User Comments, Suggestions, or Complaints | Privacy Policy | Terms of Service | Advertising
Buy MLB playoff tickets, plus 2011 World Series, 2011 ALCS tickets and NLCS game tickets. We also have Texas Rangers playoff schedule, tickets to Red Sox games and Yankees game tickets. Plus, buy Phillies baseball tickets, Tigers playoff tickets and the biggies like ALDS baseball tickets and 2011 NLDS tickets. |
Demarini, Easton and TPX Baseball Bats
|
AllianceTickets.com has cheap MLB Tickets. Get all your Colorado Rockies Tickets, Seattle Mariners Tickets, San Francisco Giants Tickets and all your favorite baseball tickets here. We also carry cheap Denver Broncos Tickets, Seattle Seahawks Tickets and Denver Nuggets Tickets. |
Page rendered in 0.8170 seconds
54 querie(s) executed
Reader Comments and Retorts
Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
That's an extremely wide definition of "luck", and one reason why so many casual fans get turned off by sabermetrics. Most people differentiate results obtained by luck (the "funny bounce" mentioned above) from those obtained by skill. Cliff Lee pitching poorly isn't bad luck - it's poor performance from Cliff Lee. The fact that his poor performance happened during the World Series might be luck, but that's not the same thing as saying he was "unlucky" last night.
I understand that, but saying something like, "Well, Cliff Lee had only an 11% chance of pitching poorly last night, so it was bad luck that he did" does a really piss-poor job of describing Cliff Lee's pitching performance.
Now, of course that singular performance will be overanalysed because of the stage he was on, and it's important to take that performance in the larger perspective of his postseason and overall career, but that doesn't make what happened last night the result of "luck". Cliff Lee is not a coin, or a deck of cards.
No reasonable person thinks that Cliff Lee is going to pitch 7 2/3 innings and allow 3 ER every time out; they understand that sometimes he's going to pitch a shutout and other times he's going to get shelled. But no one understands last night to be "bad luck"; "bad luck" is what happened to that sea gull who flew in front of Randy Johnson's fastball.
A fielding error by the SS is bad luck for Cliff Lee and the Rangers, but not for the SS. For the SS it's just bad playing.
A bad pitching performance by Cliff Lee is bad luck for the Rangers, but not for Cliff Lee.
Playing against the Rangers, instead of the Yankees, is probably good luck for the Giants, but not for the Rangers. For the Rangers it is the result of good play.
Sound right?
Every profession has jargon. Lawyers, doctors, accountants, you name it. That must be why regular folks are turned off by lawyers and doctors and accountants and morticians and the like.
Seriously, somebody overreacted here. So it's jargon. Big deal. If it helps saber-whatevers talk to each other, great. If people who don't really care, don't really understand, well, nobody cares.
Now, losing on those short odds probably does require a bit of bad luck, but also a lot of factors both indirect and direct that lead to poor performance which can be directly attributed to this or that element.
The problem is using a regular word that has a regular definition that 99% of the world uses as a "professional jargon" term. There are plenty of "technical" terms for what MGL is referring to here: variance, residual, outlier, depending on exactly what you're talking about. Professional statisticians, econometricians, and the like (I'm the middle one of these) don't use the term "luck" in this way, even if they/we intuitively get what MGL is talking about here.
That said, of course "somebody overreacted here". This is the Internet, isn't that what it's for (well, that and watching porn)?
What do sabermetricians mean by "lick"?
I remember getting a note from a copy-editor once saying that my manuscript needed to be "right justified." I thought he was calling for me to be washed in the Blood of the Lamb.
Probably not what the other 99 percent of the population does, unless mom's basement has gotten considerably more interesting since I moved out (OK thrown out).
And if saber-whatevers don't care about being misunderstood by non-saber-whatevers, then everyone's happy in their respective circle jerks.
------------------
I agree, this is terrible. Maybe we could call it "variation", or something? Pretty much anything would be better...
I completely agree with this. The problem with the term "luck" as used by saber-whatevers is simple: At least around here, its purpose often seems to be a rhetorical weapon to brandish against (a) big Pythagorean underdogs who win the World Series (the 2006 Cardinals, just to take the most recent example), and (b) pitchers with gaudy W-L records and less than stellar ERAs. Whether intended or not, it often carries the odor of disrespect and / or sour grapes.
Same thing with the term "crapshoot" as applied to playoffs. While it's not an inaccurate word to ascribe to a short series, IMO it's one that sounds a lot better when used by humble winners rather than by sore losers.
I guess I'm wondering about the ultimate purpose of the original post. Is it just an explanation of the language sabermetricians use, or is it supposed to be a lesson about a more precise definition of "luck"? I have no problem with the former purpose. But the latter one seems a tad annoying.
Talking to each other is one thing, but that's not what MGL is doing. He posted this on his blog, available to the public. And one of the goals of sabermetrics is to grow in appeal, which means talking to the average fan.
How do you all know that Lee's performance was not bad luck but was short-term poor performance? What is the evidence for this claim? It could be true. Maybe pitchf/x data will reveal that he threw a lot of very hittable pitches. But if you're just basing this on observation, or (even worse) the outcome, that's not much evidence at all. He struck out 7, gave up 1 BB, and 0 HR. So the problem was entirely hits on balls in play. That could very well be luck, even by the more commonly understood meaning of the word.
If you want to make up new words or use highly (or even slightly) technical words, fine. But if you redefine common words then demean people who don't understand you, that's just being a prick. And prickishness seems the #1 reason people are turned off by advanced stats.
To the extent that the performance was a result of poor fielding by the Rangers or good hitting by the Giants, that's still not "luck", as "luck" is commonly understood. That's still a performance issue - we're just shifting the performance issue away from Lee and onto other players.
Unless you're saying that all the Giants hits were bleeders and bloopers that found gaps in the defense. Then yeah, most people would agree that that's luck.
What complete nonsense. The #1 reason people are turned off, by far, is because advanced stats sometimes lead to conclusions that challenges what a fan already believes. And #2 is that the advanced stats are complex, and take time and energy to understand that most fans -- quite reasonably -- don't want to extend. So prickishness is at best #3, and I'd guess there are several other factors that are bigger problems.
Ironically, I find some of the greatest prickishness appearing on this board these days is attacking sabermetricians for failing to show the proper humility, or use the right words, or express enough appreciation for grit, or whatever the hell someone thinks will make millions of fans suddenly embrace sabermetrics. Sabermetricians do what they do. If they aren't reaching as big an audience as you think they should, then become a writer and try to remedy that. But getting angry at MGL or whomever for doing their thing is just, well, pathetic.
This is the problem with this discussion. MGL isn't in the business of "describing Cliff Lee's pitching performance." He's interested in understanding Lee's talent, and more broadly how teams win and lose games. He basically cares about things that allow him to project future performance. If you want an analysis of Cliff Lee's performance in a single game, MGL isn't your guy. So he's talking about "luck" as he uses it, in a very different context. If that doesn't interest you, fine, but I can't for the life of me understand getting upset about this.
Then why is he talking about "luck" at all? If all mgl is interested in is broad-based analysis and attempting to project future performance, he should never talk about individual games at all, under any context.
Whoa, big fella. No one's really getting angry at MGL. We're critical of his use of luck in this instance. Whether this overuse of the word luck is reason 1 or 3 or 17 why people are dismissive of sabermetrics, it's one of the easiest ones for statheads to fix. Why our community stubbornly chooses not to has long perplexed me, and I've just as long argued against it.
First, this isn't directed at MGL, and I'm assuming that you're just using him as a stand-in for sabermetricians in general.
But here's an exercise that I'm way too lazy to do myself, but I'd still love to see someone undertake:
Take 10 prominent sabermetricians. Then take 10 prominent MSM beat reporters and / or baseball columnists with random degrees of sabermetric knowledge.
Then every February ask all 20 of these people to predict the coming year's standings and leaderboards, and publish their predictions on a website specifically established for such a purpose. Perhaps Consumers Reports might want to sponsor it.
Do this for five years in a row to reduce the "luck" factor.
Then publish the results for all to see, and figure out which group is really better at "projecting future performance." Results will talk and after five years, bullshit will walk. If I were a sabermetrician, I'd certainly welcome such a challenge, and in fact I'd likely be issuing it myself.
And, you're wrong. Your attitude is one all too common in these circles - "the only people who have a problem with advanced stats are too stupid to understand them", and that absolutely turns people off before the conversation can even begin. You know the old saw about attracting more bees with honey than vinegar? You're closer to napalm.
Agreed. I'm not very well versed in statistics, so I may not be stating this technically correctly. However I feel comfortable saying that Cliff Lee's performance was within the normal variance of expected outcomes.
Luck seems to be an incredibly poor word to describe his performance. In addition to Dewey's comments about bleeders or bloopers, I would also say that if Lee had been dialed in, and had been getting hit on pitcher's pitches all night, that would be "luck". But that wasn't what was happening.
And your point is?
Really, is there any evidence Lee pitched poorly? Was his velocity down? Location more? Less movement than usual?
Jolly Old Nick: are you willing to put money on the old guard? Because I would happily bet on the saberists. There are lots of predictions made, shouldn't be hard to do this.
That's fine - but there are other topics besides evaluation and projection and people drop "luck" into those discussion indiscriminately.
TDF: I guess this must be one of those "situations." I suggested two reasons people might not embrace advanced metrics. And actually I left off the most important, which must be simple lack of interest. But nowhere did I say this was a question of the average fan's intelligence. Anyway, I don't have a carefully-developed theory about why every fan has not embraced advanced statistics. But I find the increasing tendency of some to focus on the linguistic or stylistic failings of sabermetricians as a major part of the explanation to be wildly implausible, as well as often unpleasant.
Let's step back here. MGL writes a post to clarify that when he uses the word "luck" he means two things: what most fans thing of as luck (ball hits pebble) and also random variation of player performance. His whole point is to say "when I say 'luck' I also mean things that many other people don't intuitively consider to be luck." And then several posters here criticize him by saying in essence "those are two different things!" Well, yes -- that's the whole point of the post!
He threw a higher percentage of balls than he typically does (71-66 percent).
Not at all, and in fact if I had to bet, I'd probably go with the sabermetricians in such a contest.
But like any good sabermetrician (even though I'm not one myself), I'd still like to see some tangible proof that the sabermetricians can back up their claims with mathematically certifiable results, over a reasonably long-range period of five years.
There are lots of predictions made, shouldn't be hard to do this.
I wouldn't think so, either, which is why I threw out the wish. I'm just too lazy to do it myself.
I noticed this too - this was a post to fix confusion problems. But I think most of the posters disagree with this solution (continuing to use "luck" but explaining it).
It's like calling all motorized, wheeled vehicles "cars". Yea, if an outsider squinted real hard and used their imagination they'd understand what you were talking about if you called a bus "car" but that's not reasonable; it's reasonable to call it a bus. It's even less reasonable to complain when an outsider doesn't understand.
There are always going to be a lot of idiots who won't accept sabermetrics no matter how many gobs of sugar and cream are heaped on top of them. But there are also a lot of people who are undecided, and using less technical jargon to reach them seems to me to be only a matter of common sense. Bill James has certainly come to understand this.
Yes, that's my position. Rather than keep using the same word that you know means something different to outsiders (and have to re-explain that position again and again as new people are introduced to sabermetrics), I'd rather see the community simply modify its language when appropriate. You can keep banging your head against the wall if you'd like, but at some point you've got to realize the resulting headache is not the wall's fault.
I think Bill James always understood that. 'Runs Created', 'Win Shares', even 'Value Approximation Method' are all pretty clearly titled, even if you don't know the formulae. 'Wins Above Replacement' is less so, but still pretty easy. WPA/LI, UZR or WOWY are not so clear, though.
The problem has arisen in explaining how we go from 'a to z'. Some of Tango's stuff is hard going, because he doesn't bother to explain clearly what one should do if you want to utilize his conclusion in such a way that someone relatively undereducated mathematically can understand.
Good lord. He's explaining how HE uses the word. He's not asking anyone else to use it, and I can't imagine that he would care if someone else prefers "random variation in performance." He might even agree to say that instead if someone could demonstrate it would facilitate communication.
But can we please stop pretending that just using a different word will make a big difference? The real problem is that most people want to attribute predictive power to short-term performance differences. They believe that if Lee had kept pitching he would have kept getting hammered, even though that probably isn't true. They believe (like LaRussa) that individual hitter-pitcher matchup histories tell us something important. They believe some hitters perform better in the clutch. And on and on. That's the larger context here. And most of the resistance to the idea that short-term performance differences are in fact random will not go away just because you stop calling it "luck."
the title is "what do sabermetricians mean by luck" so he is clearly talking about more than just himself.
It would make a big difference. People always respond better to complex concepts phrased in a familiar way than phrased in jargon or a personal vocabulary. I speak as someone who has experience of trying to convey complex concepts to 'weakly educated' readers and undergraduates.
The problem isn't understanding, the problem is expectations.
I think Lee would have kept getting hammered, because he himself might be suffering from 'I can't get anything right tonight' mentality. You start overthinking your mechanics, rather than doing things naturally. However, there's every chance he'll come back and shut the Giants out next time.
That's hilarious.
There have been countless digressions in threads like these over the years over the use of the word luck, taking discussions away from the merits and into linguistics. That its curtailment would facilitate better communication is beyond any possible doubt.
But he's not interested in that. As mgl just said, "I get to define my own definition of “luck...” And I did."
That's his attitude. Fine. But he's going to get justifiably criticized for it.
I don't think so. I think that most people, in talking about Cliff Lee's performance last night, aren't the least bit interested in its "predictive power". As Dewey says in #27 here, you should never talk about individual games or individual observations in a discussion of "predictive power". Most people, I think, who talk about a single-game performance are simply interested in understanding the specific causes of what happened last night. I think that comments like #35 and #39 in this thread are far more useful and interesting to most people than a dismissal of Lee's performance last night as "bad luck".
His command was mediocre. He frequently missed his spots and left some balls in the middle of the plate. He didn't do that against the Yankees. I don't think his performance necessarily means a lot going foward, but he didn't pitch particularly well last night.
Whether you buy that or not with respect to abbreviations, it sure as hell seems like a sensible argument when the "abbreviation" you came up is not an abbreviation, but rather a word that actually exists, and has a very similar but not identical meaning to the concept you're trying to express...
So, I guess the answer is "no?" :>) Also, the fact that he threw 66% strikes, as opposed to 69% in his Cy Young season, I think tells us very little. But maybe the pitch f/x data will reveal he really didn't have good stuff and/or command. That would be interesting.
That's all fine, Kiko. But MGL's post just used the comments made about this game as a taking off point to talk about variance/luck more broadly. Most of the discussion about how saberists use "luck" are in discussions of concepts like DIPS or clutch hitting -- they aren't usually about a single game.
Digging into Lee's performance is a fine exercise. But to me the interesting distinction is between actionable information and the rest. If Lee was throwing to the right spots but just missing more than usual (assuming he did), then he should stick with his approach. If he had a good mix of pitches and locations but SF hitters guessed right a lot, then again, there's nothing to change. Whether we call all of that luck or something else, I don't personall care -- it doesn't change anything for Lee. But if the analysis can identify something Lee was doing wrong, that's interesting. As for trying to draw the line between "luck" and "performance variation," I don't know how you would begin to do that or why it would be interesting.
Now THAT is hilarious. MGL clearly defines how he is using a word, and is "justifiably criticized" for that because it doesn't advance the cause of educating the masses about sabermetrics? Who says that is his cause? (I've never heard him say that.) You're projecting your goal onto him, and then criticizing him for not pursuing it the way you want him to. I don't see how this is "justifiable."
And note the final irony here: apparently, you want MGL to use his time and talents to persuade fans of something, but think he could do it more effectively. And you think the best way to get him to adopt your preferred strategy is through sharply-worded criticism? I don't think that recommends you very highly as a dispenser of communications advice!
Shorter Andy: I was just blowing smoke.
This is a good point.
Ultimately, it doesn't matter for Nate's analysis whether Harry Reid or Sharron Angle wins next Tuesday. Nate's looking for larger trends, and the result of a single election doesn't tell you very much (although it might or might not tell Nevada pollsters something). That election result is a data point, one of many data points that he will collect next Tuesday and compare to the polling.
But if Reid wins next Tuesday, I don't think that Silver would call it "luck". His win would be based on receiving more votes than Angle or any other candidate. Silver is always careful to say that if his model doesn't account for something, that's a shortcoming of the model. Perhaps an unavoidable shortcoming (if you could account for everything, you'd no longer be talking about probabilities, but certainties), but a shortcoming nonetheless.
I don't think this is simply a rhetorical issue ("luck" vs. "deviation"). It goes back to what you're trying to accomplish. Saying, "I'm not trying to predict how Cliff Lee will perform on any given night, and my model doesn't work that way" is just pointing out where your model stops working, and that you need to look elsewhere for those answers. I don't think anyone would have a problem with that.
Saying, "It was luck" is something else entirely. It's essentially saying, "my model is great. It's that messy reality that's broken." That's not going to go down smoothly for a lot of people.
I wish saberists would give it up on "luck". It's a very stupid battle on their end. In the words of W. C. Fields, "If at first you don't succeed, try, try again. Then quit. No use being a damn fool about it." Communication is a two way street, in fact, if you can't get your point across then it is incumbent on you to change how you are presenting your message. Luck is simply a bad choice of words. "Random", while probably not an exact word in probability world, would work quite well, especially with the now-20-somethings using it all the time. Cliff Lee throwing a clunker LAST NIGHT is a random event while at the same time, it is expected to happen from time to time. People can understand this.
We don't get to choose how we define words. That's not the way communication works. I can't say I define "jackass" as "a good guy," turn around and say, "You know that GuyM is a real jackass," and then wonder why you misunderstood me. If mgl is not interested in actually communicating fine. But some of us are.
I don't give a damn how mgl uses his time and talents. I know he could communicate more effectively, and he could certainly do it more effectively by following the criticism of me and others. Whether he wants to is up to him. But criticism is the chief currency around here, and one that mgl has dispensed rather freely both here when he was a regular and on his site. I'm not sure why you think this gentle flower should be immune from it, or unable to benefit from it.
I'm really glad he put this in. Back to the original topic..
Perhaps the term "luck" was used loosely. There are many inputs to a pitcher's performance that cannot be observed well and quantified (upset stomach, kid failed algebra,just can't concentrate well today,...). I think this is what MGL is referring to as luck. Using an example from thermodynamics, there are too many molecules of air in a room to describe the speed of all of them, so we use "temperature" and "Maxwell distribution" to describe an approximation to the speeds of the molecules. So it is with a players performance, one can't be shocked to observe a hot player get cold. After all, luck is involved.
Nothing in the original post or in this thread discusses any of that. MGL defines "luck" as he uses it - last night was something other than an average outing for Lee; that's what saber-whatevers mean by "luck". He didn't talk about matchups, or clutch, or anything else; never once does he talk or infer or even hint at prediction. Nor has anyone, anywhere in this thread, talked about those things. We've criticized MGL for one thing, and one thing only - redefining a common word with a single common connotation, which confuses the issue he and other saber-whatevers are trying to address.
No, what those of us on this side of the discussion want is MGL and those like him to use words that mean what everyone understands them to mean.
Not at all, and in fact if I had to bet, I'd probably go with the sabermetricians in such a contest.
But like any good sabermetrician (even though I'm not one myself), I'd still like to see some tangible proof that the sabermetricians can back up their claims with mathematically certifiable results, over a reasonably long-range period of five years.
Shorter Andy: I was just blowing smoke.
Ray, what in the fuck is your point? Is it so completely beyond the scope of your limited imagination that some people might want to learn something that they don't already know? I realize that your certainty to curiosity ratio runs something like 500 to 1, but that doesn't mean that everyone else suffers from that affliction.
I DON'T KNOW---and neither do you---just how "predictive" these sabermetric "projections" really are. In part that's due to my not paying enough attention, but it's even more due to the fact that there isn't any one website or print format where past "projections" can be compared to actual results, and then compared to the "predictions" of non-sabermetric writers.
I would assume---and I'd imagine you would, too---that Dan's and Tango's "projections" would outperform the "predictions" of someone like Boswell or Gammons over a five year period. But I also think it'd be kind of fun to see if this could be confirmed. What's the downside to this? (Though I'm sure you'll think of something.)
Outside of things like radioactive decay or positions of electrons in an atom, is anything really random? Almost everything that appears to be random is just a weak signal with a lot of noise.
I don't think people would be any more inclined to embrace sabermetrics if the word luck was replaced with the word noise.
You can't even go two consecutive sentences without completely contradicting yourself, but others are supposed to take communications advice from you? Good luck with that....
MGL has said very clearly what he means when he uses the word. So he's met his main obligation: you can read MGL from now on with a perfect undertanding of what he's saying (or don't read him if that's your preference). Now some people think it would be better if he used a different word. Fine. Then the rational thing to do is try to persuade him that he could, for example, reach a larger audience if he changed his vocabulary -- and he can then accept or reject this speculative advice. What is completely irrational is to attack him for his choice. This whole process of projecting onto MGL certain responsibilities and then criticizing him for not properly discharging them is, well, pretty weird. Seems like something that needs more psychological than sabermetric exploration.
I don't think that finding this magic word, by the way, is so easy. People who don't like "luck" also won't like "chance." And I'm not sure they will like "random variation" a whole lot more, as most of them won't believe it's "random."
I don't think this is simply a rhetorical issue ("luck" vs. "deviation"). It goes back to what you're trying to accomplish. Saying, "I'm not trying to predict how Cliff Lee will perform on any given night, and my model doesn't work that way" is just pointing out where your model stops working, and that you need to look elsewhere for those answers. I don't think anyone would have a problem with that.
Saying, "It was luck" is something else entirely. It's essentially saying, "my model is great. It's that messy reality that's broken." That's not going to go down smoothly for a lot of people.
I think this is a good summary.
If Reid was to win b/c a giant snowstorm paralyzed the rural areas of the state, that would be "luck". If he wins b/c Democratic turnout is higher than expected, that's just reality being different than the model.
It's a complete misunderstanding of normal variation in performance to call last night's outing "luck".
Otherwise, why didn't he call the 8 shutout innings against the Yankees "luck"? That's almost as far from his expected performance.
When mgl used to post here, some would ask why he was so caustic, when the same point could be made without calling people idiots. His response was something to the effect of hasn't he done enough that he shouldn't have to worry about adjusting his tone?
He's been resistant to adjust, no matter the approach used. Us talking about it isn't going to change anything, but then again, the vast, vast majority of things discussed here won't change anything either.
You can't even go two consecutive sentences without completely contradicting yourself, but others are supposed to take communications advice from you? Good luck with that....
I can see why you don't think mgl needs communication advice...
I didn't contradict myself.
I ultimately don't care how mgl uses his time. It won't cause me consternation if he fails to change his behavior (in fact, I'd have to be a pretty big dumbass to believe that after all this time, mgl gives a rat's ass what anyone thinks of his communication skills). That doesn't mean I must refrain from commenting on them.
If mgl wants to heed advice, fine. If not, that's fine too.
Hell, it's not even "extremely wide". It's nonsensical. It has nothing to do with "luck" whatsoever, as that term is used by everyone in every other discipline, ever. It is sabermetric fiction, bad and clunky sabermetric fiction, and it should be openly and pointedly derided as such. Use luck to denote things that are luck, and use something else to denote swings in observed performance.
Bouton's Fetish in #31 hits it precisely. The pitches that Torres and then Sanchez hit for doubles, I wouldn't expect my fresh-up-from-AA lefty to make those.
Really, is there any evidence Lee pitched poorly? Was his velocity down? Location more? Less movement than usual?
If you're asking if Lee pitched poorly last night, you really should get your nose out of the books and watch a game sometime. He wasn't appallingly bad - and Lincecum wasn't very good either, by the way - but, yes, he pitched poorly.
Results do matter. Lee's line is not the line of a guy who was dealing. Dan made good observations above.
Did you see the game? (Seriously, sometimes people miss games. I mean, it was Game 1 of the World Series and this is a baseball board, but, perhaps you really didn't see the game).
Oh, hang on there buddy. By #######, I mean an accomplished, clever analyst of baseball who just doesn't care to be all warm and fuzzy when he debates. No need to get angry.
Words have meanings. If you use words others use but use them differently, you will be misunderstood. Shakespeare gets to co-opt language. The rest of us don't.
The current understanding is that there is randomness to every physical quantity. There is uncertainty to the position of a cinder block in a wall, it's just a rather small uncertainty. More at issue here is that there are too many factors that cannot be observed well acting on a baseball score: pebbles, wind gusts, indigestion, a nightmare. These factors "average out over the long term", supposedly (but never certainly!).
It's also got more than a bit of the "heads I win, tails you lose" in it.
---"If my projections pan out, I'm brilliant, but if they don't, I was done in by an unlucky roll of the dice.
"And if your non-sabermetric predictions proved better than mine---well, so what? There was no real predictive value in your predictions!"
A snarky summary, perhaps, but not altogether inaccurate in some cases. And it's why I'd like to see the same empirical methods used to test the predictive value of sabermetric projections as the sabermetricians use in making those projections. What is there to lose?
It's the same reason I shy away from evaluating teams on a "process-oriented" basis, as opposed to the dreaded "result-oriented" basis. Because then the question becomes, how do you evaluate the process? For most people, all it means is that the person in charge of the process made decisions we like, whether or not those decisions worked out. If you evaluate the process rather than the results, you can never be wrong.
The sabermetric prognosticators have been doing that for years while comparing the results of their systems.
What do you mean by 'randomness' here? Do you just mean indeterminacy? Because surely that's not what people normally mean by random; who the next chess world champion is going to be is certainly indeterminate as of now, but surely it would be strange to call it 'random'.
Pardon the screaming, but exactly how thick are you?
What is that way?
I've always thought chance would be a much better choice in most of these instances, while being equally easy to type and read. It conveys exactly what we want to with luck, without any of the baggage.
In sports parlance, luck isn't just the explanation we give to events that we don't know the cause of. For as long as kids have been playing sports, luck is the explanation given to events when we don't want to acknowledge the real reason is true talent. "You got lucky," has been spit out for generations in an attempt to downplay the opposition's victory. It was a bad idea to use this word from the start. It's simply pointless stubborness to continue to fight this battle when it's clearly led to confusion and has derailed many a reader from the point the author was trying to make.
I can walk in front of oncoming traffic and safely reach the other side once the cars swerve and stop short to avoid me, and the result was great -- I reached the other side. The process? Not so much.
When we're dealing with probabilities, as Dan said above, a 1 in 3 chance is not 0 in 3.
The sabermetric prognosticators have been doing that for years while comparing the results of their systems.
I realize that this has probably been the case, which is why I acknowledged upthread (#59) that I could have been paying closer attention.
But that said, I still think that for the purposes of reaching out beyond the sabermetric community, it would be interesting to have one central location (preferably a website) where those results could be compared side-by-side, along with a parallel set of predictions and results coming from non-sabermetric writers, preferably from the MSM. Sort of like a PolitiFact for baseball projectors and prognosticators.
The upside of this wouldn't be hard to visualize, especially if you could talk some prominent MSM critics of sabermetrics to take part:** Free publicity for sabermetrics at two visible times of the season (Opening Day and the interim just before the playoffs); and with the likely result that the sabermetricians would win such a competition, perhaps even a bit more respect for them from the MSM, which tend to be interested more in the Bottom Line results ("Whose predictions actually came out better?") than they are in the abstract reasoning behind it.
**maybe by challenging their manhood
Many of those things are understood by the general public as "randomness", not "luck". If MGL said "Lee had a bad night", no one would disagree. If he'd said "Lee's pitched better in the post-season than he really is, but last night was worse than he really is" and explained why, that's a good place to start a discussion. But by saying "Lee's performance was luck", he's introduced a word that everyone thinks is one thing but he means as something else; further, that one thing everyone thinks makes it sound like Lee cannot control the outcome of the way he pitches, which goes against everything everyone thinks they know. After all, if pitching is "luck", how can there be good and bad pitchers?
"Luck" is a dismissive word with negative connotations ("I'd rather be lucky than good"); to use it in evaluation of performance is derogatory.
Good point. By 'randomness' I meant the uncertainty associated with the Heisenberg uncertainty principle (http://en.wikipedia.org/wiki/Uncertainty_principle).
There is a separate set of issues dealt with in probability that have to do with too many, or too difficult to measure, factors affecting the outcome of an event. If you grab marbles from a bag that has black and white marbles, you can look in your hand to count 3 black marbles and rest assured there is no quantum mechanics related error in your counting. The next chess champion has mostly to do with who is a really great chess player, and then with other factors that are too difficult to quantify like who might get sick or hit by a bus. Maybe you would like to call these other factors random events, in which case there is some randomness associated with who the next chess champ is.
This is a really long topic, and I'm a poor writer, so please don't be offended if I don't reply.
The Book.com has been comparing "saber" results side by side for years, along with The Monkey (a simple regression model which does pretty well). I think they used to include some MSM, but those "experts" were so far behind that there was no point in publishing them anymore.
That's good to know, bads85, but this is what that TheBook.com website looks like these days. Not much about baseball that I can see. If you can link to any current website that displays this sort of comparative information, I'd appreciate it.
I can't imagine that MGL or any other sabermetrician would have a problem using "chance" to describe variation from true talent. However, I think that removes 5% of the resistance to the underlying concept, if you're lucky (in the "correct" sense). I think most of the fans who reject the idea that pitchers who give up a lot of hits on BIP in a season are mostly "unlucky" would feel the same way about the claim it was just "chance."
I'd like to help you out here, but it seems to require pharmaceutical expertise I just don't have.....
This is not what MGL means when he says "luck".
There is a blog called Vegas Watch that does such a thing.
MGL's definition of "luck" is a little too broad for me. For the totality of everything he describes in his post, I would use the term "baseball".
Even if you're right and the percentage is that low, then what reason would there be for continued use of the less effective word?
He's also not using it in the right part of speech, that jackass! Clearly, your definition says that luck is an intransitive verb, but despite using the word over and over, MGL /never/ uses 'luck' as a verb in that whole article!
Snark aside, since you endorse Merriam-Webster, let's go to the definitions for the noun version of 'luck' on that website.
1.a : a force that brings good fortune or adversity
b : the events or circumstances that operate for or against an individual
2: favoring chance; also : success <had great luck growing orchids>
from here.
Um, this actually isn't what I expected to happen. I guess I think MGL's definition is really very close indeed to 1.b. MGL says that luck is the variation around what a player 'should' do in terms of true talent, and I guess I think that's really pretty close to 'the events or circumstances that operate for or against an individual.' Merriam-Webster's luck are the circumstances that cause MGL's luck, and I guess I think that's a pretty close association.
There is a blog called Vegas Watch that does such a thing.
I went to that site and found a link to this, but that's not exactly what I was looking for, as you can see. Couldn't find anything else relating to baseball there, though perhaps I could if I were to search their entire archive.
There is a fangraphs article up which shows clearly that Lee was not cutting the corners of the zone. Now, like a coin flip - a good or bad pitching performance is not truly random, just unpredictable.
You can probably reasonably predict that Lee will pitch some shutouts and get lit up some fraction of the time he takes the mound. This probably as a little to do with his opponent (as mentioned above, the Giants are not a team to take-and-rake, although they are not as hacktastic as in prior years). He did not have great command, left some pitches in the zone and they got hit hard. I am not going to bother to call up his bb-ref page, but I presume he does this occasionally.
The luck part is that he did it last night.
Note that Lincecum wasn't much better - he was not really commanding his breaking stuff and had to come in and was getting hits. HE was lucky because Lee was a little worse/Giants a little better AND his relief didn't give a 3-run dinger to Juan Uribe.
Of course, the Uribe HR was NOT luck, it was crap pitching. O'Day threw 2 out of the zone (and when Uribe doesn't hack at them...), and had to come in, and threw a meat ball.
There are two hypotheses: (1) #### happens, or (2) #### like this doesn't happen unless Cliff Lee has a bad back.
If (1), he'll come out for Game Five (if necessary, and I sure as heck hope it's necessary because I have tickets) and give the Rangers a good chance to win, with a Game Score close to his median for the season and postseason: 72, which is pretty good.
If (2), that's bad luck in the cosmic sense, but also means that in Game One he wasn't pitching within his normal range of variation; something was wrong that neither MGL nor ordinary users of English would attribute to "luck," and would account in part for him "missing" his pitches right over the plate and seeing them driven to the outfield.
Edit: I dunno if it's Cokeworthy, but zenbitz did a great job of elaborating on hypothesis (1) while I was typing :)
Here's a 2009 version from Vegas Watch. If you Google "site:vegaswatch.net pecota", more results come up.
He's actually trying to bridge the gap, explain why a sabermetrician's "luck" isn't what people commonly think of the term. I think he does a very good job of it here, and should be commended for taking the time and effort to do it.
I also agree completely with those who suggest, effectively, if it ain't luck then don't call it luck. Still, I don't think MGL was advocating that the world should adapt to his definition of it. (I'm pretty sure he was joking in his followup comment that he gets to define luck.) It's also clear that he didn't recommend a solution, nor was he required to. As much as I'd like sabermetricians to adopt a better term than luck for what is not generally called luck, I don't know why any of us should have expected MGL to solve that problem here. Maybe that's the next step.
Lastly, I read MGL's post before I came here to read the comments. I actually thought that, if we were already at 90+ posts, surely we'd gotten Backlasher back. This subject was one trigger of his rants. In his rather strict view, nothing is random; it's just unmeasured. He (and I) objected to sabermetricians dismissing the difficult-to-measure effects and the measureable-but-I-just-didn't-do-it effects as the same useless randomness. BL went further than I would, essentially saying the difficult-to-measure effects (such as the effects of wind currents on a pitch as it approaches Barry Bonds' bat) are not luck; by that measure, nothing is luck. My stance is more aligned with the last paragraph of #54 - see here - than with BL, but I knew where he was coming from.
That's great. Too bad building a championship baseball team isn't as simple as crossing the street.
No reason. I'm happy to use "chance." I just don't get the view that MGL must use the word that people here want him to -- when I'm not even sure he shares the goal of convincing skeptics to believe in advanced metrics -- much less the anger at his choosing otherwise. He's been using the word for years, and here he takes the time to explain very clearly what he means by it. For that he gets "deserved criticism?" Just a deeply weird perspective in my view.....
I have never seen him do this and I've read probably 60% of what he has published online.
While I believe that mgl would agree that he's not interested in convincing sceptics, I'm not sure he (or anyone who spends so much time attacking other people) really means it. It's a sort of corollary to Dr Johnson's: 'No-one but a fool writes, except for money.' What's the point of all the effort and energy he's put into developing, marketing and even selling something like UZR if he is not trying to convince even the small audience whose duty is to be sceptical of spending a team's money on an unproven system of performance analysis?
The anger is down to his sense of hyperbole. Not everyone gets it. Plenty of people maintain they dislike it. And yet he carries on with it and if anything has made it more intense over the years.
It's like the 'friend' who always says something nasty 'as a joke', then thinks 'sorry' puts it right. Eventually it gets tiresome, and you avoid their company.
I have never seen him do this and I've read probably 60% of what he has published online.
Sample.
You must be Registered and Logged In to post comments.
<< Back to main