PDA

View Full Version : A Thought on Sabermetrics


voodoochile
09-24-2009, 06:58 PM
Now let me premise this post with a disclaimer.

I don't hate stats. I also think some of the newer stats have value, but I am not sold on all of them.

Now for the thought I had...

It occurs to me that sabremetrics and stat driven analysis of the game has really taken off in the last 20 years, even more so in the last decade. I don't recall when Moneyball was published but that seemed to be the moment when it crossed over into mainstream thought and wider acceptance.

During that same time, baseball has been dealing with the worst doping scandal in it's history. A huge percentage of players were almost openly using PED's. This is obvious from the way records were falling like the leaves are about to start doing. Then of course there is a list of athletes including many of the top stars testing positive for steroids and other PEDs.

Is it possible that the sabermetrics have been skewed by this fact? It used to be accepted that guys who batted .300+ were good hitters, but now that point gets disputed if they don't walk or hit for much power.

However, power numbers are dropping and have been for several years. As used to be the case only one player will even come close to hitting 50 home runs and less than 10 will crack 40 this year and those numbers are actually up from last year.

Could it be that the numbers sabermetric gurus are arguing as minimum acceptable are actually horribly skewed because of the PED usage and in reality the line that differentiates good hitters from poor is actually much lower than is commonly accepted by the sabermatricians and their disciples? Do today's numbers stack up with historical numbers?

I realize this might blow up into a massive fight, but I'm asking for some serious replies.

Daver
09-24-2009, 07:01 PM
Performance enhancing drugs have been in the game since the sixties, where are you drawing the line on what numbers are skewed?

Daver
09-24-2009, 07:21 PM
Also, where are you figuring in the fact that more pitchers got caught using PED's than hitters?

voodoochile
09-24-2009, 07:35 PM
Also, where are you figuring in the fact that more pitchers got caught using PED's than hitters?
An interesting point, but offensive numbers definitely seem to be outpacing historical averages while pitching numbers are worse in general.

As to PED's, I am talking about steroids, not amphetamines. I think it's obvious they've had a bigger effect on the game.

SI1020
09-24-2009, 08:16 PM
I'm always afraid to interject on certain subjects but here goes. The "greenies" and other amphetamines of the 60s and 70s don't begin to compare with the designer drugs used in more recent times. Not only that but amphetamine use over any extended period of time is going to hurt your performance not help it. Anybody ever know a speed freak? It's not a pretty sight. To me it's like comparing a gnat to an elephant. The steroid era did greatly distort individual records, and who knows how many championships in that era are tainted. I think the scope and breadth of steroid use casts a bigger shadow then even the widespread gambling in the early part of the 20th Century that culminated in the Black Sox scandal of 1919.

asindc
09-28-2009, 12:31 PM
Now let me premise this post with a disclaimer.

I don't hate stats. I also think some of the newer stats have value, but I am not sold on all of them.

Now for the thought I had...

It occurs to me that sabremetrics and stat driven analysis of the game has really taken off in the last 20 years, even more so in the last decade. I don't recall when Moneyball was published but that seemed to be the moment when it crossed over into mainstream thought and wider acceptance.

During that same time, baseball has been dealing with the worst doping scandal in it's history. A huge percentage of players were almost openly using PED's. This is obvious from the way records were falling like the leaves are about to start doing. Then of course there is a list of athletes including many of the top stars testing positive for steroids and other PEDs.

Is it possible that the sabermetrics have been skewed by this fact? It used to be accepted that guys who batted .300+ were good hitters, but now that point gets disputed if they don't walk or hit for much power.

However, power numbers are dropping and have been for several years. As used to be the case only one player will even come close to hitting 50 home runs and less than 10 will crack 40 this year and those numbers are actually up from last year.


I realize this might blow up into a massive fight, but I'm asking for some
Could it be that the numbers sabermetric gurus are arguing as minimum acceptable are actually horribly skewed because of the PED usage and in reality the line that differentiates good hitters from poor is actually much lower than is commonly accepted by the sabermatricians and their disciples? Do today's numbers stack up with historical numbers?
serious replies.

I would be interested in seeing a statistical analysis done to compare the last 50 individual years with each other using currently-accepted stats that were not available in the 60s, such as OPS and ERA+. I say the last 50 years because the 60s was a pitching decade, the 70s a transitional decade, the 80s more of a balanced decade, and the 90s a hitter's decade (my opinion; yours might reasonably differ).

I am among those who are skeptical of the "stat of the month" approach to evaluating players that seems to be in vogue these days, but I will keep an open mind to any analysis that shows certain stats to be reliable predictors of individual player performances and overall league trends.

Zisk77
09-28-2009, 01:24 PM
Couple of thoughts...PED's were used primarily by pitchers to recover faster. They didn't really do much to enhance an individuals performance (greater velocity, command, etc.) the way they did for hitters(chicks dig the long ball).

Stats are a useful tool to evaluate players, but they don't tell the whole story. The best evaluator of talent is to have qualified people watch the player.

khan
09-28-2009, 01:33 PM
I think that statistical analysis is a tool that is usable to make appropriate judgements on players and teams. I don't see them as being all-encompassing, but that they are an invaluable tool in terms of assessment.

I think that sabrmetrics(sp?) have helped, not hindered a team's ability to do their jobs. I think that sabrmetrics(sp?) have helped, not hindered my ability to enjoy baseball. I think that proper statistical analysis helps by using numbers to create a more complete picture.

I also think that the explosion of technology and of statistical analysis has enabled us to understand which stats are useful [i.e. WHIP or ERA+ or OPS], and which ones are antiquated and less useful [i.e. ERA or BA].

khan
09-28-2009, 01:42 PM
Is it possible that the sabermetrics have been skewed by this fact? It used to be accepted that guys who batted .300+ were good hitters, but now that point gets disputed if they don't walk or hit for much power.

Sure, it's possible. But I think its more likely that years ago, we had a less-complete understanding of what numbers are useful in analysis and which ones are not. Moreover, I think that what is valuable in one era is not as valuable in another era. [The point of Moneyball was to find what is undervalued in an industry to gain a competitive edge. OBP was undervalued at the time, hence that was one of the foci of Billy Beane and his statisticians.]


Could it be that the numbers sabermetric gurus are arguing as minimum acceptable are actually horribly skewed because of the PED usage and in reality the line that differentiates good hitters from poor is actually much lower than is commonly accepted by the sabermatricians and their disciples?
It could be.

But I think that the rampant PED usage required teams to arm themselves to be able to compete with other teams that were replete with PED cheaters. [I believe that each and every WS winner in the past ~15 years had at least one PED-cheater, including our own beloved 2005 winners.]

But again, that was the point of Moneyball: To identify what is a overvalued or undervalued in MLB, and to sell high on the overvalued, while buying low on that is undervalued.


Do today's numbers stack up with historical numbers?
We can't compare, thanks to PED cheaters like Bonds and Sosa and Clemens and A-Roid and Manny and others, and all the media types and fans that let these cheaters go relatively unpunished.

I think that statistical analysis will evolve. That is, instead of comparing one era to another [i.e. ERA in this era to that of the '60s], it will be used to differentiate players WITHIN an era. [i.e. ERA+]

Moses_Scurry
09-28-2009, 02:04 PM
So what is over- and undervalued now? Obviously OBP is not undervalued anymore. Is it overvalued or is it appropriately valued?

It seems like Billy Beane's approach has taken a pretty big nosedive the last few years. He also doesn't seem to be going in a specific direction. It seems like good defense and speed is undervalued now. Shouldn't Beane be stocking up on David Ecksteins and still being successful? Or is it that specifically OBP being undervalued was the key to the Moneyball success and that it won't work with other undervalued stats?

khan
09-28-2009, 02:18 PM
So what is over- and undervalued now? Obviously OBP is not undervalued anymore. Is it overvalued or is it appropriately valued?
And herein lies one of the top task(s) for GMs in baseball: To find what is undervalued, and what is overvalued. I personally don't know for certain right now, because I make my living in an industry outside professional baseball.

It seems like Billy Beane's approach has taken a pretty big nosedive the last few years. He also doesn't seem to be going in a specific direction. It seems like good defense and speed is undervalued now. Shouldn't Beane be stocking up on David Ecksteins and still being successful? Or is it that specifically OBP being undervalued was the key to the Moneyball success and that it won't work with other undervalued stats?
Before I address this, let me ask you a question:

Did you read "Moneyball," or did you get your information about this book/these ideas from second-hand sources? [i.e. most of the morons on ESPN, Hawk's insane dislike of Billy Beane for some as-yet undisclosed reason, etc...]

voodoochile
09-28-2009, 02:31 PM
OPS is a fairly simple concept and one that may have actually been subconsciously used. Just because someone decided to finally add the two numbers together doesn't mean that players who reached base a lot and hit for power too weren't in high demand prior to that fateful day someone said, "Hey, lets add OBP and slg and see what it tells us." Obviously it doesn't tell anyone anything specific other than "guys with high OPS are good players and a lot of them are in the HOF."

Kind of the question I am asking is, have these newer stats been employed on historical data to see how those numbers stack up? That would be a test to see if today's "minimum acceptable numbers" are valid or a result of the steroid era.

WhiteSox5187
09-28-2009, 02:48 PM
I think you hit on an interesting idea of what is an "acceptable" sabermetric because really that varies from year to year. For example in a year dominated by pitching - 1968 for example, one could argue that if a guy had an OBP of .320 that was good because the average OBP that year was .300 (I don't know that for a fact, I'm just making up stats). A .350 OBP is a number that is thrown around a lot as being the minimum for a "good" leadoff hitter. That was Pods' OBP in 2005, yet the league average OBP was .330, so Pods was better than average, yet still by some accounts is evidently a "horrible baseball player."

I don't think there is any one accepted minimum for anything as it varies year from year era from era. In 1968 (or whenever the year was when only Yaz hit .300) a .250 average might have been a good average where as now it is a very mediocre average. I also think we have a tendency to focus on certain sabermetrics and ignore other facets that a player can bring to the table. For example a guy might have a 700 something OPS, but can still steal a hell of a lot of bases with a good OBP, if that is the case you are trading power for speed and there are a lot of people who either don't like that idea or don't quite get it.

Just my two cents.

Moses_Scurry
09-28-2009, 04:04 PM
And herein lies one of the top task(s) for GMs in baseball: To find what is undervalued, and what is overvalued. I personally don't know for certain right now, because I make my living in an industry outside professional baseball.


Before I address this, let me ask you a question:

Did you read "Moneyball," or did you get your information about this book/these ideas from second-hand sources? [i.e. most of the morons on ESPN, Hawk's insane dislike of Billy Beane for some as-yet undisclosed reason, etc...]

I did read Moneyball, but it has been awhile. I'm not saying that Moneyball states that he was trying to get OBP only. What I'm saying is that IF OBP is an undervalued stat, then an undervalued team full of OBP machines could be successful; however, hypothetically, if stolen bases were the undervalued stat, an undervalued team loaded with SB machines might not be successful.

khan
09-29-2009, 11:49 AM
I did read Moneyball, but it has been awhile. I'm not saying that Moneyball states that he was trying to get OBP only. What I'm saying is that IF OBP is an undervalued stat, then an undervalued team full of OBP machines could be successful; however, hypothetically, if stolen bases were the undervalued stat, an undervalued team loaded with SB machines might not be successful.

I think that's a fair statement. In fact, one of the concepts in sabrmetrics and in Moneyball is that there isn't as much correlation between SB and a team's success.

Sabrmetrics do not attempt to re-invent the game: A team still has to hit, pitch, catch the ball, and run the bases to a reasonable degree to succeed. What sabrmetrics [or more appropriately, accurate statistical analysis] have done are to more properly describe what is happening in the game. To use numbers to paint a picture. In turn, these numbers help to enable a team to properly recruit the best possible employees in the prosecution of their business.

Now, there are those observers out there [usually geezing geezers] who fear change and fear advancement and fear the application of technology to the game.

That's OK, but IMO, the world changes and evolves, and baseball is no different. Clinging to antiquated ideas will lessen a team's ability to compete. And lessen a member of the media's ability to describe what is happening. [cough-cough, Hawk, cough-cough] And lessen a fan's ability to appreciate the game.

khan
09-29-2009, 11:56 AM
OPS is a fairly simple concept and one that may have actually been subconsciously used.
Sure. But without widespread acceptance, [in an overly-traditional game] OPS could not have been taken as an analytic tool. This is but one metric that can be used to judge a player and his effect on his team/a game/a season.

Kind of the question I am asking is, have these newer stats been employed on historical data to see how those numbers stack up? That would be a test to see if today's "minimum acceptable numbers" are valid or a result of the steroid era.
It can't be done. Comparing eras is a futile exercise. Thanks to clowns like Bonds and Clemens and Sosa and A-Rod, we ESPECIALLY can't compare eras. Thanks to media outlets that serve as hype machines for the game and the players [i.e. espn], we can't compare eras, especially the 'roid era to others. Thanks to players and management that conspired to keep secrets, it can't be done. Thanks to fans that want to sweep the era of cheating under the rug, it can't be done.

I'd argue that statistical analysis can be gainfully employed to compare players/teams within a current season or recent seasons. And that is a good thing for all of us that enjoy baseball, IMO.

Oblong
09-29-2009, 12:50 PM
I don't think stats will ever give you a total answer on a general question. I like sabermetrics and think they have a place in the game but like most things it all depends on how they are used. You can always find one anomaly that could give someone an argument that a particular stat is garbage.

But consider:

Ruth, Gehrig, and Williams - top 3 all time for OPS 3 of the top 4 for OPS+(Bonds #3).

Runs Created - top 10 includes, Ruth, Musial, Aaron, Cobb, Williams, Mays, Gehrig, Rose, Rickey...

One thing I've wondered about with regard to stats and PED is with pitchers. We all know the cliche that good pitching will beat good hitting. Is that true? I don't know. But doesn't it seem like we've seen some of the best all time pitchers within the last 20 years, even more so than hitters? Maddux, Pedro, Randy, Rivera, Hoffman, even Clemens (pre doping?). My theory on that is that the predominate PED use has actually lowered the bar for average pitchers. The average pitchers are suffering because the hitters are much better because of the PED use. But these elite pitchers like those I mentioned will beat anybody so that skews their stats upward because the average pitchers are now that much worse. That's just a theory, I have no way of testing it.

ma-gaga
09-29-2009, 03:37 PM
have these newer stats been employed on historical data to see how those numbers stack up?

Well. BP's whole concept of EqA (http://www.baseballprospectus.com/glossary/index.php?search=EQA) is supposed to equalize players on a yearly basis. So the 2009 EqA Average = 0.260, the 2001 EqA Average = 0.260, the 1985 EqA Average = 0.260, etc., etc. It's similar to OPS+ (http://www.hardballtimes.com/main/statpages/glossary/#ops+) or ERA+ (http://www.hardballtimes.com/main/statpages/glossary/#era+), but the problem is that it's really hard to get a handle on HOW MUCH better an EqA = 0.280 player is compared to a EqA = 0.260 player. OPS+ and ERA+ have a much clearer way of saying, "he's 12% better than the average player!" and are more understandable to the average fan.

I think there's a stat for everyone, but it depends on your comfort level with numbers. Personally, I love VORP. Where players are evaluated against players of similar position versus just their raw numbers. Why Mauer > Pujols... [checks] Aw damn. (http://www.baseballprospectus.com/statistics/sortable/index.php?cid=204031) Why Mauer >> Teixeira despite the great disparity of RBI's. :tongue:

The question is; How do you define Sabrmetrics? It's such a big field right now, it's hard to generalize. Some stats are good and useful, some stats have predictive value, and some stats are simply worthless. If you blanket cover the entire field, you aren't going to get a good answer.

gosox41
09-29-2009, 09:53 PM
Performance enhancing drugs have been in the game since the sixties, where are you drawing the line on what numbers are skewed?

They seem to have become more potent.