A Purist's View On Statistical Analysis
I was talking with a friend of mine today and we engaged in a debate about statistical analysis. Being an A's fan, it's difficult not to get caught up in the intense debate between statheads (Bill James and so forth) and the Joe Morgans of the world. My two cents:
The emphasis on statistical analysis has gone way beyond what I can take. Baseball is an athletic contest; let's not make it into a computer game. You can easily inflate average players based on statistical analysis. Take Kevin Youkilis for the Red Sox as a prime example of a high OBP player that would be considered valuable as an offensive player. Pretty good hitter with a patient eye, we all know this from watching games. However, we also know from observation that he's slow. He's a below average fielder, both at 1B and at 3B. Would you take him over Eric Chavez? You'd have to be nuts to say yes, despite Youkilis' 40 point OBP career advantage. There are so many ways to evaluate a good baseball players, some that even metrics can't agree on (defensive range and arm, fielding prowess, speed on the basepaths, etc.). It also doesn't take someone with a Ph.D in Statistics to tell you that Adrian Beltre isn't really that good of a hitter, despite his one great year playing for the Dodgers. His swing is raw and he still regularly tries to pull pitches on the outer half, rolling over the top and creating routine groundouts. Statistics would say that the season was an outlier, but that isn't hard to figure out if you didn't watch baseball on a regular basis.
Now, I understand that part of the hype is due to the old school crowd making baseless statements like: "Moneyball advocates not stealing bases." Statistically, I've heard stealing is beneficial when you're successful anywhere around 75-80% of the time. Logically, this makes sense. You steal a base when you think you're going to be successful most of the time. But, this is what you've been trying to do anyway, you wouldn't do it if you weren't good at it. All statisticians are saying is that if you're good at stealing bases, go ahead. If you're not, don't bother. David Ortiz is a not a good basestealer; you wouldn't have him steal regularly. Not very hard to reason through something like that. And all putting a stopwatch on Pudge's throw to second is doing is reinforcing the fact that he's really good at getting the ball there quickly and that you probably don't have much of a chance of making it to second unless you're Rickey Henderson. So, I'm trying to find the middle ground here. I can't stand ignorance from either extreme.
Statistics to me are just common sense put in numerical format. What really bothers me about some is that they pretend to predict the future based on statistics. People try to extrapolate what will happen based on past stats. That is not the point of statistics; statistics are useful to formulate probabilities for certain events to give confidence to hypotheses and predictions. Using them as an end-all solution is not only crazy, it's incorrect. I can't sit here and tell you that Eric Chavez will hit 30 HRs next year. I can say confidently that he could, because he averages that many HRs, but he could hit 45 after fixing a swing fault, or he could be dealing with a degenerative shoulder and only hit 20. Baseball is random and human; no analysis will change that fact.
I just hope we don't all lose the point of watching baseball. It is a beautiful game. I cannot describe all of the memories I've had playing it and watching it with numbers. Why should I change my thought process?
0 recs |
61
comments
Comments
Yeah, I'm with that.
The stolen base example. 'The numbers' bear out that an acceptable rate of success is about 75-80% (using your numbers, since I don't want to thumb through Moneyball to get what's printed there; sounds right, though). However, as is pointed out in that book (that I would like to rely less on for examples, but who really cares?) basestealers are hearalded even when their sucess rate is below an 'acceptable' rate.
Aha! What is acceptable, though? I make no claims to knowing this. In Moneyball there is a part on the rate of scoring in the different 'situations' in a baseball game, but as I cannot recall them off the top of my head (and good for me; being able to do so would make me way too pathetic), but that doesn't make me scoff at every double steal attempt by the opposition.
These stats were/are a means to understand the game... differently, not necessarily better. And with that different perspective, those that put together a baseball team for the BUSINESS purposes of winning games have found that they can help them in that endeavor. To enjoy stats in a baseball game to the detriment of enjoying the game its self -from the fan's perspective- is missing the point and... well, a bummer. In some instances I do enjoy using my limited seamhead-stathead knowledge to enjoy baseball from a different (again, not necessarily better) perspective. But as I am not financially comitted to the team and they are for my ENTERTAINMENT purposes only, I do not get overly concerned with the stats.
So: If you don't like stats, fine. If you do, that's great, too. Either way, enjoy baseball however and whenever you can as best you can and try not to let other's... perspectives and passions grate on you too much. If either 'side' misses out on something you feel strongly about, oh well, their loss.
I'll take my peanuts, crackerjacks and an A's wins anyway I can.
by StewFan on Aug 21, 2006 10:26 PM PDT 0 recs
Right on
I still can't figure out if, all things being equal, I would rather have a runner on first with nobody out or a runner on second with one out. A ground ball double play can screw up the first scenario and a wild pitch/passed ball in the second scneario would be golden.
There are only so many stats to find inefficiencies with before everyone else catches up to you. Then it becomes one big circle so you move on to MLS, but only after you win a World Series and get a new stadium.
by southofcruiseamerica on Aug 21, 2006 11:25 PM PDT 0 recs
Exactly
by baseb3383 on
Aug 22, 2006 5:16 AM PDT
up
0 recs
Well, I can tell you which one you'd prefer.
I can totally understand why someone would rather not know or understand how the math works behind the scenes. But I am not one of those people. I am an engineer and a scientist, and I have to know how and why you need to steal 70% or better to at least break even.
I want to understand what Voros McCracken means when he says that a pitcher has no control over how well hitters do on balls put in play, no matter if he's right or wrong. And I'm absolutely interested in how fielders can accurately be judged against each other. Those are really cool questions to ask, and the game of baseball is a scientifically analyzable system, just the same as the solar system, the circulatory system, or the shipping system for amazon.com.
Anyway, this is not really meant to be a dig on people who don't care that much about this. But it is usually pretty easy to figure out when you're about to dive headfirst into a meaty statistical discussion and when you're about to plow into the sweet, cream-filled center of a session of chick talk, so I wouldn't worry too much about the amount of the one vs. the other on AN.
by hunter on
Aug 22, 2006 10:46 AM PDT
up
0 recs
To me...
by GlassHeart on Aug 22, 2006 12:51 AM PDT 0 recs
One's eyes...
Most "stat-heads" would agree with your general premise - that you can't just rely on stats any more than you can just rely on your eyes. Most would also say that there isn't one single stat that tells the whole story (although they keep working on compound stats that attempt to get close).
The issue I see is when proponents of using one's eyes hold it up as the only way. Of course, I see the same problem with people that offer stats as the only way. Among the bobble-heads that comment on baseball on television, there are far more of the former than the latter. Those people are what cause the stat heads to dial up their rhetoric. When Joe Morgan or John Kruk says something with absolute certainty and authority and it turns out to be utterly ridiculous based on the available stats, it pisses people off!
With that said, I think statistical analysis of the game is far more important for low budget teams like our A's. Teams with lots of money can afford to buy the best players available at every position (e.g. the Yankees, etc). The A's can't afford to do that. They have to pick their spots. To cite your example, I don't think anyone would argue the Kevin Youkilis is a better player than Eric Chavez, but look at their salaries. Youkilis is cheaper than Chavez - is his production good enough such that the difference in salary could buy the team another decent player? If so, would the combined production of those two players be better than the combined production of Chavez plus a replacement level player?
In this case, I think we can conclude that keeping Chavvy is preferable. What stats do is provide the A's with a method for having that conversation, and in so doing, they can have some level of security in thinking that their money is probably well spent. The A's have scouts that use their eyes to sanity check what the stats tell them. There will always be other factors as well that don't show up in the stats (injury history, ability to deal with pressure, etc), and again that is where the eyes come into play.
by RickeySteals on Aug 22, 2006 12:57 AM PDT 0 recs
Ugh...
by RickeySteals on
Aug 22, 2006 12:58 AM PDT
up
0 recs
OK.
That's ludicrious.
At the end of the day, we like baseball. I don't look up the stats for basketball because I don't like basketball. I don't like watching basketball games. I can barely stand playing it.
However, I watch pretty much every single A's game I can catch on TV. I'll listen to the games I can't watch. I'll Gameday the games I can't listen to. Why? Because I love the game.
I acknowledge the beauty of a well-turned double play. I love sitting on the edge of my seat at the bottom of the ninth with the winning run at the plate. I get ecstatic when I watch a pitcher get a batter so completely off-balance that he not only swings and misses, but turns himself around in the process.
None of us denies the beauty of the game. It's why we watch it. It's why we follow it.
What we do object to, however, is people having orgasms over watching Derek Jeter's "defense", when any objective analysis shows him as being one of the worst-fielding shortstops in the majors for his entire career. What we don't like is when "analysts" scream about a batter's .300 AVG when it's accompanied by a .310 OBP.
We aren't some sort of weird nerd caricature that only looks at spreadsheets and never watches a game of baseball. So please, stop calling us such.
------------------
It also doesn't take someone with a Ph.D in Statistics to tell you that Adrian Beltre isn't really that good of a hitter, despite his one great year playing for the Dodgers. His swing is raw and he still regularly tries to pull pitches on the outer half, rolling over the top and creating routine groundouts. Statistics would say that the season was an outlier, but that isn't hard to figure out if you didn't watch baseball on a regular basis.
The statheads were the ones who said that all signs pointed to Adrian Beltre's last year with the Dodgers being a fluke. That's why DePodesta stuck with his guns and refused to overpay for Beltre's services.
Statistics to me are just common sense put in numerical format. What really bothers me about some is that they pretend to predict the future based on statistics. People try to extrapolate what will happen based on past stats. That is not the point of statistics; statistics are useful to formulate probabilities for certain events to give confidence to hypotheses and predictions. Using them as an end-all solution is not only crazy, it's incorrect.
Umm...you just contradicted yourself. Twice.
And statistics, as a science, must necessarily be about using past data to predict future events. That's the point. There's no reason to do any science whatsoever if you aren't going to attempt to predict what will happen in the future based on what has happened in the past.
And I know I'm repeating myself here, but since you don't seem to be getting the message - none of us believes that the numbers are everything. Only an idiot who had no notion of what statistics actually is would believe that the numbers tell you everything.
by Inquisitor on Aug 22, 2006 1:40 AM PDT 0 recs
Umm... you didn't read my post carefully
You can talk about VORP, OPS, Win Shares etc. all day but that you really don't need so many metrics to predict what should be common sense; you can predict how someone will do with a fairly high level of confidence just glancing at the back of a baseball card.
by baseb3383 on
Aug 22, 2006 5:13 AM PDT
up
0 recs
*sigh*
I understand what you're saying - you believe that it's unecessary to "crunch the numbers" because "common sense" will tell you everything you need to know about the game anyway.
If that's your opinion, good for you. I'd just prefer not to see the runaround logic and poor rationalizations (with a touch of semantic quibbling). You don't like the numbers. I get it.
And yes, I'm in a bad mood. I just had to deal with some idiot blathering on about Intelligent Design as a "valid scientific theory". I'm not in the mood to get more of the same here.
by Inquisitor on
Aug 22, 2006 7:38 AM PDT
up
0 recs
I don't think so
by baseb3383 on
Aug 22, 2006 8:43 AM PDT
up
0 recs
You're not literally wrong when you say
by hunter on
Aug 22, 2006 11:00 AM PDT
up
0 recs
Whoa now
by baseb3383 on
Aug 22, 2006 12:27 PM PDT
up
0 recs
Homers and doubles are better viewed as
by hunter on
Aug 22, 2006 1:02 PM PDT
up
0 recs
I agree on some points
by baseb3383 on
Aug 22, 2006 1:07 PM PDT
up
0 recs
Homers and doubles are better viewed as
by hunter on
Aug 22, 2006 1:27 PM PDT
up
0 recs
slight quibble
by sec119 on
Aug 23, 2006 8:34 AM PDT
up
0 recs
Also
by baseb3383 on
Aug 22, 2006 5:22 AM PDT
up
0 recs
Careful and astute
by rfloh on
Aug 22, 2006 8:53 AM PDT
up
0 recs
Right, that's the whole point of trusting
by hunter on
Aug 22, 2006 11:05 AM PDT
up
0 recs
Everything has bias
by baseb3383 on
Aug 22, 2006 12:35 PM PDT
up
0 recs
Objective analysis...
Statistical analysis is far from "objective." The stats can be made to show whatever the person wants them to show. Personally, I'd take Jeter over Crosby any day, and Scutaro shouldn't even enter into the conversation. Other than Young or moving ARod back to short, I can't think of another SS I'd rather have than Jeter when you factor in all the intangibles.
by IndianaAsfan on
Aug 22, 2006 7:44 PM PDT
up
0 recs
I hate to be Petty but Don't Do Me Like That
The 'stat head' movement isn't for everyone but it sure has allowed a team like the Athletics to maximize and economize. It is tough to argue against how a team like Oakland can remain competitive while keeping its payroll, relative to the other competitive teams, in the lower end of the league's team-salary median. Whether it's assembling the team from year to year or playing percentage ball during an actual game, it would appear, at least anecdotally, that Oakland is doing something right. And of course, in general, we know what that 'something' is and its not traditional baseball purism.
To some purists this movement is bothersome. Why? Does the movement's ideology take away from the game somehow? I suppose it's the omission of the 'traditional' leadoff man that irks some or maybe seeing someone like Scott Hatteberg playing first base instead of the homerun threat that has to be, rightfully, playing that corner infield position. Maybe it is the lineups that don't seem like that they're purist assembled.
But seriously, look throughout the years and really investigate these purist thoughts. How many teams in any given season had a prototype leadoff hitter as compared to today? How many had thumping first basemen? Okay, there probably were more thumping first basemen in past seasons but could it be possible that the thump has just been better distributed to other positions? Perhaps some are even playing shortstop if you can believe that! And how often are the same lineups used during a season anyway?
My main thrust is this: yes, it is a beautiful, random game played by [sometimes overpaid] human beings, but, if you think that I, as an Athletics' fan, want to go back to the days of being competitive for one to two seasons and then suffer through two to three sub .500 seasons, then please take the purist in me behind the barn and shoot him dead!"
Oh, by the way, that was a curious way to apply wording to kick off a diary. Just curious, but at any point during the conversation between you and your friend did he happen to mention that a woman had hurt his pride?
by LowcountryJoe on Aug 22, 2006 4:46 AM PDT 0 recs
To be honest
by baseb3383 on
Aug 22, 2006 5:07 AM PDT
up
0 recs
If you're going to rely on batting average
by booya on
Aug 22, 2006 3:24 PM PDT
up
0 recs
Funny thing is, that's what this is about
Funny thing is, he [Beane] is actually staying away from overvalued statistics. He is, like any other GM, trying to fill holes. There are many ways to do that, whether you value physical tools or statistical analysis more. This years team is built on pitching and defense and the hitting is lousy. To say he's using statistics to economize by stockpiling fill-in players isn't really doing his job justice.
The whole point of the book Moneyball was to make judgments about the tradeoffs between those skills or attributes that are overvalued/undervalued -- overvalued/undervalued in the form of the player-salary costs associated with paying for said skills and attributes. The tradeoffs are the key! Which doesn't necessarily mean that Beane stays away from an overvalued skill; No! What it means is that the calculations change and with them so do the consequences of the choices between these valued skills.
As to your point about filling holes: have noticed what Michael Lewis titled his chapter when he introduced and started discussing Hattegerg? And the way the current team is built resembles a old, old school pruist's dream does it not [This years team is built on pitching and defense...]?
Oh, and to say that econometrics have not been a part of Beane's method to compose an economical yet competitive team is an equal injustice to those that respect the 'stat heads' for finding the right balance tradeoffs by sifting through data and plugging the data into statistical models.
by LowcountryJoe on Aug 22, 2006 7:16 AM PDT 0 recs
I agree with you
by baseb3383 on
Aug 22, 2006 8:33 AM PDT
up
0 recs
I'll agree with that
by LowcountryJoe on
Aug 22, 2006 10:10 AM PDT
up
0 recs
To a point
by baseb3383 on
Aug 22, 2006 12:53 PM PDT
up
0 recs
So, the new statistical methods, in your view,...
Maybe your contention is that not all statistical methods are equal and that focusing too much on some stats have failed to yield desirable performance results. Please correct me if I'm putting words in your mouth.
My point is, and always has been, that the yield should not be measured from player to player but instead on what the team accomplishes together in the win/loss column. On that score, the Oakland Athletics have done well for an unfluky-like period of time and they've done it while consistantly staying within a stable and sustainable budget. Can this be explained away without mentioning the role of statistics? It should be obvious to everyone that baseball purism is not what's predominatly at work in Oakland.
by LowcountryJoe on
Aug 22, 2006 1:14 PM PDT
up
0 recs
Well
Finally, I don't think the A's have some sort of secret winning formula. It's a consistent philosophy that has been applied over a long period. Stockpile draft picks and cheap prospects, give platoon players a chance to prove themselves, and take chances on players with injury concerns are all hallmarks of our front office. It doesn't just stop there though, the A's minor league system is top notch. We seem to also do a great job of player development at the lower levels. It's a combination of things that have led to continued success. Despite this, beyond this year there are some question marks as to whether we will be able to sustain this above .500 phase. The last few years haven't been very challenging to the front office given the gluttony of pitching talent we landed in Zito, Mulder, and Hudson (and all the pieces we got in return).
by baseb3383 on
Aug 22, 2006 1:28 PM PDT
up
0 recs
A gluttony of pitching talent
by iglew on
Aug 22, 2006 4:47 PM PDT
up
0 recs
A company declares losses instead of profits
...the next year it makes adjustments or sits back and accepts losses yet another year.
The only valid measurement for any business or MLB team is did we make money and how can we make more...
money?
As you break this down for a team you discern the effect winning will or won't have on reciepts;
- Media money (yes = more viewers)
- Revenue Sharing (no effect)
- Gate (yes)
- Concessions (yes)
- Licensing; A's apparel, royalties, "MacAfee", joint ventures, etc (yes)
- Stadium, homes, and "village" construction. (yes)
While all six items affect income/profits to some extent I would think #6 is the driving force for an ownership that has made it's profits this way for decades.
At ownership level: the stats of the profit/loss statement are keys to the A's roster being an asset or losing proposition, and what options are or are not used.
Tell me the dollars differences on income if the A's make the playoffs by a game, 10 games, or miss them by 2 games... Is there a carryover over to the next season? ...What ever those differences are they affect the strategy throughout the season and the following season. Adding a bat or starter or passing due to "costs" are all valid options.
Of larger effect; What will this ownership do if they DON'T contract on building a Stadium while enjoying the player salary subsidies their first 5 years as owners?
At team level: I have to think cumulative totals are sought. Obviously team average, obp, slg, Hrs, ERA, and omni-potent runs, all with an eye on wins is sought. At team roster level there is probably some concern over Perez's offense and much rejoicing over Frank's.
This years Runs totals indicate problems.
Take out the games Vs the lowly M's (1 loss), and we are lagging way behind our potential playoffs opponments. Regardless of the A's individual offensive, defensive, and pitching stats the only stat of meaning and consequence is did you win or lose?
How it happens can be magic, and as a spectator I much prefer magic to logic.
Go A's!
by A s Eh on Aug 22, 2006 7:28 AM PDT 0 recs
confession
by colin on Aug 22, 2006 7:43 AM PDT 0 recs
Oh math
by baseb3383 on
Aug 22, 2006 8:47 AM PDT
up
0 recs
Youkilis...
I honestly don't know how good or bad his defense is. And I can't see how your statement is common sense when it's very general and tells me almost nothing.
by Sharon on Aug 22, 2006 8:44 AM PDT 0 recs
"common sense"
by monkeyball on
Aug 22, 2006 10:37 AM PDT
up
0 recs
Youkilis' defense
They have him at a Rate of 105 this year at 1st, 104 for his career; at 3rd the have him at a rate of 107 for his career. For comparison, Eric Chavez's Rate this year is 114.
The Hardball Times gives him 2.0 fielding win shares; Swisher has 2.2 fielding win shares.
by rfloh on
Aug 22, 2006 11:38 AM PDT
up
0 recs
I love the numbers....
The numbers of baseball are a great way to butress or rebut my theories about different players and teams. I like to think of each stat as offering a different perspective of a player; by trying to weave them together you can come up with a 3-D 360 degree view.
Particularly useful because I don't get to see every game by every player.
by BleacherDave on Aug 22, 2006 9:03 AM PDT 0 recs
Lies, Damn Lies, and Statistics
With all those different stat values, one is bound to contradict the others.
(I for one hate the stat showing how many career GS a battter has in that situation - talk about small sample size and stats so old they might not be relevant!)
So I will agree that casual stat usage often leads to incorrect presumptions, just like looking soley at physical form (ie nice swing) may miss important results (nice swing, but strikes out a ton.)
by MobiusKlein on Aug 22, 2006 10:28 AM PDT 0 recs
The likelihood of statistics abuse
If your goal is to try to predict something unknown -- like how to best win a ball game, or how to cure cancer -- then you have absolutely no reason to twist statistics in your favor. That's because you have no ax to grind. All you want is the best result, and you'll use statistics as a tool as far as they really are helpful. If ZORP really is a better predictive tool than BEBOP in evaluating a player, then you're going to use ZORP -- not out of advocacy, but simply because it works.
If, on the other hand, your goal is to win an argument -- to "prove" that this guy is better than that guy, or to persuade consumers to buy your product instead of the other company's -- then you have plenty of reason to bamboozle the listener with phony statistics.
A lie is a lie. Statistics are just a tool that can be used for truth-seeking or for lying.
by iglew on
Aug 22, 2006 4:58 PM PDT
up
0 recs
Once caveat
by MobiusKlein on
Aug 23, 2006 9:26 AM PDT
up
0 recs
Yes, but that's very different than
The caveat that I would point out is that even among cancer researchers (or any other observation-based discipline) there is some bias due to scientists wanting to prove their materials/methods more correct or more likely to lead to the desired solution. The motivations are myriad; research grants, professional development, ego, a desire to promote one's work so as not to see one's professional accomplishments go by the wayside. Despite this, cancer research is generally going pretty well.
I suspect that baseball research is prone to similar problems, but on the whole, analytical research in baseball is still uncovering truths. I just think it's better to acknowledge that the problem probably exists.
by salb918 on
Aug 23, 2006 9:51 AM PDT
up
0 recs
well, obviously what we need to do is ...
by monkeyball on
Aug 23, 2006 9:55 AM PDT
up
0 recs
Of course the bigest use of stats....
by MobiusKlein on
Aug 23, 2006 10:55 AM PDT
up
0 recs
Right on.
Really? What's the league average for that situation? Is Joe Ballplayer being pitched around? When people have an intuitive distrust for "new" stats (or any at all), it is because they don't know how to contextualize the information. People are comfortable with batting average because everyone knows that .300 is a good batting average. But most people who think that VORP or Win Shares is a crock of bull don't have a way of contextualizing the information. That's not the fault of the statistics.
by salb918 on
Aug 23, 2006 11:32 AM PDT
up
0 recs
I don't like organic produce and oat bran
by monkeyball on Aug 22, 2006 10:33 AM PDT 0 recs
Yuck, I can smell the aftermath of that from here
No doubt that such a combination would create some wicked wind -- the kind of wind you refered to in your unstatesman-like reply in today's DLD -- wind that brings on the giggles.
By the way, the dacron shirt and pocket protector are signs of progress and hipness, get with the times you closed-minded knuckle-dragger.
by LowcountryJoe on
Aug 22, 2006 10:47 AM PDT
up
0 recs
Ah ha!
by peanut gallery on
Aug 22, 2006 11:37 AM PDT
up
0 recs
<inhales deeply>
by monkeyball on
Aug 22, 2006 1:08 PM PDT
up
0 recs
In August, yes!
by peanut gallery on
Aug 22, 2006 4:06 PM PDT
up
0 recs
phenomenal
by Cutthemullet on
Aug 22, 2006 10:43 PM PDT
up
0 recs
Question re pitchers and wins
by kkdaz on Aug 22, 2006 12:50 PM PDT 0 recs
Wins and Losses
by baseb3383 on
Aug 22, 2006 12:56 PM PDT
up
0 recs
Re:
by AsGirl on
Aug 22, 2006 2:19 PM PDT
up
0 recs
Wow
by baseb3383 on
Aug 22, 2006 2:26 PM PDT
up
0 recs
yes, but
by vk on
Aug 22, 2006 9:30 PM PDT
up
0 recs
Further question
by kkdaz on Aug 22, 2006 1:51 PM PDT 0 recs
It will be the pitcher of record...
by AsGirl on
Aug 22, 2006 2:06 PM PDT
up
0 recs


