voros (maybe) thought it was luck for pitchers. That notion did grab hold reasonably quickly and making an allusing to Newton's laws is just standard hyperbole, not extra double hyperbole.

But the main problem is that I simply will never understand why anyone cares much about BABIP for hitters. The notion makes sense for pitchers -- the other stuff is "defense indpendent" and there's a logic to viewing a pitcher independent of his defense (which he will play in front of for an entire season). There's no need for a defense-independent measure of hitting. If we were going to break down overall hitting performance into components, hitting should have been looked at in terms of "on-contact" as the first step, not BABIP. (Note, that's not quite the same as saying there's no utility in breaking down on-contact performance into different components to better understand on-contact hitting.)

As to "luck" ... I don't care if you call it "luck", "random error" or "unexplained heterogeneity that we have little/no choice but to treat as random error." Far more annoying are those who offer "Don't call it 'luck' ... it could be, y'know, real although I don't have any idea what could explain it."

And yes, a 460 BABIP is luck, not "luck". Nobody's true talent is at that level. Now, maybe Wright's true-talent BABIP has increased such that a small-sample 460 BABIP is less lucky than we think. It's a possibility.

But yes, the questions are: (1) has there been a real increase in BABIP or BA on-contact? (2) has there been a real increase in K-rate? (3) has there been a real decrease in HR rate? (4) Is (1) enough to offset (2) and (3)?

Answer those and you can make some judgments as to whether Wright is a better or worse hitter than he used to be. I'd argue he almost has to be a different hitter becaue it would be pretty unlikely to find three rates at fairly extreme variance from established norms to be just "luck" even in just 300 PA. Still, chances are the change in overall production isn't going to be huge either way and Sam et al are right not to be very worried.

What's weird is that HR-rate, K-rate and BA on-contact (and maybe BABIP) tend to be positively correlated. If the HR-rate was up as well, you'd just say he's turned into more of a TTO hitter. And maybe he did change his approach in such a way (just swinging harder) but the HR spike has yet to come. But these are pretty wacky numbers for a guy with Wright's track record.

In response to the anti-luck-as-a-conclusion crowd:

If you roll a die 60 times and get a four 27 times, that's luck. Statistically, you'd expect to roll a four 10 times in 60 tries.

Similarly, when David Wright puts a baseball into the field of play with his bat, he is expected to get a hit about 35% of the time. Instead, he's gotten a hit 47% of the time. This is a combination of Wright hitting the ball well, the defense being bad or in bad position to field, and the ball simply finding holes. Some of the difference between his current BABIP and his average BABIP can be explained by tangible factors, but not all of it. That last part -- Wright's batted balls simply finding holes -- is luck.

There is such a thing as luck in baseball.

“Luck-as-a-conclusion,” I think I like that.

Luck is hardly ever the conclusion to actual arguments I’ve seen here, only to theoretical ones, within the confines of parameters just like the ones set by yourself and Walt before you. More often than not it’s used as a catchall, explaining everything from a divergent BABIP, to a particularly unexpected start.

I have no problem believing luck to be a factor in a game played by men, but it’s long since become more than a “factor” to most I see use it.

Although I don't know why I think this, for some reason I have it stuck in my head that Wright's high BABIP is likely a statistical fluke, while his high K rate and lack of power are likely something to worry about. Can someone explain to me why I am wrong?

You are not wrong.

Every metric has a certain amount of noise to it. For something like K/PA, it has very little amount of noise. For something like BABIP, it has alot of noise.

The important thing is that this is not an either-or situation.

Every metric has a certain amount of noise to it. For something like K/PA, it has very little amount of noise.

Wright's K/PA from 2004 to 2008 was .16 (in 2312 PAs). Thus far in 2009, it is .23 in 299 PAs. He had shown no indication of an increase in his most recent seasons prior to this one -- last year, for instance, he was right on his career norm of .16. So it's not like this is the continuation or the acceleration of a trend.

Thus, if I accept the premise that the metric of K/PA has very little noise -- which I do -- then the version of it I would accept as the most quiet in David Wright's case is the one that has 2312 PAs of information to silence the noise.

This is something real, concrete, and tangible, unlike variation from Pythag or any of a dozen other instances where random chance and Fate distort numbers. As such, it should be much easier for someone who is against calling it luck to outline their position clearly, something that is consistently failed at in (for example) Pythag arguments. I'll ask the same question I've asked countless time before to start, and give the same response to a yes answer:

Is your quibble just with the term of art, would you be okay if it was "random chance" or "statistical variation"? If so, seriously, please stop this. It's a stupid argument, you've "lost" (people have, do, and will continue to use "luck" instead of either of those in the vast majority of cases), and you just sound like /. arguments about "hacker isn't a negative thing, the media misuses the term". You're only muddying the discussion with some petty linguistic I'll-be-kind-and-call-it-nuance.

If your quibble is with the fact that this demonstrates (or just has a strong chance of it) a skill or talent or laudable feat, such as the arguments for why Team A outperformed Pythag by 6 wins to surprise the division, please provide an explanation. At least semi-plausible. With teams you hear a bunch of crap that's too abstract, like "they know how to win", here is, again, something tangible. How is he doing it?

And if your quibble is purely of the "maybe you just can't measure the real cause because your instruments suck and you're stupid", again please stop. No sane person around Primer thinks or claims that it can all be measured and analyzed. We know this. Once again, this doesn't contribute unless you have an idea what it might be or a place to start

If you look at his splits, he is striking out more at home than away, and I remember him saying something early in the year about how he was having trouble seeing the ball at Citi. His K rate at home is .28, away it's .18, very close to his career norms. If Citi is contributing to his K rate the Wilpons better get on this quick and fix it.

This is why the work luck if not wrong is imprecise and misleading.

I'm not a great basketball player. If I had to guess I'm probably a 60% free throw shooter. As a 60% free throw shooter it is pretty unlikely that I would make ten free throws in a row. However if I were to make ten free throw in a row [with nothing but net] that wouldn't be luck. I actually made those shots and it was actually my skill that made them go in regardless of the fact that statistically speaking I am unlikely to repeat the act next time.
By contrast if I were to put a few in off the backboard that would be luck.

The widespread usage of luck by the baseball statistical community willingly and willfully obfuscates this difference and as a result does a pretty grave diservice to the understanding of baseball.

#11 and #14 are reassuring. I understand sample size and regression to mean on an intellectual level, but my gut is fooled by the crazy numbers.

The .480 or whatever really sticks out and is impossible to sustain, for anyone. It's possible that a player could have Sammy Sosa's K Rate, and Brian Robert's HR rate so if this is all you'd ever seen from Wright, you'd figure he was a lucky SOB of a AAA player about to face some adjustments.

But it's David Wright. He has 4 years of 30 HR, 110-120 K performance under his belt. He will make contact. And when he does he will hit it hard.

Everyone agrees that Wright actually got all those hits, and everyone agrees that he's very unlikely to maintain his 2009 BABIP going forward. Some people describe this situation as "luck," and some people get upset when others use the word "luck." Who cares what word people use as long as everyone understands each other?

Looking at the rest-of-season ZIPS projections over at Fangraphs, Wrights' weird 2009 has lifted his projected BABIP from .335 at the start of the season to .363 going forward from now. His K% changes from 19% to 21%, and his ISO drops from .239 to .220. The end result? He's now projected for a .414 wOBA instead of the .415 he had at the start of the year.

That seems like an awfully large jump in BABIP projection from 300 PA of data, but I'm not the guy who's run regressions...

Austin Jackson is doing the same thing in AAA right now. Dan, any chance you could do a MLE/update projection based on his season so far?

There are multiple parts to a batted ball event. Some batted balls are hit really hard, such that if the fielder hadn't been right in front of them, they'd have been hits. In that situation, it was the hitter's skill that allowed him to hit it like that, but it was bad luck that made it not a hit.*

Conversely, some batted balls are little dribbling grounders that sneak through the infield, because no fielder was able to get to them. If they'd been hit on a slightly different trajectory, or if the defense had been positioned slightly differently, they'd have been outs. In this case, the hitter had good luck in getting a hit, because his skill didn't result this time in his hitting a ball that was likely to be a hit, and yet it was.**



* From his point of view. There's also, of course, the defensive team's point of view, from which it might be that they had him positioned exactly right.. That's skill on their part, and still luck for the hitter.

** And, of course, it was also bad luck for the pitcher, and some combination of bad luck and wrong positioning or poor range (skill) by one or more defensive players.

So while these "luck" events involve skill of some of the actors, they really are a matter of luck, as the concept is normally understood, from the point of view of at least one of the actors, and individual statistics, like BABIP, are measurements of events from the point of view of individual actors.

However if I were to make ten free throw in a row [with nothing but net] that wouldn't be luck.

It would be the way most people use the term: an unexpected beneficent event or series of events.

Everyone agrees that Wright actually got all those hits, and everyone agrees that he's very unlikely to maintain his 2009 BABIP going forward. Some people describe this situation as "luck," and some people get upset when others use the word "luck." Who cares what word people use as long as everyone understands each other?

That's the thing. Everybody on this board knows that people mean this when they say luck, but we still have to have this discussion anyway.

Austin Jackson's wacky BABIP, when made a little saner, takes the happiness out of the translation, which drops all the way to 258/320/322.

I'm not a great basketball player. If I had to guess I'm probably a 60% free throw shooter. As a 60% free throw shooter it is pretty unlikely that I would make ten free throws in a row. However if I were to make ten free throw in a row

Except there is nothing in baseball that players have as much control over as a free throw.

It would be the way most people use the term: an unexpected beneficent event or series of events.

I described two completely different situations. Only illiterate people use the same word to describe different things.

Only illiterate people use the same word to describe different things.

You have the right to say that, but you are not right.

(For those as literate as Gaelan: in English, right can either mean that which one is entitled to do, and that which is correct.)

Only illiterate people use the same word to describe different things.

Welcome to the English language.

For all of the handwringing, David Wright is providing just about as much value as he always has. The shape of his performance is a bit odd, but at the end of the day, it just falls into the category of "Good Player Playing Well". That's hardly an earth-shattering headline. Maybe he just figured out another way to be good for some reason.

Now, of course his BABIP is unsustainable, but players are dynamic and David Wright is not a player who shies away from tinkering with his swing. Maybe he's swinging the from his heels all of the time on a more level plane. Certainly, I would have expected some player in baseball history to have taken this approach and to have had similar results, but maybe not. It's a boring answe, but the most likely conclusion is that Wright's performance is due in part to skill and in part to "luck".

Does the rampant overuse of “luck” to explain any presently unexplainable outcome grate on anyone else?

Oh god yes. It's sabremetric fugacity.

And yes, a 460 BABIP is luck, not "luck". Nobody's true talent is at that level. Now, maybe Wright's true-talent BABIP has increased such that a small-sample 460 BABIP is less lucky than we think. It's a possibility.

Luck is not a replacement word for repeatable skill, it's imprecise language and just causes all kinds of aggravation and confusion because "luck" has all kinds of institutional meaning behind it outside of the sabremetric lexicon.

A 460 BABIP is statistically and historically very unlikely to be a repeatable skill.

You have the right to say that, but you are not right.

Oh snap.

One of the Fangraphs guys said something only a handful of the guys in his database having a BABIP over .400, yet he fails to mention Rogers Hornsby or Ted Williams. I know they were before the "Retrosheet Era" which is ever-changing, but surely you can figure out BABIP without play by play info.

In any case, people throw out the word luck in the sabersphere to explain things in the same answer someone 800 years ago might have used the word God.

#27:In response to the anti-luck-as-a-conclusion crowd:

If you roll a die 60 times and get a four 27 times, that's luck. Statistically, you'd expect to roll a four 10 times in 60 tries.

Similarly, when David Wright puts a baseball into the field of play with his bat, he is expected to get a hit about 35% of the time.

My apologies as some likely addressed this between your post and mine. You cannot equate rolling dice and variables in the results with BABIP; there is a difference between the two. One (dice roll) is a measurement of physical and mathematical probable outcomes - we can definitively say what the probable outcome of a given number of die rolls are assuming we know the number of possible outcomes and each outcome has an equally likely chance of occurring. We do not need to know what, historically, the particular die has rolled in the past.

BABIP is a measurement of historic performance - either what an individual player has achieved or an aggregate of what all players have achieved. We certainly cannot say definitely what any one individual's BABIP is expected to be as we can with a die roll given the physical difference of outcomes and the variables that go into them between the two. With the imperfect effects of the human body and mind, the game of baseball itself and the passing of time I would caution against equating definitive probabilities and historic BABIP results and tossing any variability of the latter away as "luck".

yet he fails to mention Rogers Hornsby or Ted Williams. I know they were before the "Retrosheet Era" which is ever-changing, but surely you can figure out BABIP without play by play info.

Per BBRef Williams' career BABIP was .328 with a high of .378 (1941) and a low of .246 (1959). Hornsby's was .365 with a high of .422 (1924) and a low of .304 (1918). In both cases I excluded low AB seasons.

For some reason, I thought that Williams struck out less or got confused and was thinking about BACON. In any case, I can't link the Fangraph's post for some reason, but here is what R.J. Anderson had to say:

I ran a query through my database for the highest BABIP with 300+ at-bats, and the best I came up with is Reggie Jefferson’s 1996 (.408), Rod Carew’s 1977 (.408), and Jose Hernandez’ 2002 (.405). A couple of others topped .400, but the highest of highs is just shy of .410. Nobody comes near .420, or .450, or .470.

So it was seasonal and still no Hornsby.

Wright hit .280 for April and I was reading stories about how he was through and would soon join Joe Charboneau in the ranks of great fizzlers-out. Now he's hit .382 since April and the stories are about how that means he's obviously in big trouble. Jeez, it's like Sam says: Wright has been one of the best players in the league for several years, he's only 26, his season line looks a little odd but he's been ... one of the best players in the league. Big surprise.

Well, as far as the New York media goes, the fact that Wright isn't a Yankee has a lot do with it, as well...

Edgar Martinez had a career BABIP near Wright's and had a few seasons with ~.380 BABIP. A quick query has Wright in the top 50 for career BABIP for players with 1500ABs since 1950, not including this year. I think it is certainly possible that he could end up with a BABIP near .400, though obviously it is likely to be less than that.

All the .400 BABIP seasons with 450 ABs. Not suprisingly, the top few are all 1922-1924



Post-Jackie:

Thus, if I accept the premise that the metric of K/PA has very little noise -- which I do -- then the version of it I would accept as the most quiet in David Wright's case is the one that has 2312 PAs of information to silence the noise.

What I was trying to say is that the metric K/PA allows you to weight recent performance more heavily.

If, for example, the "standard" weights for the last 3 years is 5/4/3, then you would weight BABIP as 4.5/4/3.5, and you would weight K/PA as 7/4/1 (or some such).

You NEVER ignore any past performance. What you do is weight them based on their persistency to forecast the future.

Since players are human beings, then we want a metric that more closely aligns to his current conditioning, strength, and speed. K/PA has limited noise, and so, you weight his recent performance more.

If players were NOT humans, then you would have no reason to weight recent performance more, and you'd stick with his career totals.

I'm not a great basketball player. If I had to guess I'm probably a 60% free throw shooter. As a 60% free throw shooter it is pretty unlikely that I would make ten free throws in a row. However if I were to make ten free throw in a row [with nothing but net] that wouldn't be luck. I actually made those shots and it was actually my skill that made them go in regardless of the fact that statistically speaking I am unlikely to repeat the act next time.
By contrast if I were to put a few in off the backboard that would be luck.

The widespread usage of luck by the baseball statistical community willingly and willfully obfuscates this difference and as a result does a pretty grave diservice to the understanding of baseball.

For purposes of projecting your future free throw percentage, there really is no difference, unless of course the fact that you made ten in a row with "nothing but net" is truly indicative of the acquisition of a new and higher skill level. You did something that is not repeatable or sustainable, which by many people's definition is luck, either way.

Anybody with a 60% free throw percentage is going to hit 10 in a row periodically. The act of making the shots may not be luck, but it is certainly a "lucky streak". There will also be stretches where you will make only two or three out of ten. These are unlucky stretches and by no means indicates that you have lost your 60% ability. To attribute the differences between the 10/10 stretches and the 2/10 stretches to a skill difference as opposed to a luck difference is not correct.

I described two completely different situations. Only illiterate people use the same word to describe different things.

All you're doing is describing two different kinds of luck.

A) a beneficial result resulting from minimal skill

B) a series of unexpected, unrepeatable, and unsustainable beneficial results resulting from a higher skill level than is normally demonstrated.

These are the two situations you describe. They both require "luck" to have occurred.

With the imperfect effects of the human body and mind, the game of baseball itself and the passing of time I would caution against equating definitive probabilities and historic BABIP results and tossing any variability of the latter away as "luck".

Thanks for the caution. In that case I will simply refer to it as an "unsustainable and unrepeatable level of BABIP" in future at bats. "Luck" is a four letter word and really offends people who like to believe that there is no "****" in baseball.

Thanks for the caution. In that case I will simply refer to it as an "unsustainable and unrepeatable level of BABIP" in future at bats. "Luck" is a four letter word and really offends people who like to believe that there is no "****" in baseball.

You could use chance, which has just as few syllables and more accurately describes what you mean. But if it's more important to charge ahead with a less perfect word because that's the one the gang used from the outset, fine. But please don't get exasperated when this argument crops up again and again.

Who cares what word people use as long as everyone understands each other?

The problem is that "people don't understand each other" and it often leads to inference trails that are incorrect, which then leads to more inference trails that are incorrect. The best example is the faulty "Pitchers have no effect on hits on balls in play". That has lead to all kinds of nonsense over time, about measuring and rewarding actual output as well as projecting future output. Even in this thread, you have posited multiple definitions for "luck".

It often still leads with people misinterpreting the value of an event based on the projectability of a future event.

When people usually make such inartful statements, there are usually armies of their acolytes stepping in saying, "What so-and-so really meant was..." Most often this isn't correct and is contradicted by the person themselves. Even if its not, the speaker's apologists will issue clarifications that are logically impossible to co-exist.

It occurs all the time at a mild level. Even in this thread you have, Treder saying:

"You're wrong because the player in question is David Wright, who has a well-established track record as a tremendous major league hitter. "

and Tango saying:

"You are not wrong."

as a response to the same statement. (Incidentally, Tango has the most correct answer). The reason that one of the statement's is wrong is because they don't understand the concept of when there is a controllable causal element and when there is an externality. DISTINGUISHING BETWEEN THESE TWO THINGS IS THE SINE QUA NON OF ANALYSIS.

The other thing the frequent misuse of the term "luck" does, even among those that can understand what is actually occuring, is that it sets up fictions. For instance, a constant point of disagreement I would have with real analysts like Tango, is the term of art that has slipped in their craft of "true talent level". IMHO, they are searching for a Platonic and ontological measure of a player. Even if such thing existed, it would be fluid and not static (ie the "true talent level" today is different than the "ture talent level" tomorrow, and if you tried to use any intergral analysis to measure changes, I opine it would move faster than for any such construct to have meaning.)

Most important, the confusion regarding the meaning of the term "luck" becomes annoying. In addition to impeding analysis by not differentiating between controllable and external elements of performance, and obfuscating calculations of value, its used as fanboys for excuse-making on any non-desired outcome and to put such outcome in trappings that makes it look like there is some science that their people are better.

Third order pennant flags fly forever.

But the main problem is that I simply will never understand why anyone cares much about BABIP for hitters.

Because if a player hits .250, .260, .250, and .300
and his BABIP is: .300, .300, .300, .350

then that .300 average is less likely to be repeated/sustained (ie represent real improvement)
than if his BABIP is .300, .300, .300, .300

other than that...

it was useful to poke at Phillies Phans who insisted based upon 2006 that Ryan Howard was a ".300 hitter"
well, Howard was going to hit .300 only if he sustained the greatest ion con6act numbers in the history of baseball (possible, but very unlikely)

For purposes of projecting your future free throw percentage, there really is no difference, unless of course

For purposes of calculating the value of his past free throws, it is very important. So is the context of WHEN they happened, as the contribution of a specific value will often different from event set to event set, depending on the definition of the event set.

Since players are human beings, then we want a metric that more closely aligns to his current conditioning, strength, and speed. K/PA has limited noise, and so, you weight his recent performance more.

If players were NOT humans, then you would have no reason to weight recent performance more, and you'd stick with his career totals.


I agree totally. That isolates the biggest problems that occur from the use of the term "luck" and the robotofication of players by the weaker sabermetrician(s) (or sabermetric fans).

They treat any variation of performance as CAUSED (not described as an additional circumstance) by luck. That leads to no further investigation on which of the various outputs (e.g. release point, mechanics, K/PA output, BABIP) is caused by a temporal or permanent attribute that will affect future performance over some defined interval of time.

That leads to the most classic examples of fallacies you see, such as:

(1) Assuming an inning of Pedro Martinez is always better than an inning of Mike Timlin.

And a specific Treder special,

(2) Assuming a PF of John Smoltz is always better than the PF by Ray King (when Smoltz is so injured he is put on the DL the next day).

this thread is an interesting read. On behalf of other semi-lurkers, thanks to everyone who puts so much effort and thought into their writing here.

the "true talent level" today is different than the "ture talent level" tomorrow

You are correct.

Seeing that a player is a human being, his true talent level changes by the second. As it does for me in my profession, and anyone else here in theirs.

Yes, it only makes the teeniest, tiniest of differences second-to-second. And, as the more time elapses between two data points, the more relevant the player-is-a-human becomes.

This is why I weight performance based on this:
weight = .9994^daysAgo

I can just as easily change the "daysAgo" term to "secondsAgo/86400".

so..... in the context of the whole space/time continum I guess we're all "lucky" that we were born on earth at a time when a game called baseball is played.
And the internet was invented; and a group of guys decided to build baseballthinkfactory.
Of course there was probably a degree of "luck" exhibited by our parents in their meeting, finding one another attractive enough to copulate thus creating each of us.
Or, perhaps, one of the two used their "skill" to persuade the other that they were an attractive donor/conceiver.
Thus, they "got lucky", as we used to call it back in the day whenever we were able to convince some sweet young thing that "yes, I'll still respect you in the morning".

Of course, we can sweep all of this stuff under the rug and just call it God's will. :-)

(ie the "true talent level" today is different than the "ture talent level" tomorrow, and if you tried to use any intergral analysis to measure changes, I opine it would move faster than for any such construct to have meaning

There should be some mention here of the Heisenberg Uncertainty Principle -- or Schrödinger's One-o-Cat

On behalf of other semi-lurkers, thanks to everyone who puts so much effort and thought into their writing here.

Seconded. I'm glad I asked what seemed at the time like a stupid question.

You cannot equate rolling dice and variables in the results with BABIP; there is a difference between the two. One (dice roll) is a measurement of physical and mathematical probable outcomes - we can definitively say what the probable outcome of a given number of die rolls are assuming we know the number of possible outcomes and each outcome has an equally likely chance of occurring. We do not need to know what, historically, the particular die has rolled in the past.

BABIP is a measurement of historic performance - either what an individual player has achieved or an aggregate of what all players have achieved.

What? You're saying that the two cases are different because in one we have a probabilistic model and we could measure historical numbers (but don't because we don't need to) and in the other we can only use historical data. So tell me this:

I'm stuck on a yacht in choppy seas, and after battening the hatches, I sit down with a pair of dice and throw them 10,000 times, meticulously recording each result. My results aren't exactly matched with probability, but that's okay because I have never even heard of probability. I have 4 kids, and every time I see the number 4, I gain a little bit of strength and hope from thinking about them; they are what keeps me pressing on through the storm. In those 10,000 rolls I get a 4 on 800 of the rolls (instead of the 850 probability would suggest.) This is all the information I have.

It's still rough out there, the boat is pitching to and fro, so even if I were a dice shark (they do exist, though it's a grueling skill to master) I can't hope to throw the dice the same way every time. This time, I roll the dice 100 times, again noting the results. I get a 4 on 25 of the 100 rolls.

What would you call that? And what, if any, differences are there between that and BABIP that make comparing the two impossible?

Dr. Fluke is 4 for 4 today, with 3 runs scored.

Ba-da-bing, ba-da-bang, BA-da-BIP.

The best example is the faulty "Pitchers have no effect on hits on balls in play"

This is not faulty.

It is the essence of why DERA or xERA is a much better predictor of future ERA than any method that assumes that pitchers can somehow influence the rate at which balls in play become hits. Or that they will be able to somehow repeat their success or failure from previous seasons with regard to BABIP.

Dr. Fluke is 4 for 4 today

Can people who think it's all just luck go through a game like that and separate the wheat from the chaff? "Fluke, fluke, legitimate hit, fluke."

Can people who think it's all just luck go through a game like that and separate the wheat from the chaff? "Fluke, fluke, legitimate hit, fluke."

Can you go through a week's worth of games and tell which hit is the one that separates a .300 hitter from a .252 hitter? (1 hit/wk * 4 wks * 6 months = 48 points of BA.)

It was that broken bat single that landed just out of the shortstop's reach

Can people who think it's all just luck

You did realize that the "Dr. Fluke" thing was just a touch tongue-in-cheek, hmmmm?

37:

Except there is nothing in baseball that players have as much control over as a free throw.

Hitting an Adam Eaton curveball says hi.

/hi-fives everyone

Yes, it only makes the teeniest, tiniest of differences second-to-second. And, as the more time elapses between two data points, the more relevant the player-is-a-human becomes.

That depends on the two seconds in question. It is not a continuous function. The second after Dave Dravecky broke is arm is going to be a rapid jump in any TTL measure you have over the second before such TTL measure.

This is not faulty.

That shows you precisely why everybody doesn't understand what is meant. There are people that still believe that pitchers have no effect on balls in play. Its the same people that so expand the meaning of the word "luck" so that any attempt to use what you are left with will end up in an incorrect outcome.

There are people that still believe that pitchers have no effect on balls in play.

There are people that don't believe in dinosaurs.

You want to spend your life arguing with the lunatic fringe, go ahead. But you should save it for court, because at least you're getting paid for it.

There are people that don't believe in dinosaurs.

Carl Everett = Vance Law Revue?

You did realize that the "Dr. Fluke" thing was just a touch tongue-in-cheek, hmmmm?

Uh, yeah. I wasn't including you in that group.

Dr. Fluke helped win the game today, making a double play ball bounce off Carpenter's glove, allowing him to safely reach first in the only inning the Mets scored.