What do x,y,z,a,b,c represent, BL? the games played?


The known information that you will use in your test, e.g.

x= OBP
y=ERA
etc.

One of the problems is that there is not sufficient instrumentation to capture all the relevant output.

Ruth, Wagner, Foxx, Munson, et al were alive today and agreed to an ordinal list of the top 10 teams, I think I would weight their opinions higher

I'm not sure how much weight I'd give their opinion. I'd give it more than your run of the mill fanboy, or faux baseball historian, but I'd try to find some measure if such a thing were possible.

Why is a season of 35-5 and 61-61 worse than 16-12, 15-11, 16-11, 15-12, 16-11, 16-11?

It depends on if anything is distinguishable about those intervals, and what it is you are deciding to measure. If you go 35-5, and then have injuries to your best players and finish 61-61; you still have injured players come playoff time, I don't like your odds as much as some triangle theorem might state. I'm not sure we disagree on that point at all.

The major problem with some of the fanboy thought is a presumption that all events are equal and probability of outcome is the same at all times. They have trouble comprehending the dimension of time.

the very small sample of games played.

What are you watching in October? I see full games being played. I haven't seen a sample of a game being played. More important I see series being played.

With difficulty, as everyone else does.

So what do you do? I don't think everyone has "difficulty". Is there something you do to adjust your triangle formula? In specific terms, how do you account for the varying contextual elements that take place at different times?

I'll acknowledge the counterexamples, certainly. But anecdotal counterexamples to every general rule abound. They don't invalidate the rule at all.


What rule. If a condition exists that makes a statement false, it is false. That is not a matter of opinion.

I'll acknowledge the counterexamples, certainly. But anecdotal counterexamples to every general rule abound. They don't invalidate the rule at all.


So if one team is five games ahead with seven to play, loses 7 straight, while the other team sweeps the week, its the first 155 games that matter?

This is kind of silly. In most years, very many divisional titles will be won by a handful of games. So how on earth do you determine "best" Is it just triangle theorems?

Should we comb back through history and find some THIRD place teams from the regular season and claim they were actually the BEST or greates team becuase of their triangle record. The regular season is an interval contest to have qualifiers for the postseason tournmanet. There are winners and losers based on set criteria in both cases.

The likelihood of equally critical injuries to competing teams is also far greater, the longer the span of time. One need not be a mathmetician to perceive this.


One need not use flowery language to know that the impact of the injury is based on the percentage of time there is lost or diminished performance.

One should have some logical capability to know that injuries are causal and not random, and if you project them using a randomized function, then there is equipotential at any point in time. If you do more discerning analysis, you would have to show why the probability and impact will lower the goal maximizing utility.

Luck is just part of the game, period.

Then prove it, offer evidence of this causal force. At least try to explain it. Saying the same fifty times doesn't make it more true. Saying it emphatically doesn't make it more true.

I'd call the Giants' ability to tie the Dodgers that year every bit as "lucky" as the placement of that Willie McCovey line drive.


Bingo. Consistency in definition is key.

Depth is a critical element of team strength.

Over 162 games, yes. Over a championship series, less so. Why is the former inherently better than the latter. Its just a different game. I could make every team play triple headers every day for 300 games. that would impact final standings. You want to take one type of game and make it superior to all others. AND THAT IS NOT THE GAME THAT THE TEAM IS ULTIMATELY STRIVING TOWARD.

To be continued

such as the degree to which the team's key players appeared to be playing significantly above or below their likely "true" talent.


Huh. How do you know their "true" talent? What is even your definition. If you use the same definition that the saberists are using then one cannot play above their "true" talent.

Yeah, they were red hot.

The Cardinals were never red hot? Why are you preferring an interval or a metric where someone can not maximize output and win, over an interval where its nearly impossible to not maximize output and win?

I think I know now. I'm surprised I didn't see this before. Its because you are a FAN of a team that has virtually no postseason success. Andy nailed it.

Nobody's "system" can determine the best team, MLB's postseason tournament included.


You still haven't answered this basic question: What is the "best" team. And if you can't do it, if you refuse to even try to do it, why are you so adamant about things that are provably false?

The problem is your throwing out terms you have heard others say. You like how they sound so you adopt them. But you misapply the terms and use them inconsistently. If "luck" is just bad fortune, it applies to all bad fortune. If the test is "impact of an event" it must be used on all events. I'm not sure anyone even can discern what you mean by "best team".

A sample is a subset of a universe. You aren't even defining your Universe when you throw sample around. Because if you did, I doubt any sample, whether random, selected or cherry picked would be indicive of what you want to measure (which you also choose to define).

You want to talk about "samples of games", ok, I'll play along. If I randomly select one inning from every game next year, can you accurately predict who will win?

Of course you can't. The reason is that the context of the game changes so much. Likewise, you can't take 162 games and determine who is the "greatest" Because you are defining greatest with other ambiguous and unknowable terms, such as "true talent" You can't get that from a career output. Plus all of this is time dependent. A team's greatness as you would define it varies at each moment in time.

However, you can choose teams that had remarkable achievements, etc. You want to set up some construct whose only rules are your proclamations, even when they are inconsistent. And you want to promote yourself to some level of expertise that leaves such proclamations not subject to any criticality.

The words "sample", "luck" and "great" change definitions for you from post to post.

In the postseason, the loss of your best pitcher for a ten game stretch costs you 4 starts out of a maximum of 19 games, or more than 20% of the playoffs (and that assumes every series runs to the maximum number of games).

The injury is no more or less significant in terms of its effect on the player, but in terms of the impact in percentage of games, it's much greater in the playoffs.



Yeah, if you assume the injury occurs on the first day of the playoffs. And if you realize that time is a little different in the playoffs because of more days off.

It's a big part of why the postseason isn't played as single elimination games.


That and the fact that we won't to measure a pitching staff rather than each team's best pitcher, and the fact of the influence pitchers have on a game. Then its just a contest of who has Sandy Koufax.

Unless the outcome of a single baseball game is dominated by luck more than by skill, a smaller sample has to, by definition, suffer from luck having a greater impact.

Define "luck"? And that explanation is just wrong. The longer the trial, the more likely the tested quality will manifest itself in its true proportion, IF AND ONLY IF, the length of the trial does not effect that attribute.

So if players were dice that do not have wear and tear, you would be right. Otherwise, that is not true.

If what you want to test is between two teams, which would be the first to win four games; you need to run that test a million or so times to see the distribution. That is correct. Since you can't run it a million times without effecting the result, its not much of a question.

It was just another way of saying that it's a lot easier to replace a Mike McCormick than to replace a Sandy Koufax.

Yes, but your tortured point that Koufax's injury was the sole reason the Giants and Dodgers tied that year is shown to be empty simply by looking at McCormick, whose own injury may very well have cost the Giants more wins than Koufax's cost the Dodgers, or at least nearly as many. Your focus on one piece of the puzzle of both teams is myopic; your analysis of that season doesn't stand up to even mild scrutiny.

Here I'll quote my self-damning words:

The Giants' "ability to tie the Dodgers that year" was wholly due to the Koufax injury.

By which I meant that without that injury, the Giants would have been toast. And there's no doubt about this. Koufax may have missed only a third of the season, but he pitched fewer than 200 innings and his finger had been bothering him from late May on. By July he was allowing thirteen hits against the Mets---the 1962 Mets. This was not the real Sandy Koufax.

I'm not minimizing the loss of Mike McCormick, but he was no Sandy Koufax. Certainly not the early 1962 Sandy Koufax. And while the 1962 Giants were a great team (certainly better than the Yankees), I'd still maintain that they were lucky to tie the Dodgers.

IOW your generalization is interesting and valid, but not much use beyond that in trying to determining how luck influenced any given season.

Of course not. Yet it remains true that, in general, teenagers drive more recklessly than senior citizens, just as it remains true that, in general, short series results are influenced by luck to a greater degree than full season results.

Again, interesting, possibly true, but wholly irrelevant in deciding particular cases. The luck factor in a pennant race is not a legacy that can be handed down from year to year.

Or to put it another way: Past luck is NOT PREDICTIVE in anything but the abstract. The father's luck (good or bad) is not magically bestowed on his children. It is truly RANDOM in its pattern. Some teams can get a RANDOM sack of shlt dropped on their season's fortunes, and it's little consolation that it will all even out "in the long run."

The Giants didn't replace McCormick, though. They bought up a greenhorn Gaylord Perry and gave a few starts to Bob Bolin, but neither was particularly good.

If McCormick is healthy the entire season, he probably posts an ERA of around 3, and with the Giants offense, might have won 20 games.

Yes, but your tortured point that Koufax's injury was the sole reason the Giants and Dodgers tied that year is shown to be empty simply by looking at McCormick, whose own injury may very well have cost the Giants more wins than Koufax's cost the Dodgers, or at least nearly as many. Your focus on one piece of the puzzle of both teams is myopic; your analysis of that season doesn't stand up to even mild scrutiny.


LOL. Tortured Point. So it was the Giants ability to adapt that made them great, but it was bad luck when they lose in the post season.

Anyone who is ever confused about my points, look no further. I'd probably not even worry about the misuse of "luck" if it was used consistently. If it just meant an "external event", or a "low probability event from a pre-time instant", or if you used it to mean entropy in your projection.

But this is what you get. "Luck" is the REASON for bad things happening. "Skill" is the reason for GOOD things happening.

Come on Steve, Cowboy Up on this one. No imprecise language, tell us how you measure "great" Tell us what "Luck" means.

By which I meant that without that injury, the Giants would have been toast. And there's no doubt about this. Koufax may have missed only a third of the season, but he pitched fewer than 200 innings and his finger had been bothering him from late May on. By July he was allowing thirteen hits against the Mets---the 1962 Mets. This was not the real Sandy Koufax.

I'm not minimizing the loss of Mike McCormick, but he was no Sandy Koufax.

Also, to your meta-point it doesn't really matter. I can look in any event space and find "disastrous events" or "low probability events." I can find them in one of Steve's crapshoots.

What is deemed "luck" is just those things somebody wants to remove from the event space because they don't like the result, or don't like a future projection.

Its bad luck that Wohlers hung a breaking ball to Leyritz from my perspective. From yours, I bet it looked like bad luck that Pettite laid one up for Andruw Jones.

But to the loser one was "randomness" and the other was "skill". To the winner, who cares, we won.

and it's little consolation that it will all even out "in the long run."

It can't "even out" by definition that team can't play again.

But I mildly disagree on the Random part. I don't know that many things that are that "RANDOM" If some general manager is constantly "exploiting market inefficiencies" and getting players that injure themselves in bar fights or choking at opportune moments, I think its time to start wondering if there isn't a lack of a certain skill.

By which I meant that without that injury, the Giants would have been toast.

Of course they would have. But it's just as true to say that without the McCormick injury, the Dodgers would have been toast. The Koufax injury was not the only significant injury impacting the two teams that year, not the only manifestation of "luck" that impacted the race. It was a big deal, that essentially was "evened out" by the equally big deal of McCormick's injury. It's one of the many ways that the length of a full season does, in fact, often provide an "evening out" effect that a short series has far less capacity to provide.

And while the 1962 Giants were a great team (certainly better than the Yankees), I'd still maintain that they were lucky to tie the Dodgers.

I disagree. I think the '62 Giants and '62 Dodgers were about equally impacted by fortune. Koufax got hurt, but so did McCormick. Tommy Davis had a flukey power-production year for the Dodgers, but Chuck Hiller had a flukey batting-average year for the Giants. Wally Moon had an off-year for the Dodgers, but Stu Miller had an off-year for the Giants. Most everyone else on both rosters performed pretty close to a reasonable expectation.

If anything, the main factor of that season was that an obviously brain-damaged Alvin Dark gave Harvey Kuenn more than twice as many at-bats as Willie McCovey. Reverse that, and the Giants probably waltz away with the pennant -- although it's certainly the case that managerial performance is part of team quality.

The '62 Giants and '62 Dodgers were something very close to dead-even in quality. It's one of the reasons that was such a compelling pennant race.

Some teams can get a RANDOM sack of shlt dropped on their season's fortunes, and it's little consolation that it will all even out "in the long run."

Absolutely. But though that is rightfully little consolation, it's more than the zip consolation that when the random sack of doo-doo drops on you in a short series, the likelihood of it all evening out is very close to nil.

Funny thing about luck. Branch Rickey, a patron saint of many in the baseball blogosphere called it "... the residue of design."

Personally, I think that the truth lay somewhere between the Trederian and Backlasheresque positions, but what the heck do I know? I'm no Karl Popper or Bertrand Russell.

If McCormick is healthy the entire season, he probably posts an ERA of around 3, and with the Giants offense, might have won 20 games.

At the start of the 1962 season, McCormick was 42-42 lifetime, and had a 3.62 lifetime ERA.

If you can't win with an offense that sports Mays, Cepeda , the Alou brothers and McCovey, then maybe you just aren't that good.

McCormick had one big year in 1967 but otherwise was a pretty pedestrian pitcher, 134-128 and a career ERA+ of 95.

At the start of the 1962 season, McCormick was 42-42 lifetime, and had a 3.62 lifetime ERA.

This includes his age 17-20 seasons as a Bonus Baby kid developing without the benefit of minor league play, and thus niftily overlooks the fact that in 1960, McCormick was 15-12 and led the NL with a 2.70 ERA (129 ERA+), and in 1961, was 13-16 despite a 3.20 ERA (120 ERA+).

In 1962, the Giants got nothing resembling that caliber of performance from McCormick, instead getting 99 innings of ghastliness. After July, he was basically just dead weight at the back of the bullpen, appearing in only 9 games and posting a 7.71 ERA.

His sore-armed struggle of that season was a very impactful eventuality, every bit as impactful as Koufax's late-summer absence from the Dodgers, after pitching brilliantly through mid-July.

And Andy's assertion that Koufax had been struggling before he went down with his injury is vacuous. Koufax was 4-2 with a 1.23 ERA in June of '62 (24 hits allowed in 58 innings!), and 3-1 with a 1.07 ERA in his 4 July starts.

If you can't win with an offense that sports Mays, Cepeda , the Alou brothers and McCovey, then maybe you just aren't that good.

Which is every bit as true as saying that if you can't win with a pitching staff that sports Drysdale, Koufax, Podres, Perranoski, Roebuck, and Sherry, then maybe you just aren't that good.

In other words, it's vapid.

McCormick had one big year in 1967 but otherwise was a pretty pedestrian pitcher, 134-128 and a career ERA+ of 95.

Yes, and the huge depression in McCormick's career that began with his arm trouble of 1962 and lasted through 1964, is a huge explanation of that. The difference between the McCormick performance of 1961 and that of 1962 is enormous, and his aggregate career stats are irrelevant to that.

t the start of the 1962 season, McCormick was 42-42 lifetime, and had a 3.62 lifetime ERA.

That is Steve's evening out? Is this what this all about. There is malaise in the Bay so we have to blame it on something supernatural.

Which is every bit as true as saying that if you can't win with a pitching staff that sports Drysdale, Koufax, Podres, Perranoski, Roebuck, and Sherry, then maybe you just aren't that good.

Wasn't Andy saying they didn't have that lineup.

His sore-armed struggle of that season was a very impactful eventuality

Maybe the organization didn't realize they couldn't shred his arm? Maybe if he had the benefit of a better pitcher utilization model he would have been a dominant set up man? But it doesn't matter that much. The gods didn't come down and say, we'll make this guy sore arm. Actual causal events took place. They came down on Koufax too.

And that one year, the team the Giants could field performed better than the team the Dodgers could field.

So, I take it now, your definition of great means that you don't evaluate how well an organization uses its resources. We just choose which team has the roster that projects with the best rate stats per 162 games. That is your definition of great.

And whatever that median projection is, if someone actually plays better, then they are above their '"true" talent' and lower they are lower than their '"true" talent'. No matter what causes that performance, its luck, because all that matters is their per/162 game performance. Anything else is "luck".

Personally, I think that the truth lay somewhere between the Trederian and Backlasheresque positions, but what the heck do I know?

This actually answers one of Werr's questions. I have made no statement on who a "great" team is. In fact, I've mostly just talked about the misapplication and biassed application of terms. The Trederian position can be shown to facially false, and often invalid. Its not opinion, its decipherable. Yet, I am projected a position that is the negation of the Trederian position, and someone is willing to see merit in unsubstantiated pronouncements just because they are said over and over and over again. The only counter to this is more speech.

And I'm not the one throwing around "you must be joking" and "vapid" and what not.

Personally, I think that the truth lay somewhere between the Trederian and Backlasheresque positions, but what the heck do I know?

I should also add, you are giving me too much credit. Its strong silence and Andy that are making Steve spin around like a top, contradict himself, and show the lack of support for his premise. They are the one taking out the hooey. I'm like Maury, I'm just eating some popcorn.

Which is every bit as true as saying that if you can't win with a pitching staff that sports Drysdale, Koufax, Podres, Perranoski, Roebuck, and Sherry, then maybe you just aren't that good.

Steve, I think all those other guys were on the bench when Koufax was pitching, unless they were using Drysdale as a pinch hitter. The rest of the Dodgers pitching staff wasn't putting runs up on the board to help Koufax out. They weren't playing behind him.

Yes, and the huge depression in McCormick's career that began with his arm trouble of 1962 and lasted through 1964, is a huge explanation of that. The difference between the McCormick performance of 1961 and that of 1962 is enormous, and his aggregate career stats are irrelevant to that.

I remember doing a post about this, using the American League of the early-mid sixties. It is amazing looking back on that era, all the promising pitchers who ended up with dead arms-Hargan, Siebert, Tiant, Chance, Pasqual, Grant, John, Pizarro, Barber, Bunker, Aguirre, Bouton, Newman, Richert, Morehead, Radatz... I could name at least a dozen more.

McCormick is in the same boat. A young pitcher who was let loose and developed a dead arm by the age of 23. It was entirely predictable what happened to McCormick. It wasn't unlucky at all. It was inevitable.

Define "luck"? And that explanation is just wrong. The longer the trial, the more likely the tested quality will manifest itself in its true proportion, IF AND ONLY IF, the length of the trial does not effect that attribute.

So if players were dice that do not have wear and tear, you would be right. Otherwise, that is not true.

If what you want to test is between two teams, which would be the first to win four games; you need to run that test a million or so times to see the distribution. That is correct. Since you can't run it a million times without effecting the result, its not much of a question.

I would define luck as that the sum of the random factors that comprise an event, or the "wiggle" in combining two sets of skills. If there was no luck, the entity more suited to win a particular contest would win that particular contest 100% of the time. The weather, for example, is beyond the influence of the team. A batted ball that hits the base or an irregular piece of turf and bounces off at a strange angle. A fan like Steve Bartman or Jeffrey Maier.

You can't run a test with human beings a million times because they won't play the same way each time; that's true. But a good simulation can run a million times because the players in a sim season never get bored playing the same series ad nauseum.

We can never truly have the 1927 Yankees play a game against the 2001 Mariners unless we develop time travel. We can only approximate such a contest. To say that it will be imperfect because the simulation doesn't perfectly model every interaction goes without saying.

I don't think anyone seriously considers this exercise to produce the undisputed champion of all-time. However, there is something valuable in running a few types of simulations for the purpose of discussion, education, and perhaps even limited prediction.

There has been discussion about the Yankees remarkable postseason success in the late 90s coming from the composition of the team... front-loaded rotation, dominant closer. Similarly, I remember an article about the Braves and how they were generally less built for the postseason because of the percentage of payroll spent on starting pitching, for example, high-quality 4th or 5th starters become relatively unimportant in the postseason, but can get a significant amount of wins in the regular season.

It should be interesting to see if there is a pattern for regular season success and a different pattern for postseason success.

The rest of the Dodgers pitching staff wasn't putting runs up on the board to help Koufax out. They weren't playing behind him.

True, and completely irrelevant to any blind assertion that either the '62 Dodgers or '62 Giants "didn't deserve to win."

I remember doing a post about this, using the American League of the early-mid sixties. It is amazing looking back on that era, all the promising pitchers who ended up with dead arms-Hargan, Siebert, Tiant, Chance, Pasqual, Grant, John, Pizarro, Barber, Bunker, Aguirre, Bouton, Newman, Richert, Morehead, Radatz... I could name at least a dozen more.

And it was pointed out in that very thread that your post's myopic focus on the 1960s AL didn't comprehend that there wasn't anything particularly unusual about the 1960s AL in this regard.

McCormick is in the same boat. A young pitcher who was let loose and developed a dead arm by the age of 23. It was entirely predictable what happened to McCormick. It wasn't unlucky at all. It was inevitable.

If so, no more or less inevitable than the injury trouble that vexed Koufax in 1962 and later. Again, completely irrelevant to the issue of whether the 1962 Giants or 1962 Dodgers were any more or less victims or beneficiaries of outrageous fortune.

True, and completely irrelevant to any blind assertion that either the '62 Dodgers or '62 Giants "didn't deserve to win."


I wasn't making any assertion about who should win and who shouldn't. All I was saying is that McCormick was an average pitcher.

And it was pointed out in that very thread that your post's myopic focus on the 1960s AL didn't comprehend that there wasn't anything particularly unusual about the 1960s AL in this regard.

I don't remember but if it was, it was pointed out wrongly. You never see 20-21 year old pitchers throw 250 innings 2 years in a row anymore. That ended sometime in the late seventies or early eighties. Only the truly exceptional, like Gooden and Saberhagen, have been allowed to do that now.

And what's with the "myopic" comment? Did you want me do do a study going back to Abner Doubleday?

OK, let's make a list. Let's try to make a list of all the pitchers who threw at least 300 innings before the age of 23 in both leagues between 1962 and 1966 and see which ones didn't end up with sore arms within 2 years after that Iindicated by *:

McCormick
McCool
Cloninger*
Blasingame
Culp
Sadecki
Ellsworth*
Dierker
Fisher
Boswell
Bunker
Palmer
McNally*
Pappas*
McLain*
McDowell
Stafford
Newman
Chance
Lopez
Morehead
Hunter*
Draboswky


Of the ones with the asterisks, Cloninger Ellsworth and McLain all hurt their arms withiong 3 years. The only ones who seemed to survive intact were Hunter, Mcnally and Pappas.

I didn't really want to get in this argument, only to point out that McCromick wasn't that good.

Steve, without dragging this out until 2006, I guess I just can't make myself believe that with a healthy Koufax and a healthy McCormick, that the Dodgers wouldn't have won that year. Even granting my misextrapolation of the exact timing of Koufax's drop in effectiveness, which was in fact, as you say, about six weeks after he first noticed his finger going numb, which was in mid-May. McCormick was a fine young pitcher, but numbers like his (even granting that they would have equalled his 1960 and 1961 seasons) are certainly easier to replace than the sort of numbers Sandy Koufax was putting up before mid-July of 1962. And after mid-July, for the last two and a half months, Sandy Koufax's record was 0 and 2. Zero and two for the last ten+ weeks..

But here's what's even more interesting to me than our little Koufax-McCormick dispute: It's that with your argument, you're implicitly conceding my point, which is that the particulars of those random fingers of fate are what often determines the fate of a team in a pennant race.

Koufax out in mid-May: Giants win, no playoff.
Koufax out in mid-July: Dead heat.
Koufax out in mid-October: Dodgers World Champs (the Series was over by then, as you know).

OR, McCormick fully effective, entire season, Koufax out in mid-July: Giants win.

We only disagree about what would have happened if Koufax and McCormick had been healthy for the entire season. I'd take LA; you'd likely take the Jints.

But this isn't what the underlying argument has been about.

My underlying argument had been that random fate (call it luck, call it whatever you want) often determines a particular year's pennant winner.

In the long run, you say, luck evens out. And who can argue with that? Well, maybe Herb Score and the late 50's Indians, but that's for another day.

But the "long run" is often much longer than the course of a pennant race. Luck often doesn't always even out that quickly, even if sorting out luck's distribution over the course of a season isn't always as easy as it is in the postseason.

And you can't just dismiss this long-term luck as "the breaks of the game," as if it were some sort of a species separate from the sort of "luck" you seem to find in all sorts of postseasons. Luck is luck, and it can be just as determinative of a champion if it strikes in April as it does in October.

And unless you're espousing some sort of Seinfeldian "everything evens out" philosophy and applying it to the narrow confines of each and every close pennant race, including but not limited to the 1962 National League, then in fact you're conceding my point---which is that close pennant races are every bit as possibly determinable by luck as a best of seven (and even best of five) postseason series.

This is not to say that all close races are determined by luck, only that it's ludicrous to pretend that many of them aren't. And I've never maintained anything more than that.
You can check my past posts if you don't believe me.

And to say that these "exceptions don't alter the rule" is rather meaningless, since my argument has never been about the general rule, but rather about the necessity to examine each pennant race by itself and to examine the relative damaage of all those fingers of fickle fate, like Koufax's numbed finger or McCormick's sore arm.

Which I note you have done yourself, and rather eloquently, I might add. You didn't just assume that the luck of that 1962 NL race equalled out; you felt an obligation to present a factual argument for your side of the case---which presupposes that the argument might have gone in the other direction, does it not? That the luck did not "equal out."

But as always, I appreciate your forcing me to stay on my toes, and as a side benefit, re-appreciating the lost potential of Mike McCormick.

I would like to see the video of every game between the 62 Giants and Dodgers ... I rebel against the "noise" guys...I rebel against the "Luck" guys.......can't quantify it...just do...Games turn on real events...make the plays ...it is not easy...did your strike zone get bigger.. then.. in that situation....did it

Having said that I managed a 12 yr old Tournament team that lost a game in the bottom of the 7th inning on a wild pitch that hit the backstop went directly down to the ground and somehow spun and picked up velocity and ended up in the 3rd base coaches box.....the baserunner on 3rd scored...

I would define luck as that the sum of the random factors that comprise an event, or the "wiggle" in combining two sets of skills. If there was no luck, the entity more suited to win a particular contest would win that particular contest 100% of the time. The weather, for example, is beyond the influence of the team. A batted ball that hits the base or an irregular piece of turf and bounces off at a strange angle. A fan like Steve Bartman or Jeffrey Maier.

I have no idea what the "wiggle" is, but everything you describe is not "random"; many, but not all areexternal to your control. And your second statement is untrue as stated. If there were no external factors, then an outcome can be analytically predicted. But after the event space is closed it is true. The 1927 Yankees won the 1927 World Series 100% of the time.

Some of those things you describe are not external to the actors control, but are the output of both teams control. And the other external factors are part of the game. If it effects your team disproportionately, then that often is a flaw of your team. Weather being one of the big factors. In basketball, we use to have a mantra, "A good shooter must adjust" The same is true in baseball. That is one of the skills being tested. Its not hitting a baseball off a tee to an empty field.

Moreover, other than weather, I haven't seen you list anything that is not partially under the actors control. And even if you assert its not fair to penalize someone for not acting in accordance with an outcome (e.g. Jeffrey Maier) in a seven game series the impact of that event on the outcome is dispositive, but in the entire trial pretty neglible. The force of luck does not pick up a base and move it.

If you aren't good enough to overcome a neglible impact that is bounded by a few runs, do you deserve that title of "great." Now, I can't answer that because so far the only defintion I've seen of great is "what Steve proclaims" and Andy's test, which is based on testing of a team able to thrive in both game spaces. You don't meet Andy's test. To meet Steve's test, I think you have to play your home games in teh Bay Area.

You are assuming that your forecast at time t for event at time t + n is some sort of Platonic truth that never changes. Its not. As time approaches t + n, you can greatly increase the precision of your inference and narrow the distribution of your probability function. That' not "luck" at least not how you guys are defining it.

And there is no mulligans in this. Once the event space is closed, everything happened with 100% probability. Again, if the question you want to answer is: What happens when the 1927 Yanks play the 2005 White Sox under X set of criteria, then if you list all possible criteria, there is no randomness. You just solve the problem. If you want to don't have all the attributes of each team, or you want to add a randomizer, then you will get a probability distribution on the outcome. That distribution is going to be some aggregation of that which YOU CHOSE to treat as a random variable. And when that aggregation is based on teh current set of baseball knowledge, at best its only solvable in polynomial time. It may be intractable. Therefore, you can approximate that distribution by running a lot of simulations.

To develop plans its perfectly acceptable to say, I can't worry about this set of events, I'll just maximize this set of events. But that's not luck. That's a decision, which depending on the circumstances will have varying levels of accountability.

If so, no more or less inevitable than the injury trouble that vexed Koufax in 1962 and later.

That is the major problem. You really think that McCormick and Koufax innately have the exact same injury potential. Which is obviously not true, its a function of their physiology, utilization, AND EXECUTION at time t. You think some force comes down and makes it happen.

If there was no luck, the entity more suited to win a particular contest would win that particular contest 100% of the time.

On further reflection, I wish to expound more, since so many people think I'm discussing philosophy or polity instead of mathematical and definitional truisms. Your next paragraph, "human beings don't play the same way ..." is very indicative of the connotational problem with the above statement.

Your statement at any relevant point is belied by the fact that human performance makes the event a game.

I'm fairly interested, do you think golf is subject to your luck? And let's go to the other extreme, how about Vegas games?

In the former case, the game is almost entirely in the control of the actor. The only real externality is crowd noise and weather. The former is heavily controlled. The latter is considered part of the skill. If someone hits it in a bunker, do you say they are are unlucky. You know who is most likely to win going in a golf match, but they still have to perform to get the result. The human is not a robot, so the output varies. However, you can probably get a distribution function on the output.

No let's go to the other extreme. In most Vegas games, the game is set up so that most decisions are made prior to the acquisition of knowledge that you could use to predict the event. Its rules are based so that the output is not subject to individual skills. And if someone does develop a skill, the game is shutdown. Human performance and mechancial performance still control the output, but its heavily biased to create specific distributions.

Baseball is a middling game, where its set to have most information known to the participants except the other sides performance. Almost all outcomes are based on the intersection of performance between human actors. The forward distribution varies to the level of that human performance. The prior distribution is a result of that human performance.

The result is fixed. What you want to call luck is something that you didn't execute, didn't plan for, and didn't react to. I have not, nor will I ever, say that any one person is accountable for all unfortunate events. Nevertheless, because your model didn't predict something, it doesn't mean that you are not responsible for the result. The exception might be a Macha Man, who isn't allowed any authority in the model. He may be doomed no matter what he does.

Its not luck to have Rivera in your bullpen. Its not luck to have no hitter be able to make good contact on a cutter. Its not luck to get big hits.

Treder isn't even talking about a Jeff Maier, a situation that has happened maybe five times in 100 years. He's saying the fatigue of a Gordon; the injury to a Schilling; the homer by a Leyritz. He's describing these events as luck. I presume he thinks that Leyritz hits a homer 1 out of every 50 times at bat, and if you run that particular pitch 49 more times, he gets out. But that's not how it works. Leyritz probably hits that pitch for a homer a much smaller number of times. In that pressure Wohlers probably serves that pitch more than 1 out of every 50 times. Its those intersections on a micro level in that type of contest that make the difference. That is not "luck"

OR, McCormick fully effective, entire season, Koufax out in mid-July: Giants win.

I really don't see this. He was 13-16 the year before. It's likely he would have been a .500 pitcher again. I don't think him getting hurt changed things much one way or the other. If he hadn't gotten hurt, he would have taken innings from Bolin and Pierce, who were just as good if not better.

I wasn't making any assertion about who should win and who shouldn't.

Oh, please. In a discussion about why the Giants and Dodgers tied in the regular season in 1962, you insert this comment:

"If you can't win with an offense that sports Mays, Cepeda , the Alou brothers and McCovey, then maybe you just aren't that good."

And then you want to say that you weren't making any assertion about who should win and who shouldn't?

I really don't see this. He was 13-16 the year before. It's likely he would have been a .500 pitcher again. I don't think him getting hurt changed things much one way or the other.

What nonsense. He was 13-16 the year before -- with a 120 ERA+ in 250 innings. And then he sported a 70 ERA+ in 99 innings. The notion that this enormous reduction in contribution didn't change things much one way or the other for the Giants is ridiculous.

And then you want to say that you weren't making any assertion about who should win and who shouldn't?

Yeah, I think most people interpreted that to point out the ridiculous comparison of losing McCormick to losing Koufax. Rhetorically it was quite powerful too as it just ripped apart a point and showed the disingenuousness of the assertion.

One way you can tell is the post starts by showing the prior performance of McCormick.

I guess I just can't make myself believe that with a healthy Koufax and a healthy McCormick, that the Dodgers wouldn't have won that year.

Yes, but you continue to pretend that Koufax was injured all year. He wasn't. He had a hell of a good year for the Dodgers in 1962, going 14-7 in 184 innings, leading the league in ERA. He earned 15 Win Shares.

Meanwhile, McCormick's contribution to the Giants was very nearly negative. In 28 games and 99 innings, he earned a grand total of 1 Win Share.

The Dodgers had a healthy Koufax for two-thirds of the season. The Giants had a healthy McCormick for none of the season, and worse yet, they gave him 15 starts and 13 relief appearances in which he stunk up the ballpark.

The issue isn't that both pitchers were equally absent. Nothing close to that. Please stop pretending that it was.

We only disagree about what would have happened if Koufax and McCormick had been healthy for the entire season. I'd take LA; you'd likely take the Jints.

No, we don't. You made the ridiculous assertion that the only reason the Giants were able to tie the Dodgers for the pennant that year was because Koufax got hurt. I pointed out the obvious fact that the truth of the matter is far more complicated than that; that Koufax's injury was hardly the only factor that influenced the pennant race that year.

Koufax and McCormick were not equally hampered in 1962. McCormick's disablement was far more severe than Koufax's. Therefore, even though McCormick wasn't nearly as good as Koufax, the effect of his impairment -- worthless for the full season -- had something close to an equal effect on the Giants that Koufax's impairment -- brilliant for two-thirds of the season, absent otherwise -- had on the Dodgers. Please stop pretending it wasn't as it was.

In the long run, you say, luck evens out.

No, I don't. I do say that luck has a vastly better opportunity of evening out over a long time frame -- say, a full season -- than over a short time frame -- say, a week.

Luck is luck, and it can be just as determinative of a champion if it strikes in April as it does in October.

Yes, it can. But the likelihood of it actually doing so is far greater if it "strikes" in a short series in October than in a short series in April.

then in fact you're conceding my point---which is that close pennant races are every bit as possibly determinable by luck as a best of seven (and even best of five) postseason series.

Of course, close pennant races are. But that's completely different than your original notion that the quality of teams is revealed by their performance in short postseason series, more than by their performance over the full season.

And to say that these "exceptions don't alter the rule" is rather meaningless, since my argument has never been about the general rule, but rather about the necessity to examine each pennant race by itself and to examine the relative damaage of all those fingers of fickle fate, like Koufax's numbed finger or McCormick's sore arm.

No, it isn't meaningless in the least. The rule remains completely true. Your attempt to explain away the 1962 NL pennant race as being a function of luck is a failure, based on the facts of the matter.

Of course every pennant race and every postseason series must be examined on its own. But that doesn't for one instant change the always-relevant fact that sample size matters, and the smaller the sample of games, the greater the likelihood that the result will be impacted by random influences.

You didn't just assume that the luck of that 1962 NL race equalled out; you felt an obligation to present a factual argument for your side of the case---which presupposes that the argument might have gone in the other direction, does it not? That the luck did not "equal out."

Of course, it might have. But it factually didn't. Your attempt to invoke the 1962 NL pennant race as an example of one hugely impacted by the luck of one contender over the other has been demonstrated to be invalid. The 1962 NL pennant race, in fact, stands as a very good example instead of one in which the breaks that befell the two primary contenders actually did pretty much even out.

You've made my argument for me. I thank you for that.

BTW, in 1961, McCormick earned 17 Win Shares, Koufax 20. In 1960, McCormick earned 18, Koufax 9.

I have no idea what the "wiggle" is, but everything you describe is not "random"; many, but not all areexternal to your control. And your second statement is untrue as stated. If there were no external factors, then an outcome can be analytically predicted. But after the event space is closed it is true. The 1927 Yankees won the 1927 World Series 100% of the time.

As I'm sure you realize, the question isn't the patently obvious "what percentage of the time did the 1927 Yankees win the 1927 World Series?" but "If we could run X independent trials that model the 1927 World Series with an acceptable degree of accuracy, what percentage of the time would the 1927 Yankees win?" or "If we could place team A and team B who never played each other in a matchup, what percentage of the time would A win?"

A matchup between a good hitter and a good pitcher contains an element of luck as it often becomes akin to a game of rock-paper-scissors. If the pitcher throws what the batter is guessing on that pitch, he's a lot more likely to give up a hit. RPS is the classic example of a game in which there is not randomness in the individual components, yet the overall game becomes a random exercise (no strategy is superior to a purely random strategy, yet each player is completely in control of what is thrown by him or her). Baseball isn't nearly as simplistic as RPS, but there are often decisions that mirror that style of gaming (steal vs. wheel play; green light on 3-0 vs. strike down the middle; pitchout vs. stolen base; bunt vs. infielders charging).

If you aren't good enough to overcome a neglible impact that is bounded by a few runs, do you deserve that title of "great." Now, I can't answer that because so far the only defintion I've seen of great is "what Steve proclaims" and Andy's test, which is based on testing of a team able to thrive in both game spaces. You don't meet Andy's test. To meet Steve's test, I think you have to play your home games in teh Bay Area.

Andy and I are not particularly far apart. Both of us believe that the regular season and the postseason are factors to be considered in determining the "greatness" of a team. He may weigh the postseason more heavily than I do, but he has not said the worst WS winner is superior to every team that didn't win the WS, nor have I said that that the postseason is meaningless.

Again, if the question you want to answer is: What happens when the 1927 Yanks play the 2005 White Sox under X set of criteria, then if you list all possible criteria, there is no randomness.

Well, of course. Much like if we simulate a contest in which you and I play a series of rock-paper-scissors and part of the criteria of our simulation involves the precise symbols thrown in each match. There would be no confusion as to the outcome.

So if you don't like the term "luck," then call it "things that happen for reasons beyond the players' control."

If a meteorite falls out of the sky while I'm driving and hits the car next to me instead of my car, I'm comfortable saying that I was lucky that it didn't hit me. Even though in reality "luck" had nothing to do with it... the laws of physics didn't change on my behalf.

To take your argument to the logical conclusion, outside of the quantum world, there is no such thing as randomness at all. After all, dice move according to the laws of physics and under the same throwing and landing conditions, will always result in the same total. A deck of cards will always deal according to its prearranged order.

Such statements are literally true, but are valueless in calculating the odds at the craps or blackjack table until we become so proficient in tracking the events that we are like unto gods. The results of such skills would be indistiguishable from mystical prognostication.

strong silence- The playoffs are not, by definition, championship Major League Baseball games. Their statistics are not part of the official record tally. They are an exhibition that people care about and the players try to win, but an exhibition nonetheless.

The only reason we have a tiered tournament postseason in baseball is because the owners believe that system will provide the most profit. If it was determined that having every team play a 175 game season would be more lucrative teams would do that and suddenly you and your ilk would be without their absolute determination of team quality. Whether you'd switch to "best regular season record" or actually confront the horrors of an uncertain world I can't say.

Their are a number of non-MLB playoff tournaments. If a country that has only 5 MLB caliber players on their roster wins the World Baseball Classic, will you believe that team to be the best baseball team in the world?

And hey, since Backlasher and his team are coming dangerously close to Maynardism in this thread, I'll whip out one of his dandies with a minor twist:
Why is it so important to you to rank one team above all others?


Oh, everyone seems to be into poker analogies these days: Is there anyone who thinks the winner of the World Series Main Event is the best poker player in the world?

you and your ilk

Why the venom? Have I done something? Is it the tone of my writing? What?

If I had known that my observance of Koufax's birthday would have become ar argument about 1962, I wouldn't have mentioned it.

Steve, injuries aren't caused by luck.

Steve, luck often doesn't even out over the course of a season. Your system for determing the best team, most wins in the regular season, should account for the differences in luck over a season. And the impact of luck.

Baseball isn't nearly as simplistic as RPS, but there are often decisions that mirror that style of gaming (steal vs. wheel play; green light on 3-0 vs. strike down the middle; pitchout vs. stolen base; bunt vs. infielders charging).

So what Scioscia and Torre are doing when they have their hands in their pockets is playing a game of RPS!

Crosby, these decisions aren't random and they aren't lucky. It could be a hunch, which isn't luck. Managers have information and make decisions with that information.

So if you don't like the term "luck," then call it "things that happen for reasons beyond the players' control."

As I understand BL, a team still has the ability to win despite those types of things. Their response to these things gives us information about their ability.

Why is it so important to you to rank one team above all others?

Ben, mostly because it's fun. Its 600+ posts of fun. For some reason, I'm a sucker for Top 10 Lists. I'm like millions of people.

That is the major problem. You really think that McCormick and Koufax innately have the exact same injury potential. Which is obviously not true, its a function of their physiology, utilization, AND EXECUTION at time t. You think some force comes down and makes it happen.

Steve, since you apparently have decided not to talk to BL, would you respond to this? It's BL's point and expresses it better than I could. I'm not trying to do anything except get your view on whether injuries are determined by luck.

What you want to call luck is something that you didn't execute, didn't plan for, and didn't react to. I have not, nor will I ever, say that any one person is accountable for all unfortunate events. Nevertheless, because your model didn't predict something, it doesn't mean that you are not responsible for the result.

This is the heart of the issue and it's well said.

It changes the debate to those things which are not luck which could be planned for, executed better and reacted to. Now that would be an interesting discussion!

strong silence Posted: December 31, 2005 at 08:00 AM (#1800981)

That Your Any system for determing the best team, most wins in the regular season, should account for the differences in luck over a season. And the impact of luck.

To which I would only re-emphasize, "One season at a time, please."

And all the generalities in the world about luck "evening out in the long run," however mathematically sound, won't tell us a damn thing about how luck will impact next year's regular season---either the extent of it, or which teams it will help or hurt.

Oh, please. In a discussion about why the Giants and Dodgers tied in the regular season in 1962, you insert this comment:

And then you want to say that you weren't making any assertion about who should win and who shouldn't?

That was in reference to McCormick, Steve. And it was in response to something Flynn wrote. Please read my posts more carefully.


What nonsense. He was 13-16 the year before -- with a 120 ERA+ in 250 innings. And then he sported a 70 ERA+ in 99 innings.

Why is it nonsense? McCormick won 20 games once in his career, and that was a fluke. His next highest total after 1961 was 12 wins, and that occurred 6 years later.

It's really very simple. The Giants burned out McCormick by allowing him to throw nearly 1000 major league innings before the age of 23. Of course he got hurt. McCormick was done by the age of 30.

These days, teams have made all kinds of adaptations to limit the damage of their young pitchers. They start them in the majors as relievers, they went to 5 man rotations, they have extended bullpens, they put pitchers on pitch counts, they limit the innings they throw in the minors. You think these changes all happened for spurious reasons?

What happened to McCormick was not only predictable, it was inevitable.

Steve, injuries aren't caused by luck.

They aren't "caused" by luck. But to the extent that they're unpredictable, and beyond the control of players and teams, they represent one of the elements of whatever term you prefer as it pertains to the performance of teams: luck, randomness, chance, fortune.

Steve, luck often doesn't even out over the course of a season.

Of course it might not. But it also might. The point isn't that it necessarily does, but rather that the possibility of it evening out increases over a longer time frame -- thus the possibility of luck evening out over a full season is much higher than over a short series.

Your system for determing the best team, most wins in the regular season, should account for the differences in luck over a season. And the impact of luck.

I don't pretend to have any such "system."

And all the generalities in the world about luck "evening out in the long run," however mathematically sound, won't tell us a damn thing about how luck will impact next year's regular season---either the extent of it, or which teams it will help or hurt.

That's unquestionably true. Asserting it, however, does nothing to change the fact that the likelihood of luck evening out increases over a longer time frame.

Why is it nonsense? McCormick won 20 games once in his career, and that was a fluke. His next highest total after 1961 was 12 wins, and that occurred 6 years later.

It's nonsense to consider only a pitcher's win total in assessing his effectiveness and contribution.

It's nonsense to look at a pitcher who'd put together back to back seasons of 250+ innings, with ERA+ figures of 129 and 120, being named to the All-Star team both seasons, who then is able to contribute only 99 innings of extremely ineffective work, and say, as you do:

"I don't think him getting hurt changed things much one way or the other."

Of course it changed things, significantly.

You assert:

"If he hadn't gotten hurt, he would have taken innings from Bolin and Pierce, who were just as good if not better."

In the first place, neither Bolin nor Pierce were "just as good if not better." Respectively they put up ERA+ performances of 108 and 105 in 1962, significantly less good than McCormick had fashioned in both '60 and '61. Moreover, practically the impact of McCormick's struggles in 1962 were more likely:

- It forced Billy O'Dell into an extremely heavy workload, probably diminishing his rate stats

- It forced Pierce and Bolin to start more games than they otherwise would have, meaning they weren't available in the bullpen as much as they otherwise would have been

- Which forced Miller, Larsen, and Duffalo to work more games/innings than they otherwise would have, and none was especially effective in 1962

These days, teams have made all kinds of adaptations to limit the damage of their young pitchers. They start them in the majors as relievers, they went to 5 man rotations, they have extended bullpens, they put pitchers on pitch counts, they limit the innings they throw in the minors. You think these changes all happened for spurious reasons?

Of course not. As you well know, I've written voluminously about all of these historical developments.

What happened to McCormick was not only predictable, it was inevitable.

I disagree that it was "inevitable." If it were, then none of the many cases of pitchers with extremely heavy workloads in their early 20s who didn't break down at age 23 -- Drysdale, Seaver, Pappas, Dierker, Hunter, Sutton, etc. etc. etc. -- would have taken place.

But let's say, if not "inevitable," that McCormick's injury problems were clearly not surprising. I'll heartily agree to that. But to that degree, neither were the injury problems that Koufax encountered. And most definitely, the injury problems of neither McCormick nor Koufax were completely predictable down to the moment they would occur. Neither the Giants nor the Dodgers knew, at the start of the 1962 season, that either or both would encounter injury problems that season, or how severe they would be if they did encounter them. We know it in retrospect, but they couldn't have known it before the fact.

Therefore to point to Koufax's injury as bad luck, and dismiss McCormick's as inevitable is, yes, nonsense.

Moreover, practically the impact of McCormick's struggles in 1962 were more likely:

And I forgot to mention this one, the most obvious and direct impact:

- It injected 99 innings of extremely ineffective pitching into the Giants' 1962 season.

Oh, and one other thing.

At the start of the 1962 season, McCormick was 42-42 lifetime, and had a 3.62 lifetime ERA.

He was 54-54, not 42-42.

In the first place, neither Bolin nor Pierce were "just as good if not better." Respectively they put up ERA+ performances of 108 and 105 in 1962, significantly less good than McCormick had fashioned in both '60 and '61.

But 60 and 61 were the best years of McCormick's career, ERA-wise. He never did that well either before or after. What you are saying is that you would have expected McCormick to sustain the pinnacle of his career for another year. I think that is unlikely.

My bad about the record miscount but it doesn't change the conclusion drawn from it that McCormick was an average pitcher.

And I don't know why youy say that Mccormick was better than Pierce and Bolin. Maybe Pierce was near the end but he was still a good pitcher. And Bolin was just coming into his own. Bolin put up a 120 ERA+ in 1961 too. His 105 in 1962 is right about where I would guess McCormick would check in at.

But 60 and 61 were the best years of McCormick's career, ERA-wise. He never did that well either before or after.

Well, the fact that he didn't do as well after is kind of the whole point here, isn't it. And before 1960, McCormick wasn't yet old enough to vote; it really wasn't too shocking to see him developing beyond his performance through 1959.

What you are saying is that you would have expected McCormick to sustain the pinnacle of his career for another year. I think that is unlikely.

We only know it was the pinnacle in retrospect. Nobody could have known it at the time. What I'm saying is that following back-to-back years of 253 and 250 innings, with ERA+ of 129 and 120, then a 99-inning, 70 ERA+ performance was just a wee bit of a letdown.

McCormick was an average pitcher.

His aggregate career stats are very close to league-average, yes. But "average pitchers" don't rack up 262 lifetime decisions, and pitch over 2,300 lifetime innings.

More to the point, McCormick most definitely was not an "average pitcher" in 1960 and 1961. He was an All-Star both seasons, among the better starting pitchers in baseball. He was then terrible in 1962. Aggregating his performance across all three seasons isn't relevant or useful to the discussion.

And I don't know why youy say that Mccormick was better than Pierce and Bolin.

Only because he was, in both 1960 and 1961, and thus could reasonably have been expected to be, by the Giants and everyone else, in 1962. No reason other than that.

Maybe Pierce was near the end but he was still a good pitcher.

Yes, he was still a good pitcher. But his performance in 1962 wasn't as good as McCormick's had been in either 1960 or 1961, on the basis of either innings pitched or ERA or ERA+. Pierce's last season at that level of performance had been 1958.

And Bolin was just coming into his own. Bolin put up a 120 ERA+ in 1961 too.

He was a good young pitcher, and he had put up a 120 ERA+ in 1961 -- in 48 innings of short relief. Just a bit of a difference between that, and a 120 ERA+ in 250 innings as a starter.

His 105 in 1962 is right about where I would guess McCormick would check in at.

A 105 ERA+ in 1962 wouldn't be an unlikely performance from a healthy McCormick, sure. It would be a comedown from his '60 and '61 years, but not a dramatic one. But here's the thing, again: Bolin put up that 105 ERA+ in 1962 in 92 innings, as a long reliever and spot starter. A healthy McCormick in 1962 would unquestionably have handled well over 200 innings, quite possibly over 250 again.

That kind of contribution, with an ERA+ in the neighborhood of 105, would represent vastly more value than the same ERA+ in 92 fill-in innings. A healthy McCormick in 1962, even if he hadn't been able to sustain his ERA+ performances of the previous two seasons, would have provided vastly more benefit to the Giants than the sore-armed McCormick actually did.

So what Scioscia and Torre are doing when they have their hands in their pockets is playing a game of RPS!

That's a great image.

Crosby, these decisions aren't random and they aren't lucky. It could be a hunch, which isn't luck. Managers have information and make decisions with that information.

The decisions on each side are not luck, just like if I played you in a series of rock paper scissors and anticipated your next play based on a pattern, it wouldn't be luck. The solution to these sorts of problems (where predictability would result in you opponent anticipating your strategy more accurately) is often to inject a pseudo-random element into your strategy.

Example in poker: Only playing monster hands is a recipe for never being called unless you are mathematically beat. Therefore, many good players add a weaker hand that they'll play maybe 25% of the time. The method of determining whether to play this hand could be based on the location of the second hand on a clock, or the suit of one of the cards in the flop, or something else. Note that this isn't precisely random (if my opponent knew how I was calculating it, he could anticipate whether it might be one of those bluffs), but assuming my opponent couldn't figure out my system, the effect is no different than purely random play.

Example in baseball: Joe Slugger swings at a 3-0 pitch 10% of the time, and takes 90% of the time, to prevent opposing pitchers from getting free strikes against him. He makes this decision based on selecting a number prior to each AB and then glancing at the ones digit of the minute... if it matches, he will swing 3-0 for this AB. The pitcher can make an educated guess that the hitter is likely not to swing, and even if he knows the exact probability and the proper weights of how to respond, he cannot predict what the hitter is doing, and it has the same effect as if the hitter is truly behaving in a random fashion.

In each of these cases, the ideal strategy for defying predictability (in a mathematical sense) is to inject a random element (in the sense that it is independent of the event in question) into play. In RPS, the best strategy against an opponent with no "tells" is to play in a purely random fashion. If you don't believe me, try playing RPS against a six sided die (assigning to two numbers to each possible throw). No strategy you can use will result in better than a 1/3 win, 1/3 loss, and 1/3 tie in the long run. All strategies are equally viable (or non-viable) against such an opponent.

Again, the individual components (my choice to bluff or not, my opponent's choice to call or not, the strike down the middle, and the swing/take decision) are determined by the actors so they could be said to be non-random or not "luck." But without the ability to crack the code, it behaves in a manner identically to simple random fluctuation, and can be modeled as such in a simulation. We are arguing semantics.

By the way, happy new year to everyone and their loved ones. I'm keeping this thread bookmarked and hope the interesting discussion continues into 2006.

By the way, happy new year to everyone and their loved ones.

I second that emotion!

Why the venom? Have I done something? Is it the tone of my writing? What?

That's an easy answer, but you won't believe me. Its because you have a minority opinion. Moreover, you have one an opinion that meets the three criteria for being insulted:

(1) Its logically complete.
(2) Is shows the flaws in the counter assertion.
(3) The counter assertion is part of others' core belief system.

I don't pretend to have any such "system."

Then why all the posts talking about 'best' teams? Why the original rebuke? Why did you say "luck...." almost verbatim almost ten times without providing any evidence? I'm legitimately curious about this, although I think I know the answer.

I'll play too. I didn't post when it was list making. If you wanted to directly compare qualities of the 27 Yankees with teh 95 Braves, it would have been something of minor interest to me. I posted after about the 10th time that assertions without evidence where made about this force of "luck".

So do you have a "system"? Do you want to assert who the X number of best teams are? Or did you just want to sermonize about "luck"?

I will tip my cap once though. When you forced the conversation into the 1962 Giants, you did start actually presenting evidence about their team. Its not particularlly persusive because I doubt many are going to believe that a loss of McCormick has the same impact as a loss of Koufax.

However, you did move from proclaiming to actually advocating with evidence. You also picked up the tone quite a bit, but I'm sure that no one would believe that you are uncivil and get one of those Furtado warnings.

Asserting it, however, does nothing to change the fact that the likelihood of luck evening out increases over a longer time frame.

You just kind of said that luck was an externality, and are applying it to injuries. So do you think that injury is equipotentail for all players, no matter how they are used? Do you think injuries will "even out" over a 162 game schedule. Do you think Jeffrey Maier will even out over 162 games. Personally, in the latter, I think it happens very, very rarely. Its more likely to happen in the playoffs, and its a function of among others the configuration of the park. But you think everybody is going to get one of these for a home run each year.

- It forced Billy O'Dell into an extremely heavy workload, probably diminishing his rate stats

I thought you believed that you should just pitch them till they dropped.

Of course not. As you well know, I've written voluminously about all of these historical developments.

You wrote three short articles on the Hardhat Times. All of them were immediately debunked.

That kind of contribution, with an ERA+ in the neighborhood of 105, would represent vastly more value than the same ERA+ in 92 fill-in innings.

I'll let you argue with Andy and Kevin about McCormick was as good as Koufax. I don't know the pitchers. But this I do know: performance from a staff is what counts. I'm sorry if McCormick was just a flash in the pan that didn't have what it takes. I'm sorry if Giants management overused him to injury. And I'm sure that if sabermetricians were actually around they would have falsely projected him to be a 250 inning, 120 ERA+ beast. And they would have been wrong. Because they would have failed to account for his injury performance. That isn't luck. This game isn't about coming up with a model that you read from Bill James, and declaring a team a victor or a player an all star before the season counts. Its about actual performance.

We are arguing semantics.

No we aren't because "But without the ability to crack the code, it behaves in a manner identically to simple random fluctuation, and can be modeled as such in a simulation." has already been said by those you are arguing against.

Its not semantics. Its the exact same thing of taking "low variance between all ML pitchers with more than 150 IP in their year-to-year BABIP" and using it to say:

(1) BABIP is not a skill.
(2) Pitchers have no effect on BABIP

and then starting to extrapolate that to:

(1) Batters have no effect on BABIP

and then start criticizing choices when skills for BABIP are used for selection. And culminating in discounting PAST PERFORMANCE based on a person achieving a good BABIP in rating said performance. Its something that I call "creeping fanboyism" And for somebody to punt out of the situation saying, "We really don't disagree" is just disengenious. Because people do disagree and are promoting things that are facially false.

You have one poster that is asserting:

(1) It evens out over 162 games.
(2) Injuries are random.
(3) Giving some teams credit for the best possible performance that could have occurred in the past, and sticking other teams with the most disastarous outcomes of the past
(4) Ignoring actual trials that provide real information
(5) Ignoring accomplishment

AND MOST IMPORTANT

Refusing to analyze the basis of accomplishment. If you can't "crack the code" then if you want to be an analyst, you should be working hard to "crack the code" And the last thing you should do is throw your hands up and say, "That was luck." Remember what direction you are going. This discussion isn't about moving forward; its about looking backward at what already occurred. And people want to ignore outcomes they don't like.

And its exasperating when you are very exact, and people still try to do a "gotcha" by showing an inconsistency that doesn't exist. If you are going to use DMB, WHICH USES FUNCTIONS DESIGNED TO RANDOMIZE, then you absolutely need to run it a bunch of times to get better information. And as has been expressly stated, the information will still show the bias of the simulator. But one thing it shouldn't do is have McCormick as good as Koufax. Because it will base its output on the actual performances of the players.

It will not account for any attribute of the "team" If there is an advantage to a manager decision, team psychology, etc. It won't be found in that simulation. It in effect will hold all those events equal, and randomize those things it does not have data for that you choose to randomize. Its going to take rosters, assume equipotential for all outcomes like injury, etc., have them managed the same way.

None of that, or the need to do that based on either expediency or lack of knowledge, is going to effect this discussion about "luck" How you model is your choice; but you should be accountable for its output. When your wrong, you should be looking to improve your model, not make excuses. And if the innate ability of another produces better decisions then you, well tough darts, (s)he beat you in that contest.

But how you project has nothing at all to do with how you evaluate past results. That is only true if you so much want your model to be correct, you will develop an almost religious like view of the world, so that it blinds you to real outcomes. If you really, really, really want to try to take your bias and push it into past results, you can do that too. There are plenty of algorithms that will fairly remove output based on some criteria with a specific distribution. None of them would just remove the post season. And more importantly, none of them would just remove events that were adverse to the team where you were a fan.

But you are seeing why this is not semantic. You have one poster here that is legitimately trying to use modelling techniques on past problems, and removal techniques to argue that McCormick was just a good a pitcher as Koufax. That will cherry pick an interval to make an argument that the Cardinals were a better team than the White Sox. That will say a championship run through the postseason was luck, but a good performance during a 162 or 155 game interval is skill.

That is why there is discussion on this. As I said, if people were just using "luck" to mean "externality" or just using luck to mean "entropy" or just using "luck" to mean "unquantified information" it wouldn't bother me in the least. Then it is semantics and linguistics. But that is not what is happening. People are using characteristics of "external" and characteristics of "entropy" and characteristics of "unknown" and lopping it into the word "luck"; giving it causal properties; and then applying it for a rhetorical purpose to advance an argument or hypothesis. Then when you show it doesn't apply, they become more rhetorical saying over an over again "luck is part of life" rather than discussing the causality of the events under discussion. I can't see how that leads to any rational investigation, selection, or honesty in discourse.

So its not semantic; its integral to any conversation where we are analyzing or hypothesizing about prior events.

Steve, luck often doesn't even out over the course of a season.

Of course it might not. But it also might. The point isn't that it necessarily does, but rather that the possibility of it evening out increases over a longer time frame -- thus the possibility of luck evening out over a full season is much higher than over a short series.

And all the generalities in the world about luck "evening out in the long run," however mathematically sound, won't tell us a damn thing about how luck will impact next year's regular season---either the extent of it, or which teams it will help or hurt.

That's unquestionably true. Asserting it, however, does nothing to change the fact that the likelihood of luck evening out increases over a longer time frame.

Two responses which more or less say the same thing. But when a weatherman says that there's a 10% chance of rain tomorrow, that isn't much consolation when you leave the house without an umbrella and get drenched. And it doesn't matter whether you call it "luck" or "bad weather forecasting." You're still drenched, even as you think, "well, the next nine times he says that he'll probably be right."

Because although you know that a coin will come up heads about half the time over a thousand tosses, it does you no good to know that if you're betting on the outcome of any specific flip. You might run out of money before your luck "evens out," and the pennant race may be over, too.

And in comparing the strength of teams, the fact that luck evened out in pennant race #1 has nothing to do with whether it influenced pennant race #2, and awarding strength points to a lucky pennant winner (if it can be determined that luck gave them the pennant) makes no more sense than awarding them to a lucky postseason winner (ditto). Especially when the definition of short series "luck" begins to include things like "a pitcher got hot for three weeks," while not saying the same thing about a pitcher who gets hot in the last three weeks of a pennant race. A hot streak in September is no less "random" than a hot streak in October, and the close of the season gives the "unlucky" team no more chance to "even out" that luck than the end of the World Series.

I don't accept much of what Steve has said and attributed to fortune, but it does seem clear that from the perspective of March 1962, the loss of McCormick for the year (as represented by his earning 1 WS) was a worse loss than the loss of Koufax for two months. Even if you didn't consider McCormick to be a two time All Star coming off two well above average seasons and simply considered him to be a league average 100 era+ pitcher, his loss has to be worth more than the 7 or 8 WS the Dodgers lost with Koufax' injury. The reason this isn't intuitive is because, I believe, people overestimate the replaceability of 200 league average innings.

But when a weatherman says that there's a 10% chance of rain tomorrow, that isn't much consolation when you leave the house without an umbrella and get drenched.

Agreed. I don't get the concern with "consolation," however.

Because although you know that a coin will come up heads about half the time over a thousand tosses, it does you no good to know that if you're betting on the outcome of any specific flip. You might run out of money before your luck "evens out," and the pennant race may be over, too.

Agreed.

And in comparing the strength of teams, the fact that luck evened out in pennant race #1 has nothing to do with whether it influenced pennant race #2

Agreed.

awarding strength points to a lucky pennant winner (if it can be determined that luck gave them the pennant) makes no more sense than awarding them to a lucky postseason winner (ditto).

Agreed.

Especially when the definition of short series "luck" begins to include things like "a pitcher got hot for three weeks," while not saying the same thing about a pitcher who gets hot in the last three weeks of a pennant race.

Agreed.

A hot streak in September is no less "random" than a hot streak in October, and the close of the season gives the "unlucky" team no more chance to "even out" that luck than the end of the World Series.

Agreed. But here's the crucial distinction: the three weeks the pitcher got hot during the regular season represented a very small portion of that season. He may well have (probably did, in fact) had a cold three weeks at some other point. Seasons are compilations of the entire performance, hot and cold streaks included. When the hot and cold streaks occur matters little; every game all season long is counted equally.

If he gets hot for the three weeks in the postseason, that represents the entire postseason. Thus the impact of a three-week hot or cold streak from a key player has vastly more impact on his team's postseason performance than it has on his team's regular season performance.

That's the issue. That's one of the essential reasons why the postseason is a far less reliable metric of a team's genuine quality than the regular season is. It's all about sample size.

t's all about sample size.

I'm sorry if you have issues with size, but a sample of what? If you want things to "even out" you have to show there is equipotential. There is not equipotential. Players aren't robots.

Most people that have had no statistical training view your "hot streak" as an indicator of quality or skill. Most people that have had analytical training view one large sack of information as being a mess. In fact, that one little normal distribution is an aggregation of many different distribution curves, each of which having different statistical properties and characteristics. They look for those other curves, because they convey far more information. And if those sets are too small, they increase the amount of information until they have something that generates a reliable inference, or an inference that has better weight over other sets of inferences.

People that have heard of statistics, just cherry pick information and then use statistical operators as a means of advocacy. The context of that "hot streak" is important. Its not an equipotential event that is just as likely to occur at any point in time. Players don't operate at peak efficiency, output is not always maximized. In fact, with the extra round of playoffs, I hypothesize managers and teams take even more steps to prepare to optimize output in the post season than they did in the past. They will rest players more often, take more medical steps etc. If that is true than the 162 game interval is less reliable now than in the past for obtaining most any relevant information, and the post season contains more information about the bounds of team output particularly against quality competition.

This isn't Strat-O-Matic. If someone learned about baseball by playing that game, then I could understand those gaps in thought. The outcome is determined by your dice roll. DMB isn't even that, its a much more sophisticated simulator, but its output is still less reliable than actual occurences.

But most importantly, if you are "Agreed" on most of those points that Andy made, you really shouldn't be disingenous by the way that you are throwing the word "luck" around. If your intereted in examining, comparing, or extracting useful information, examine the causes of events, don't just come into a thread sputtering "Luck this" and "luck that" and misuse terminology because it sounds good to express things in such fashions.

I think you will find that many people are able to discuss things at a meaningful level. The certainty with which you posit things doesn't relate to their veracity. And if you want to talk about "somebody joking" or