The superstats of today

[holyroller]

Originally posted by Plutoro
Really, im shocked...

I don't doubt it.

Any system can win, ANY. You can be a light power hitting, great defence team that runs alot and you can win. If you do not think that you are ignorant and need to watch a baseball game.

I have watched plenty of baseball, thank you. I see that you are using the 'ignorant' word since I used it, fair enough. As to your point, find me one example of a team that did not get on base or hit for power yet won a world series. Just one.

Im sorry but you come off as a Bill James kiss ass, no offence as you actually are rather knowledgable.

A Bill James kiss ass? Huh? Explain, please.

I do actually understand the majority of the systems, and I do not agree with what is favored over what. You need a mix of everything to be extremely successful.

You are right, you don't. As for a mix, you are correct in a sense. Baseball is half offense and half pitching and defense. What percentage of the other half is pitching and what is defense is up for debate, but there is no doubt that offense is more important. Regardless, a winning team does have to be good at scoring runs and avoiding runs. I agree. So you know, Win Shares takes offense, pitching, and defense together and places on value on it. That is why it is so valuable. It isn't perfect, but you would have to understand it before you could point out the flaws.

So to simply dismiss any argument against a player you like by saying he has x amount more win shares in a season would make you the ignorant one.

I agree, Win Shares isn't perfect. I don't recall saying anything contrary to that.

i will GURANTEE you that in 10 years a new favorite stat will apear and OPS will now be considered obsolete. It has happnened countless times before.

Actually, there are better statistics than OPS already. You are right though, sabermetrics evolves like any other scientific field. People are constantly coming up with better measurements to replace inadequate models or to fill gaps. I don't see how this supports you argument. On the contrary, it is a point in sabermetrics favor.

[Dwolfson20]

Originally posted by Plutoro
That is complete and total bull.

They could differentiate a little if they wanted too, draft a player out of the Hatteberg mold or add a player that walks a little less but gets a few more hits/steals.

But the whole point is that those guys are overvalued. When you have to watch every dollar, like the A's do, you have to buy only the things that give you the most bang for your buck. Speed and batting average are not that. Walks are.

[holyroller]

Originally posted by Plutoro
That is complete and total bull.

They could differentiate a little if they wanted too, draft a player out of the Hatteberg mold or add a player that walks a little less but gets a few more hits/steals.

I think you are still missing the point. It isn't about any single stat or even two statistics. It is the entire value of the player to the team. If you don't think Beane understands that...well, there's not much I can say to you.

[Plutoro]

Most teams before 1920 won without power and walks.

It supports my argument (which you have yet to acknowledge) THAT THERE IS NO ONE STAT THAT DEFINES SUCCESS.

So why is sabmeritics followers simply throw winshares or RC at anybody during an argument.

[crackpott]

Originally posted by Plutoro
Really, im shocked...

Any system can win, ANY. You can be a light power hitting, great defence team that runs alot and you can win. If you do not think that you are ignorant and need to watch a baseball game.

...It's not looked kindly upon to make ill-informed generalizations of members by saying "If you do not think that, you are ignorant and need to watch a baseball game." First off, check the OT forums for all the different threads involving GForce recently and threads of the past (I think he's cleaned his act up significantly) regarding Bigphesta. No one respects comments like this and they just make you look like an ass. Regardless, it would also put many on this forum into the "ignorant" category, as simply witnessing a baseball game is irrelevant - statistical analysis is how you can determine what will win.

Originally posted by Plutoro

Im sorry but you come off as a Bill James kiss ass, no offence as you actually are rather knowledgable.

I haven't seen "praise Beane" anywhere in the thread. I still disagree with a few of his philosophies, but on the whole he's done a damn good job, and you should be giving credit where it's deserved.

Originally posted by Plutoro

I do actually understand the majority of the systems, and I do not agree with what is favored over what. You need a mix of everything to be extremely successful.

If you need a mix of everything, why do the A's not resort to "small ball"?

Originally posted by Plutoro

So to simply dismiss any argument against a player you like by saying he has x amount more win shares in a season would make you the ignorant one.

I'll agree with that. But it's equally foolish to just dismiss the stat because you don't know how it works. Personally, I don't like win shares all that much. I think it's a good way to analyze a players contribution to his own team vs other players on his team, but I think it's flawed in comparing players between teams. That doesn't mean I'm going to blast anyone who uses it. However, those who do choose to use it should (and for the most part, I think they do) realize it's not the only stat that means anything. It's just one of the stats which takes everything into account, which is why it has become popular.

Originally posted by Plutoro

i will GURANTEE you that in 10 years a new favorite stat will apear and OPS will now be considered obsolete. It has happnened countless times before.

Yes it will. And the smart ones will move on to that stat. That's not to say OPS wasn't a useful stat, but it can and will be improved upon. It's all about fine tuning. Say you're using a 2 GHz computer now. And in 10 years you're up to... whatever. 10 GHz? It doesn't matter. To say that the 2 GHz computer was meaningless and shouldn't have been used is asinine. At the time, it was the best thing you could be using. There's no reason to stick with a 486 just because technology keeps advancing, just like there is no reason to stick with only primary statistics because sabrmetrics is advancing.

[Eckstein 4 Prez]

Originally posted by Plutoro
I do actually understand the majority of the systems

Originally posted by Plutoro
I think there should be some sort of RBI rating of some sort. The ability to drive in runs, some players just have it, period.

These two statements cannot possibly both be true. Since you seemed pretty adamant about the second one, I'm guessing the first one is not true.

[Plutoro]

Where have I not giving him credit, hes a great GM (beane) but hes to smug.

I do understand winshares and such, why you say I dont is beyond me. I understand them, but I dont like them.

I was more curious as to why it is people want to try and put everything into 1 stat.

[holyroller]

Originally posted by Plutoro
I do understand winshares and such, why you say I dont is beyond me. I understand them, but I dont like them.

I was more curious as to why it is people want to try and put everything into 1 stat.

If you don't understand why you want to put one value on a player then no, you don't understand "win shares and such."

[JML]

Originally posted by Plutoro

I was more curious as to why it is people want to try and put everything into 1 stat.

Well, by doing that, you can compare players at different positions with different levels of performance as measured by different stats (hi-BA/Low SLG vs. Low-BA/Hi-OBP, for example). With Win Shares, you can even compare a pitcher's contribution to a slugger's. I think that is pretty useful.

Are you having a problem with the accuracy of these stats, or the use of them in general?

[crackpott]

Originally posted by Plutoro
Most teams before 1920 won without power and walks.

It supports my argument (which you have yet to acknowledge) THAT THERE IS NO ONE STAT THAT DEFINES SUCCESS.

So why is sabmeritics followers simply throw winshares or RC at anybody during an argument.

Without power or walks, eh?

1919
In the NL:
Cincinnati
1st in walks
2nd in slugging percentage

In the AL:
White Sox
4th in walks
2nd in slugging (.001 behind 1st)

------------------------------------------------------

1918
In the NL:
Chicago Cubs
1st in walks
2nd in slugging percentage

In the AL:
Boston Red Sox
3rd in walks
3rd in slugging percentage (they also finished 3rd in R/G... coincidence?)

---------------------------------------------------------
1917
In the NL:
New York Giants
5th in walks (they made up for it by dominating in average, having the league's highest OBP)
2nd in slugging percentage

In the AL:
Chicago White Sox
2nd in walks
2nd in slugging

----------------------------------------------------------
1916
In the NL:
Brooklyn Dodgers
7th in walks (but first in OBP)
1st in slugging percentage

In the AL:
Boston Red Sox
6th in walks
5th in SLG
...I should point out they were also dead last in stolen bases, so it's not like they succeeded with the other approach. They were actually the third worst team offensively in the league... they just had some good pitching.

-----------------------------------------------------------
1915:
In the NL:
Philadelphia:
2nd in walks
2nd in slugging

In the AL:
Boston Red Sox
4th in walks
3rd in slugging

------------------------


I could keep going back further...

anyway, the observations I uncovered:

No, you don't need power and walks to win. However, you appear to need them to be offensively successful, as those two things seemed to almost always correspond to runs per game. Even when walks didn't, OBP was still high, which is a more used statistic than walks anyway.

Check out the numbers yourself. May I suggest baseball-reference. Care to retract this statement or would you like to post some more unreferenced assumptions?

[TLBOrioles]

AVG - ok
OPS - better
RC (et. al.) - better still

Win shares - contrived

[crackpott]

Originally posted by sixto
I still got absolutely zero discussion about whether a HR is correctly valued at twice as good as a double. I seriously doubt it.

I doubt it too. I've never seen it statistically analyzed although if you think about it...

either will likely drive everyone on base home. So the homerun doesn't have much of an advantage there.

Where it comes into play is a homerun will automatically score another runner, whereas with a double, you need someone to hit a single to score you.

Assuming the average AB has one runner on base (this is all very non-scientific, I know.. just giving you my opinion) a homerun will score 2 runs.

A double will score 1, with probably about a 50/50 chance of the guy on second scoring. (depends on outs and so forth). Thus, likely 1.5 runs. Which would make a homerun actually only 1.333 times more valuable than a double. But I'm sure there's a lot more to this.

[crackpott]

Let's round it up to 1.4 to account for those times the slow ass runner at 1st doesn't score on a double.

[magnet]

Originally posted by Plutoro
Most teams before 1920 won without power and walks.

if what i think you meant is true (that these teams had nowhere near the power of the teams that followed) then you truly dont understand baseball. These teams played in a completely different game (deadball era, maybe you've heard of it ) comparing them to stats of today would be ludicrous, or, if i may borrow a word of yours, ignorant.

[crackpott]

Originally posted by magnet
if what i think you meant is true (that these teams had nowhere near the power of the teams that followed) then you truly dont understand baseball. These teams played in a completely different game (deadball era, maybe you've heard of it ) comparing them to stats of today would be ludicrous, or, if i may borrow a word of yours, ignorant.

hey hey... I actually did work to disprove this claim. Don't think you can just waltz in here and state something we'd all assume is obvious but might not be and get away with it!

[kq76]

Sorry for getting to the party late.. And no, I won't be offended by what you said and start flaming you like others have.

Originally posted by Plutoro
Why do people put so much stock in them? All of them have flaws and most (NOT ALL) are developed by likeminded individuals and the current ones will die off in a few years.

...

But why are they so valued and taken so strongly, I mean why is the Winshare the choice stat when decideing the best players in the league by the "statheads"

I think CP hit the nail on the head when he said that they are thought of as the best available stats out there. I agree, it would be oh so nice to have 1 all encompassing stat to compare players, but like TLBOrioles said, it is contrived, and like CP said, flawed when comparing players on other teams. IMO, it's garbage (yes, I have looked at it extensively) and I refuse to put any weight into it. However, I think the desire for this holy grail stat is so overwhelming that people can't help themselves from this poisonous fruit and so they look past its faults and use it as much as they want. That's fine though, others can use whatever stat they want, I'll just have to give their arguments less weight.

RC, although it is fairly accurate when compared to the most accurate run production stats, is similar to Win Shares imo. It is, to again borrow TLBO's perfect word for this, contrived. In fact, I find most things that James comes up with to be contrived. I'm shocked so many people put so much into his stats. Don't get me wrong, I'm very glad James is out there to spur on controversy and to get others interested in coming up with better stats. But for people to look at him as the sabermetric savior as so many seem to is just silly imo.

Originally posted by Plutoro
What makes stat A more valuble then stat B.

The point of statistics is to predict the future. If stat A predicts the future better than stat B, then stat A should be given more credence, no? You'll often see sabermetricians test these stats against future data (comparing it against past data is faulty as you can make it "fit" the data as James apparently loves to do) and the one that has a higher correlation is assumed to be the better stat. However, the true test is time or sample size. If stat A is consistently better than stat B when compared against more and more data, then even more credence is given to it. By that time though, a stat C has usually developed that is suspected to be even better and the test is now between it and A, while B is forgotten about.

I think the most important thing is to understand the stats yourself. What do they really tell _you_, not what do others tell you about them. If you look at how a stat is formulated and it makes sense to you (sense meaning that you a) can understand it and b) don't see any major faults with it), then feel free to put weight into it. If not, then don't. Rely on the stats that make sense to you.

[Skipaway]

Originally posted by kq76
The point of statistics is to predict the future. If stat A predicts the future better than stat B, then stat A should be given more credence, no? You'll often see sabermetricians test these stats against future data (comparing it against past data is faulty as you can make it "fit" the data as James apparently loves to do) and the one that has a higher correlation is assumed to be the better stat. However, the true test is time or sample size. If stat A is consistently better than stat B when compared against more and more data, then even more credence is given to it. By that time though, a stat C has usually developed that is suspected to be even better and the test is now between it and A, while B is forgotten about.

Most statistics are not about predicting the future. And I also believe you don't mean it that way. Usually when we talk about "predicting the future", it's about how the future will play out: how would a player perform next week? how well would a team do next month?

Most sabermatrics are not about that. K/9 or K/BB might be better indicators for future success than ERA+ or OPS+, but they are not as good in GRADING THE PAST!

Statistics are compiled from records of what happened, and the analysis works done on these records. What you mean by testing the stat consistency is actually just a way to ensure the stat is still a good indicator for "newer past". Do you actually believe something like win share is a good stat for "predicting the future"?

Maybe there will be one day that sabermetrics is advanced enough for us to say " hmmm, check out these stats, I'd say this guy got 89.6% of chance of hitting a homerun in today's game". I seriously doubt it though.

[holyroller]

Indeed. The point of statistics is to measure what a player has done. The analysis of those statistics, among other factors, is what predicts future success to the extent that it can be predicted. That extent is, despite systems like Brock2, Vlad, etc, quite limited.

[kq76]

Originally posted by holyroller
Indeed. The point of statistics is to measure what a player has done. The analysis of those statistics, among other factors, is what predicts future success to the extent that it can be predicted.

God, I hate it when I type up a post and somehow it doesn't get posted.. Try #2..

Yes, that is essentially what I meant. I did not think that I had to draw so fine a picture by pointing out the connection that HR has so kindly done for me.

To me, what is really important is to try to predict the future, not to just analyze the past. You see so many people trying to make comparisons like Bonds vs Ruth, Bonds vs ARod, etc. That I don't care about and I don't think you really can fully. Sure, one day someone might come up with a statistic that can compare them to my satisfaction, but I certainly won't hold my breath.

As an OOTP GM, what I really want to know is how can I make my team better? And to do that I do think you can use some statistics to help predict the future. Sure, the accuracy won't be 100%, but any added information that helps is welcome.

In hindsight, you are correct though. I should have said, "IMO, the most important point of statistics is...", not "The point of statistics is...". And I do understand that for people who could not care less about the future, that "grading the past" is all that they use statistics for.

[Spielman]

Originally posted by sixto
I still got absolutely zero discussion about whether a HR is correctly valued at twice as good as a double. I seriously doubt it.

Neyer had a bit on this a long time back. Unfortunately, it looks like the article in question is no longer in his ESPN archive.

Anyway, he made the assumption that 2(double)=HR and got reamed for it in his mail. He checked into it and found that homers are actually worth more than two doubles because they're more efficient in actually scoring runs. Too many times that a guy who hits the double gets stranded on second, I guess.

Unfortunately, I don't remember the exact numbers he came up with.

-Spielman