Sabermetrics are a bunch of malarkey . . .

[The Wolf]

Originally Posted by BoofBonser26

Point valid, but I'm certain he was being satirical.

Oh, I know he was being satirical. I was just helping to point out how wrong the anti-statheads are.

[CBL-Commish]

You'd think that almost all fans of a simulation of baseball that's, at its core, a big number-cruncher might be accepting of newer statistical measures of production. Especially those like VORP that have been pretty well vetted. And those same folks might be a little skeptical of other simpler, context-ignorant measures defined in an off-hand way in 1910. But I guess if you thought that you'd be a little wrong.

[RonCo]

It's not their fault. Human beings are not born with a fundamental understanding of math, variance, and other forms of mathematical logic. They are also pre-conditioned to a certain ways of thinking, and want to believe in what their eyes see despite the numerous psychological studies that show that eye witnesses are often not trustworthy.

Example #2: Only in a world where human beings rely on their eyes over their brains can Derek Jeter be a Gold Glove winner...

Not to open another can of sabermetric worms.

[BoofBonser26]

Originally Posted by RonCo

Not to open another can of sabermetric worms.

Too late!

[The Wolf]

The problem isn't with people's eyes; the eyes (sensors) see just fine. The problem is in the sensory post-processing the brain does.

[RonCo]

Originally Posted by The Wolf

The problem isn't with people's eyes; the eyes (sensors) see just fine. The problem is in the sensory post-processing the brain does.

No reason to get personal. I never like it when someone tells me I'm wrong, but it hurts the most when they're right.

[ctorg]

This thread reminds me of a post I made some years back in which I pondered whether OOTP should strive to make the AI more realistic or more competitive. The implication is that behaving like a real manager/GM is not equivalent to maximizing competitiveness, at least not in OOTP (where "player personalities" consist of a few numbers tied to traits).

I think this is something the game continues to struggle with. It isn't really possible to have aspects of the game work both in the most sound way possible and in the most realistic way possible, simply because reality is not sound.

[Russ]

Originally Posted by andymac

They produced the hits, but the RBI's are a production of more than just that player. If player A can do the exact same thing as player B but you are going to consider player A more valuable because there were people on base when he did it and there weren't when player B did it then IMO you are missing something. To everyone their own though.

I'm on the sabr side of stuff, but this doesn't pass the common sense test. In fact, this is a shortcoming of a lot of the new metrics. I don't believe this was the goal of any of the designers.

Let's take a look at two players in an imaginary game:

Player A:

1st AB, bases loaded: K
2nd AB, bases empty: Home Run

Player B:

1st AB, bases loaded: Home Run
2nd AB, bases empty: K

Who's more valuable to his team this game, this series, this season? It's an underlying foundation of VORP to ignore situations because it's out of the player's control, but it also ignores wasted opportunities.

If it's the goal of a batter to not get out and if it's the goal of the offense of a team to score runs, then player B is more valuable to his team even if VORP is equal.

[Tony M]

Originally Posted by Russ

Let's take a look at two players in an imaginary game:

Player A:

1st AB, bases loaded: K
2nd AB, bases empty: Home Run

Player B:

1st AB, bases loaded: Home Run
2nd AB, bases empty: K

Who's more valuable to his team this game, this series, this season? It's an underlying foundation of VORP to ignore situations because it's out of the player's control, but it also ignores wasted opportunities.

Even the 4RBI v 1RBI comparison is situationally vague.

What if Player A's 2nd AB is in the bottom of the 9th with a tie-score, and player B's 1st AB is in the 7th inning with a 10 run lead?

Who then is more valuable? I know that sabremetrics don't deal with that either but then neither does just raw stats like that.

[RonCo]

Sabermetrics deals with these kinds of questions. It just doesn't use VORP to do so.

[Tony M]

Originally Posted by RonCo

Sabermetrics deals with these kinds of questions. It just doesn't use VORP to do so.

Is there a good site to look at for this sort of thing, because I don't think I've ever seen anywhere that has these sort of situational stats and a subsequent 'valuing' of them.

[sfeldkamp]

VORP is really more of a predictive stat than a descriptive stat. The purpose and utility of it is not so much to decide which player's had a better season, but to compare player's contributions isolated from his team-mates so you can decide which player might be better in another setting, say for instance, your own team.

You can definitely make the case that a player with higher RBI totals had a better season, even if you wouldn't say that he was necessarily a better hitter than another player with higher VORP. The first statement (a better season) is descriptive of an outcome while I think that the second statement (a better hitter) is predictive.

I really prefer WPA as a measure of a player's performance to RBI though. Too bad it's not in the game.

[sfeldkamp]

Originally Posted by redsoxford

Even the 4RBI v 1RBI comparison is situationally vague.

What if Player A's 2nd AB is in the bottom of the 9th with a tie-score, and player B's 1st AB is in the 7th inning with a 10 run lead?

Who then is more valuable? I know that sabremetrics don't deal with that either but then neither does just raw stats like that.

Actually it does. You are looking for WPA, Win Probability Added. It basically compares outcome of an at bat in the game situation (score, inning, out, runner's on base) with the historical outcome of the same situation. So if a team with a runner on third and two outs in the bottom of the eighth inning of a tie game resulted in that team winning 65% of the time (historically), and the player at bat hit a single, after which that team would (historically) win 75% of the time. The player is credited with a WPA added for the difference, and this is totalled over the course of the game and the season for each player.

http://en.wikipedia.org/wiki/Win_Probability_Added
http://www.fangraphs.com/ - one of my favorite sites graphs this stat and does so live. It's a great way to see the drama of a baseball game at a glance...

[ctorg]

I think most SABR stats are meant as predictors rather than descriptors. The overall general goal of sabermetrics, as far as it has one, has been to create a model that will predict what a player will do in the future. Not that every measure is aimed at this, but in general, sabermetrics came about as a way of beating other fantasy baseball players by figuring out a way to predict performance. Its biggest value is in its ability to say, "Yes, you should sign that guy" or "that's a bad trade" or something like that, not in its ability to look back at a season and say, "This was the best guy that year."

Now, there are some measures aimed at doing that, but the thing is, you can always make a good argument that opportunity for accomplishment is not relevant to the fact of accomplishment. IOW, it doesn't matter how much of a chance someone has to do something, only whether they did it or not. This doesn't work for predicting the future, but it works for rating the past, and that is where traditional stats people can legitimately argue that their way is better, although I still tend to favor sabermetrics personally.

It's a matter of philosophy rather than logic, in a way.

When we want to judge who the richest person is, we can simply look at who has the most money. When we want to judge who has the best ability to make more money, we need to set up theoretical predictive models. Traditional stats are the equivalent to figuring out who the richest person is. One doesn't take into account luck or opportunity or talent or anything. It's all about how much money there is. While it may hint at someone's ability to generate money, there is no way of seeing through the other factors.

So if you want the baseball equivalent of figuring out who the richest person is, you go with traditional stats. If you want the baseball equivalent of figuring out who is best at generating money, you go with sabermetrics.

It's not a perfect analogy (iie. the goal in baseball isn't to get the best stats, but to contribute to wins, and while these are related, they are not the same thing), but it's the best simple one I can think of.

[Jestre]

Originally Posted by ctorg

I think most SABR stats are meant as predictors rather than descriptors. The overall general goal of sabermetrics, as far as it has one, has been to create a model that will predict what a player will do in the future. Not that every measure is aimed at this, but in general, sabermetrics came about as a way of beating other fantasy baseball players by figuring out a way to predict performance. Its biggest value is in its ability to say, "Yes, you should sign that guy" or "that's a bad trade" or something like that, not in its ability to look back at a season and say, "This was the best guy that year."

This is precisely what we anti-VORP'ers have been trying to get across to the "eggheads" this entire thread. VORP is not a good tool to measure what a players contribution 'WAS' to his team over a past period. Only what his individual value was independant of team value and team contribution. In the real world the team with the most runs wins, not the highest VORP.

[The Wolf]

Originally Posted by ctorg

The overall general goal of sabermetrics, as far as it has one, has been to create a model that will predict what a player will do in the future.

From "The Sabermetric Manifesto" by David Grabiner

"The basic goal of sabermetrics is to evaluate a measure for a given purpose. The most common uses of statistics are to evaluate past performance (such as to determine who should win the MVP award) and to predict future performance (such as to evaluate a trade that was just made)."

Read the whole thing:

http://www.baseball1.com/bb-data/gra...manifesto.html

[The Wolf]

Originally Posted by Jestre

In the real world the team with the most runs wins, not the highest VORP.

And knowing who wins tells you nothing at all about the value of individual players, or which player is best at which position.

Winning is a goal. You can reach that goal by luck, having the best players, by outmanaging the opposition, or by some combination thereof. You can't control luck, management is a skill you can learn, but VORP will tell you who the best players are at each position.

And without tools like VORP you are just guessing. And if you go by RBI's then you're so far behind the curve that you're road kill for those using better stats.

Ever wonder why the Red Sox have done so much better (four playoff berths and two world championships in five years) since Bill James joined their staff? I don't, because it's obvious. Better stats = better player decisions = better players available.

[Jestre]

Originally Posted by The Wolf

And knowing who wins tells you nothing at all about the value of individual players, or which player is best at which position.

Winning is a goal. You can reach that goal by luck, having the best players, by outmanaging the opposition, or by some combination thereof. You can't control luck, management is a skill you can learn, but VORP will tell you who the best players are at each position.

And without tools like VORP you are just guessing. And if you go by RBI's then you're so far behind the curve that you're road kill for those using better stats.

Ever wonder why the Red Sox have done so much better (four playoff berths and two world championships in five years) since Bill James joined their staff? I don't, because it's obvious. Better stats = better player decisions = better players available.

Once again you are describing VORP as a tool to measure future production!!!! THAT IS WHAT WE ARE SAYING!!!!!

[BoofBonser26]

Originally Posted by ctorg

I think most SABR stats are meant as predictors rather than descriptors. The overall general goal of sabermetrics, as far as it has one, has been to create a model that will predict what a player will do in the future. Not that every measure is aimed at this, but in general, sabermetrics came about as a way of beating other fantasy baseball players by figuring out a way to predict performance. Its biggest value is in its ability to say, "Yes, you should sign that guy" or "that's a bad trade" or something like that, not in its ability to look back at a season and say, "This was the best guy that year."

No and yes. SABR stats are designed to do both.

Originally Posted by ctorg

Now, there are some measures aimed at doing that, but the thing is, you can always make a good argument that opportunity for accomplishment is not relevant to the fact of accomplishment. IOW, it doesn't matter how much of a chance someone has to do something, only whether they did it or not.

Correct. This is a basic principle of SABR stats.

Originally Posted by ctorg

This doesn't work for predicting the future, but it works for rating the past, and that is where traditional stats people can legitimately argue that their way is better, although I still tend to favor sabermetrics personally.

It's a matter of philosophy rather than logic, in a way.

No, it's not. Do you really simply want to say the guy with more RBIs was better without considering that stat is largely controlled by his team-mates? Really, until we get over this, we can't advance our understanding very far.

SABR stats will tell you what it was the player did. In other words, it will tell you which player was better, for his team, in real life. Not in a vacuum.

Casey Blake and Grady Sizemore had the same number of RBIs last year.

Originally Posted by ctorg

When we want to judge who the richest person is, we can simply look at who has the most money. When we want to judge who has the best ability to make more money, we need to set up theoretical predictive models. Traditional stats are the equivalent to figuring out who the richest person is. One doesn't take into account luck or opportunity or talent or anything. It's all about how much money there is. While it may hint at someone's ability to generate money, there is no way of seeing through the other factors.

So if you want the baseball equivalent of figuring out who the richest person is, you go with traditional stats. If you want the baseball equivalent of figuring out who is best at generating money, you go with sabermetrics.

It's not a perfect analogy (iie. the goal in baseball isn't to get the best stats, but to contribute to wins, and while these are related, they are not the same thing), but it's the best simple one I can think of.

And you've just made a strong case for SABRmetrics and why traditional stats are nice but don't really begin to tell the whole picture.

[Anisoptera]

I feel like I stepped into Bizarro World reading this thread.

I have to say, however, that if a case could be made for using RBIs as the determinant stat for the MVP award that Charley made it when he said you don't come away from a game remembering VORP; you come away from a game remembering the HRs and RBIs. Should that be the way you select who you want on your team (past, present, or future)? Certainly not. But I wouldn't at all mind an award for it. I'd just rather it be called the Hero award or something. The only time I might want to take them into account when deciding on MVP would be in very close contests, but even then I'd want to look at other things like defense first.