Stats vs. Traditional

[Chisox]

Sounds like you're closest to 75% stats/25% traditional

That's what I was thinking. But it's an overall aspect, and it's not accurate for any one aspect of the game.

[RuthMayBond]

This one is really difficult for me to answer.
I rely on stats 100% for offense, 100% on pitching (although I probably shouldn't, especially with the lack of readily available stats), and about .01% for defense.

Take that for what it's worth.

Sounds like you're closest to 75% stats/25% traditional

[Chisox]

This one is really difficult for me to answer.
I rely on stats 100% for offense, 100% on pitching (although I probably shouldn't, especially with the lack of readily available stats), and about .01% for defense.

Take that for what it's worth.

[Bill Burgess]

I doubt if anyone will ever truly be able to resolve these age-old discussions. And maybe that is not a bad thing.

Bill

[SHOELESSJOE3]

50/50 is the closest to describe it for me.

the numbers many times lead to stories and many stories include or lead to numbers.

they are inseparatable.

I did choose 75/25 because that was the closest one of the picks available on the board. I think where I actually come in is some where between 75/25 and 50/50 leaning a bit closer to that 75/25.

From the voting results those two seem to be the ones most choose.

[barzilla]

True Barzilla,

My comment is textured, but I don't really want to start explaining it by writing pages upon pages of minute clarifications of my point. Maybe I should write a book of my own.

Very fair, in fact my impetus for developing the formula was a book by James Vail called "Outrageous Fortune". Vail argued we could develop a baseline and completely take the voting out of the process. My index is a way to refine the debate instead of removing it. My formula has what I call a "subjective zone" where people can insert their own contextual evidence, eye witness evidence, other statistical evidence, or even "intangibles" as some have said.

However, the formula interjects my belief that these "other" criteria are used far too often and used to put people in that are nowhere near worthy. So, the index separates players into definitely out, definitely in, and the subjective zone.

As for your other question I do think people on both sides look for ways to lift themselves above the crowd. I would be remiss not to admit that sometimes it makes me feel better when I learn more. However, I would also say it is important (and all good sabermetricians believe this) that we understand that we can be wrong and often are.

[digglahhh]

True Barzilla,

My comment is textured, but I don't really want to start explaining it by writing pages upon pages of minute clarifications of my point. Maybe I should write a book of my own.

But can we at least agree that the total player metrics are the form of sabermetrics most frequently used irresponsibly? And if many begin there, as many do, are they not putting the cart before the horse in terms of those metrics reflecting the deeper relationships between numbers. Even if they are valid, isn't a good chunk of the value of them, then, knowing why they are valid.

The HOF index, would actually be something that I would not condemn as some stupid metric. The Hall of Fame is a real entitiy within baseball. Determining who is and who is not worthy is an important task. The ranking of players in its crude form, for its own sake is something that I don't even think GMs do. Additionally, your HOF index, I'm sure, no matter how good it is, can't be the whole of somebody's HOF argument, especially when you approach the bubble. There is so much context to the numbers that is lost. Again, like the Commish said, the greatest problem of all, is the feigned science behind it, the illusion, through complex mathematical procedures of unquestionable credibility.

I'm not talking to you right now, or anybody in specific, but I wonder how many who throw out these numbers could, if given raw data, produce statistically advanced arguments on their own. How many can really define a Z-score, or linear regression?

Further, the real discomforting question, how does our desire to use these esoteric formula relate to the varying degrees of intellectual, and baseball fan/historian related elitism we hold. Is it possible, that psychologically, part of this movement is to pronounce ourselves the authorities and to self-segregate in an ego-massaging and self-aggrandizing manner?

[barzilla]

Total player metrics are, IMO, the lowest, or least evolved form of sabermetrics. They are the designer T-shirts with the logo printed all over them, the ones that those seriously into fashion shun and never buy, but the favorite amongst insecure kids eager to impress their peers.

I hate to speak for Matt here, but I hope you are not boiling down PCA into some "stupid total player metric" and if you say it is the "lowest, least evolved form of sabermetrics" then you haven't seen anything he's done.

Personally, I've developed a metric called the Hall of Fame index which is a total player metric. Rife with holes? Perhaps. Useless? If I thought it was useless I wouldn't have spent six months studying, writing, and trying to get published (still waiting although I have an agent). Perhaps, I am just trying to impress my peers, after all, I want people to buy the book, but there is a greater goal there and I'm sure Matt would echo that sentiment about his research.

Making lists and such is about as far from what a good sabermetrician is looking for as you can get. The good ones are into finding out what makes the game tick, what situation provides the best run scoring, what type of hitter best takes advantage of certain chances, what is the best way to define a good fielder, etc. etc.

That would be true if this is all that total player metrics do. If I simply said Player A is the best and Player B is the second best then it would be elementary. Total player metrics attempt to say by how much. Are total player metrics perfect? Certainly not and I'm sure Matt would say the same. However, there is inherent value to them because they do attempt to answer those questions in a macro sort of way. At their heart, the empirical studies and results describe a relationship between hitting and fielding, OBP and SLG, or any other relationship a sabermetrician might look for.

[baseballPAP]

Total player metrics are, IMO, the lowest, or least evolved form of sabermetrics. They are the designer T-shirts with the logo printed all over them, the ones that those seriously into fashion shun and never buy, but the favorite amongst insecure kids eager to impress their peers.

Further, I think we have to ask the uncomfortable, and self-evaluating question of how worthwhile a pursuit, in terms of overall baseball knowledge, creating ranks and lists of the best players ever in numbered order, is in the first place. Is the best way to appreciate greatness to pit one form of it against another?

Now you've done it....
I actually agree with you

Making lists and such is about as far from what a good sabermetrician is looking for as you can get. The good ones are into finding out what makes the game tick, what situation provides the best run scoring, what type of hitter best takes advantage of certain chances, what is the best way to define a good fielder, etc. etc.

Making lists is just the way we pass our time. If no one ever goes out on a limb and puts forth an opinion on anything, then there is nothing to discuss. If no one ever says Billy Ballgame is better than Stanley Shortstop because...... then we're all just clicking away with nothing to do until someone comes up with something truly interesting to say.

Appreciating greatness in one form or another is the basis of all discussions here. Listing of players is the most fundamental form of this, and look at all the great discussions that have come from it.

[SABR Matt]

Actually the sway isn't toward traditional as defined by this poll.

Call 50/50 votes neutrally swayed...that leaves 15 statistical votes and 5 traditional votes.

[digglahhh]

Praise the Commish.

I think he would agree with me when I say that I have a deep appreciation and interest in the questions sabermetrics asks in regards to game theory and strategy. When is it productive, overall to bunt, is it ever? What is the probabilistic value of being ahead in the count, or hitting in specific situations. That stuff, I love.

Total player metrics are, IMO, the lowest, or least evolved form of sabermetrics. They are the designer T-shirts with the logo printed all over them, the ones that those seriously into fashion shun and never buy, but the favorite amongst insecure kids eager to impress their peers.

Further, I think we have to ask the uncomfortable, and self-evaluating question of how worthwhile a pursuit, in terms of overall baseball knowledge, creating ranks and lists of the best players ever in numbered order, is in the first place. Is the best way to appreciate greatness to pit one form of it against another?

[The Commissioner]

I can't even answer this question, because it's not a quantifiable topic.

First of all, we need to define being "pro-stats" and/or "anti-stats". I'm hugely into stats. However, for the most part I'm against most sabermetrics that I've read. Would that make me a stathead or an anti-stathead? Then again, I say that I'm against MOST (as in NOT ALL) sabermetrics. This appears to be quite a difficult concept for many to comprehend. (How can that be? Can you quantify to what exact degree you are into sabermetrics and what degree you are not? *sigh*)

Actually I would say that this poll itself is quite analogous to some of the problems that I have with most sabermetrics in general. This poll is designed to convey in quantifiable terms something that cannot be quantified. Let me amend that. If some system were invented that could read my brainwaves and analyze the results, then, perhaps theoretically my "stats approval rating" could be measured. However, until that time, any answer that I give would not be a true measure of what transpires in my brain. Yet my answer to this poll would give off the false impression of actually being able to gauge it. To me, this is the mistake that most sabermetrics (or perhaps just self-labelled "sabermaticians") that I have read, make. They assume to take into account all factors involved when for each factor that they integrate, they are leaving out several others which remain unaccounted for.

Where I see this as a problem over "traditional stats" is that I believe most people know and understand the limitations of these traditional stats. They get some of the flaws. Most die hard baseball fans know that Palmeiro's having hit more homeruns in his career than Foxx does make him the better homerun hitter of the two. However, I have seen way too many arguments where it is "proven" that one player was better because his "win shares" were 2 points above the other player or he is more "statistically similar" to another certain player. Until people learn to take sabermetrics with the same grains of salt that they do other numbers they will continue to muddle historical questions more than they resolve them.

[Ubiquitous]

But the same thing happens when anybody fills out a ranking system. One guy looks at a player and say he stole a lot of bases, another guy looks at the list and sees another guy who was playing everyday. One guy values the extra 20 stolen bases, one guy values the amount time playing.

So in the end they are similar. In order to trust a system or person one must know how the system works or what experience the person has in baseball. If some 6 year old ranks the top 100 players you really are not going to take it seriously, but if Mel Allen does one you are going to give it more weight, right Bill?

The same applies to metrics. If someone puts up a list of players based on HAA or Hoban points I'm not going to take it seriously. But if somebody presents me with a list based on linear weights with PBP data looking at peak and longevity in a detailed manner I'm going to give it more weight.

But again in the end the differences between most respectable metrics on offense are quite small. We are not talking about vast differences but that does remind me of another reason why rankings differ. Rankings differ because people rank them differently. Some people will rank them based on the total career number. Some people will rank based on a combination of career and peak. Some will throw in league quality, some will throw in lost time because of military or segregation, and so forth. For Total baseball ranks its players on a straight career total. Bill James does a combination, if we were to rank BPro using WARP3 we would get a ranking based on career total plus league quality thrown in there. Matt has a ranking system I believe that does a combination as well. So right there is another reason why metrics will have different rankings becuase the people behind them value different things. Just like people who do rankings without metrics value different things.

[Bill Burgess]

The problem with the different systems isn't really a problem....they're just slightly different. What makes the results so different is the players. There are a lot of factors that go into any ranking system. You have to be great in many areas to come out on top. Then you have the players who are great in one or two areas and good other places filling out 10-50 or so. Then once you hit 50 and up in the rankings, the method to make the list allows a multitude of ways to get there. In other words, the players are grouped A LOT closer together further down the list. So if one system values a stolen base slightly higher, then a player with 50 more steals might jump 15 places.

But aren't you making my argument? The various variables are what makes stats dissimilar. So trusting in one system, is to distrust another. It's the very differences which makes me not able to trust in one's results, when another system gives other results.

[baseballPAP]

Let's go slow. And I'm not trying to win the conversation as much as learn something here.

If we take Win Shares, and another stat system, take their top 100 lists, why shouldn't we come to at least similar results? Even if we make the exercise easier, and limit our conversation to hitting alone, which should be theoretically simpler, why shouldn't we get verifiable results?

That would make traditionalists feel a lot more comfortable with numbers. I bet if Chris the elder were to compare our top 100 players, we'd share a heavy consensus. And I bet if I and another traditionalist member were to do the same thing, we'd also have a heavy consensus.

Bill

The problem with the different systems isn't really a problem....they're just slightly different. What makes the results so different is the players. There are a lot of factors that go into any ranking system. You have to be great in many areas to come out on top. Then you have the players who are great in one or two areas and good other places filling out 10-50 or so. Then once you hit 50 and up in the rankings, the method to make the list allows a multitude of ways to get there. In other words, the players are grouped A LOT closer together further down the list. So if one system values a stolen base slightly higher, then a player with 50 more steals might jump 15 places.