Hardy/Bush arbitration numbers (Latest... Bush signs: $2.55 million; see reply #90)

[sbrylski]

Re: Monty: It may come across as speaking in absolutes, but I think most "stat-heads" understand the flaws. It would be more accurate to say that deeper stats speak in odds, and when you gravitate to the statistical most likely event, it appears to be advocating an absolute.

 

Statistical analysis is just one tool in player evaluation. However, when some come with ERA and Wins, which are stats, to argue their point, Ennder and others are going to come with stats that are more likely to be accurate representations. This has been a straight stats verse stats debate, in which someone can be more correct than the other. (That's strong language, I don't mean to offend...)

[Ennder]

I wrote the previous post before reading "Page 4."

 

Tbadder, excellent post with some excellent quotes!

 

From sbrylski06's graph what I see is that a perfect correlation is 1.00, and statistical analysts are claiming victory by using a stat with a year-to-year correlation of 0.271. Granted, there is less room for variance than a stat with a 0.068 beta, but it's still not enough proof to convict someone who still believes in good ol' ERA of heresy.

xERA is not trying to predict the next years ERA though so it shouldn't be expected to have a great correlation. It is trying to describe what happened the current year. If it were trying to predict the next season you would have to run age related regressions on it etc. Also by it's very nature ERA is a random stat over a single season, I'm sure if you look at the correlation of xERA to itself you will get a much stronger correlation. ERA just takes way more than 200 IP to be a meaningful measure of anything.

[monty57]

sbrylski, well put, and as I mentioned, I do like stats for that reason... you're helping to put the odds in your favor. I wasn't trying to point any fingers at you as being elitist in any way, in fact I always enjoy your posts. I simply saw the graph and used it in some small way to add to my previous post. I don't want to take any sides, as I try to learn from all of the posters here, and I greatly appreciate the efforts and thought most put into their work.

 

I thought I'd add my two cents when the poster I quoted essentially said he spoke for the "scientific community" and this community isn't impressed by Geno and his "speculative/mythological" tendancies of using ERA, while this highly enlightened community, with a "pretty good track record of being right," only utilized information on which "Rigorous statistical analysis has been done to weed out which stats predict future performance, which stats don't, and to what degree."

 

While Geno may have his faults, I'd hardly consider him "just one ignorant person following another following another." Geno doesn't need me to defend him, but he does bring a lot of good points to light, and I generally enjoy reading his posts, whether I agree with his conslusions or not. I doubt that he has "scientific communities" up in arms against him.

 

As far as this supposed "scientific community," I simply wanted to point out that most scientific communities would cringe at the lack of accuracy in baseball statistical analysis. Apparently, the "pretty good track record" still can't get more than 15% of 1st round picks to play more than three years in the majors, and can't reasonably predict which teams will win more games than others any better than your average sports journalist. As this poster stated "It was universally accepted at one point that the Earth was flat, until science proved that wrong." In the world of science, baseball statistical analysis is somewhere around the point of trying to disprove that the world is flat.

 

Statistical analysis is a key tool for the future of baseball executives, and so is sending someone to go out and watch a game. The people who best utilize all of the tools at their disposal, while recognizing their inherent strenghts and weaknesses, will likely be more successful than those who choose to put blind faith into any one thing.

[endaround]

The social scientific community would love to have baseball's statistical track record.

[monty57]

Right, and my degree's in economics, which is very good at determining what happened 2-3 years ago, but not so good at predicting what's going on right now or what will happen next. I've always found it somewhat odd that economists consider themselves scientists for the same reasons I've described above. There is absolutely nothing wrong with running the numbers to do your best to put the odds in your favor, but the best economists can't accurately predict the effects of the upcoming interest rate cut on production (GDP) or on inflation, and the best SABRmatrician can't accurately predict what will happen in next year's baseball season. There is some degree of understanding, but there are far too many moving parts to make accurate predictions (at least accurate as a physicist or chemist would acknowledge). One small for-instance is that statistical analysis doesn't seem to take much consideration for injuries, and takes no consideration for psychology. But then, how can it, as many injuries are never reported, and no one knows lingering effects, and psychology is a wild card that's different for everyone. You can't predict if someone is going to get fat after signing a big contract or if they're going to have marriage problems that lead to poor performance.

 

GMs are put in the position of taking all of the information at hand and making a best guess estimate as to what will be the best route to follow. As I said above, I believe statistical analysis to be a very important tool for GMs, but I'd still like to talk with the players before signing them, and I'd like someone on staff to go out and watch them play once or twice. The reason the Brewers selected Fielder is that they went out and talked with him. When they asked him to step on a scale, he refused, but invited them to watch him work out. They were so impressed with his workout regimen that they drafted him. I believe much of his talent comes from his drive to prove wrong everyone that told him he was too fat. That's good psychology. Chuck Knoblauch's sudden inability to throw the ball to first from second is bad psychology, but it's something no one could have predicted.

 

You put two hydrogens with one oxygen and you get water everytime. You select 30 players in the first round and about five of them will have even decent major league careers.

[TerraceReserved]

Apparently, the "pretty good track record" still can't get more than 15% of 1st round picks to play more than three years in the majors, and can't reasonably predict which teams will win more games than others any better than your average sports journalist

 

That's incorrect. Statistical models can predict fairly accurately the amount of runs scored and runs allowed for a team, thus predicting who will win more games a whole lot better than your average journalist. In The Hardball Times Annual, MGL shows preseason projections on how teams were expected to do, and it turned out pretty dang close.

 

Now are these predictions 100% accurate? Of course not. But they are out there and can do a pretty good job.

[monty57]

I said that realtively tounge-in-cheek, but as a good statistician, wouldn't you have to take their predictions back five to ten years and compare their average accuracy to maybe Sports Illustrated and ESPN's predictions? After all, one year for one magazine would be a relatively small sample size.

 

If they found that they're able to "beat the odds" on a regular basis, I'd bet they'd make more money in Vegas than they make as a hardballtimes reporter. If anyone accurately predicted Colorado to win the NL they'd have cleaned house.

 

Again, I'll state that I believe statistical analysis to be a great tool to help baseball GMs do their job. I just don't believe it is as good a predictor as some make it out to be. If it is indeed so accurate, why was Braun drafted after Zimmerman and Gordon, why was Hart drafted in the 11th round, why do we have to play the games, and why doesn't every SABRmatrician simply bring their predictions to Vegas and retire with their winnings?

[Ennder]

I just don't believe it is as good a predictor as some make it out to be. If it is indeed so accurate, why was Braun drafted after Zimmerman and Gordon, why was Hart drafted in the 11th round, why do we have to play the games, and why doesn't every SABRmatrician simply bring their predictions to Vegas and retire with their winnings?

 

Because nobody said it was perfect, especially 3+ years into the future. All they said was it is more accurate than using ERA, W's and AVG to judge players. I don't think anyone ever came close to saying stats are 100% reliable, just that some are more reliable than others.

 

Btw Zimmerman was more valuable than Braun last year and Gordon and Braun were both rookies so it is pretty hard to make much out of them yet.

[homer]

I've used this analogy before in reference to stats and it seems appropriate to do so again:

 

If I read the weather page in the morning and there's a 60% chance of rain but it doesn't rain all day, that doesn't mean that I stop reading the weather page in the morning.

[sbrylski]

I've used this analogy before in reference to stats and it seems appropriate to do so again:

 

If I read the weather page in the morning and there's a 60% chance of rain but it doesn't rain all day, that doesn't mean that I stop reading the weather page in the morning.

No, but umbrella's don't cost millions of dollars. And people have been using rings around the moon, the positions of cows in the field, and the color of the catepillars in the garden to predict rain for centuries - can't blame the weatherman when he uses those traditional techniques.

 

I'm not sure what I said makes sense. I, of course, actually agree with Homer's point - I just saw a nice opportunity to try and expand the analogy. http://forum.brewerfan.net/images/smilies/wink.gif

 

[monty57]

I think a lot of people who's opinions I really respect are defending against something I'm not really trying to argue.

 

Basically, one poster made a statement in a very elitist tone to the effect that statistical analysis is an absolute science and everyone who doesn't adhere to the beliefs is a moron. I know that most people here don't believe that. I was simply stating that it is not an absolute science, and is actually much less accurate a science than the original poster seemed to insinuate. It is much more like my preferred field of economics (although I'm far from an expert in this field as well), in that it is more of a guideline using mathematics and historical statistics to help make the best decisions/predictions possible for the future.

 

I do have concern when someone absolutely believes something that is not absolutely true, and looks down on those those who don't absolutely believe it. To clarify, I do not believe many of the posters on this forum fit into this category.

[rluzinski]

Again, I'll state that I believe statistical analysis to be a great tool to help baseball GMs do their job. I just don't believe it is as good a predictor as some make it out to be.

 

Anyone who knows anything about sabermetrics should be all too aware of the limitations of any projection system. A perfect projection system that perfectly predicted the skill level of a player for a given year is going to look very wrong, often because of the limitations of a season size sample. Of course, you know this already, as does anyone else who's had to take stats 101. And that's just the first layer of uncertainty. You mentioned injuries, the mental aspect and other factors that might effect a player's performance. Still, by using fairly rudimentary statistics, you can get a pretty firm handle on how predictive current projection systems are.

 

I agree that some people might need to work on their delivery, however. I don't agree that baseball is as difficult as Economics to predict. When you are comparing a statistical model of the American economy vs. a model trying to predict Fielder's 2008 season, it isn't even close.

[logan82]

I think this needed a bump since the arbitration dates are coming up. I am still hopeful that things get worked out before the hearings. I found the bolded part dissapointing.

 

Melvin said there was no progress to report in talks with representatives for Hardy and right-hander Dave Bush, the two arbitration-eligible players that remain unsigned. The sides can continue to talk until an arbitration hearing in February. Those dates have already been set, likely for the middle part of the month after the Brewers report for Spring Training, but for the first time Major League Baseball is instructing clubs to keep the schedule secret.

 

http://milwaukee.brewers.mlb.com/news/article.jsp?ymd=20080129&content_id=2359514&vkey=news_mil&fext=.jsp&c_id=mil

[mothershipconnection]

I guess I missed this discussion the first time around, but I'll throw my hat into the ring.

It appears that the biggest misinterpretation of stats is that people fail to make a molar/molecular discrimination. What many do is look at the failure of stats to predict with amazing accuracy each individual occurrence and declare them invalid. However, that's not what SABR types are doing; they're predicting general trends and "the average case".

 

As a behaviorist, the same misconception is thrown at my field as a criticism, but it comes from people not understanding what we're saying. Nobody is claiming to be able to predict human behavior and performance with 100% certainty, though opponents will have you believe that and claim victory when it's not true. It's the same problem encountered when Alfred North Whitehead told B.F. Skinner to account for his statement "no black scorpion is falling upon this table". It was a straw man thrown into an argument over whether human verbal behavior was operant or attributable to some inner causal agent that couldn't be observed (for hyperbole's sake, we'll call it "grit and heart"). What he asked for was an account of each individual word, and in that order, and at that exact moment. The claim wasn't, however, that such prediction was possible, but rather than a more molar explanation of human verbal behavior was possible through operant analysis. I could say, for example "this stimulus will evoke this response 50% of the time", but the statement is not false when the stimulus occurs twice and the response is not evoked, as I never claimed that it would occur every other time.

 

Statistical projections are often wrong when you look trough a powerful lens at individual and specific instances. They become much more valid, however, the further back you step.

[rluzinski]

To be fair, it's not always the "traditional" fans abusing sabermetric principles, though. Sometimes, the stat guys can overstate the power of the tools they are clumsily wielding. I have been guilty of doing that on more than one occasion myself (I'd like to think that I've learned my lesson, however).

[logan82]

Bush comes to terms.

 

http://blogs.jsonline.com/brewers/archive/2008/02/05/brewers-bush-come-to-terms.aspx

[Oldcity]

Met less than halfway. Just waiting for the Vargas trade now...

[ILuvDaBush]

I would have rather seen him locked into a 2 or 3-year deal, though year-to-year is probably best with the decisions that have to be made soon enough regarding our young cornerstones.

[MillerParkSouth]

I find it interesting that Melvin said that Hardy will probably go to a hearing. For some reason this gives me the impression that he'll be the odd man out when it comes to locking up our young guys and we'll turn the job over to Escobar if he starts hitting in a few years.

[brewmann04]

It does sound like Hardy will either move to 3rd or Might be trade bait down the Line.

[SmartBall]

If I was a betting man I would guess that they are working on a multi-year deal, so if they don't get it done in time, they will just go to a hearing and then replace the one-year contract with the long-term deal once the details are finalized.

[logan82]

If I was a betting man I would guess that they are working on a multi-year deal, so if they don't get it done in time, they will just go to a hearing and then replace the one-year contract with the long-term deal once the details are finalized.

In the past they have gotten a deal done around the midway point before the hearing and then went to a multi year contract from there. I don't know if they plan on something like that with Bush, but it wouldn't surprise me. On the other hand it doesn't seem like Melvin likes to hand out contracts longer than 2-3 years for pitchers.