Defending Harold Reynolds

by Sky Kalkman on Jun 22, 2009 6:00 AM EDT

Harold Reynolds recently wrote an article on his blog (he has a blog?) about, uh, OPS and Dick Williams and some other stuff. You see, part of the problem is that it's awfully hard to figure out what he was actually trying to say. Harold doesn't appear to be a great writer -- which is okay, there's no need to be good at everything -- and he was trying to discuss a topic he's not an expert in, measuring player value via OPS. Ok, not a big deal, move on to something more substantial.

Except that Joe Posnanski got involved (and while his article could be read as FJM re-incarnated, he actually stated he somewhat agreed with Harold's probable thesis, the fact that OPS is a flawed stat) and when Poz gets involved, everybody else does, too, because, well, everybody reads Poz. (You don't? Go, read him, now.) And then there was merriment among those fans who would probably call themselves statheads about how bad a writer Harold is and how little he understands about stats.

Both those points may be true. But the first point is trivial -- there are tons of bad writers out there and simply writing poorly does not make you wrong, it just makes you ineffective. The second point misses the boat, too, and is a prime example of why many people find statheads annoying. You see, there's a difference between harassing Harold because he's wrong and harassing him because he doesn't understand the math to a large enough degree. The first is fine when appropriate, while the second can be elitist.

Is OPS a great stat? No. Will a slugger in a weak lineup receive more walks? Yes. Will his home runs be less valuable because pitchers will actually give him something good to hit only when it matters less? Probably. And -- allowing for some writing gaffes -- that's Harold's main point. If someone explained to him that WPA/LI accounted for the context of every game situation and gave appropriate credit for advancing runners when necessary and hitting home runs when they mattered the most, don't you think he'd agree it was better than OPS? I do. One of the largest mantras of statheads is that context matters. Well, Harold just wrote an article about how context matters.

The problem many statheads have with an article like this is that Harold used the language of a traditionalist. Yes, numbers and analytical research would have made me like the article a lot more, but Harold is still right, if you give him a little leeway. Maybe he doesn't deserve that leeway because of past analytical mistakes, but the point remains even if we move on from Harold Reynolds specifically to non-saberists in general. Speaking French while in China makes you more difficult to understand, but not any more right or wrong.

Let me make this clear: Poz's article wasn't elitist (although perhaps a bit rude) and there's a bunch of stuff in Harold's article that's crazy (heck, he uses everyone's favorite "clogging the bases"). I'm just using this situation as an example to discuss a larger theme. Many statheads (and yes, we're a collection of individuals, not a group that meets every Tuesday and issues collective PR statements) are obnoxious and elitist. And those same people can be right or wrong on a case by case basis. Attitude and correctness are two different things. Statheads often value the second one more and sometimes don't even pay heed to the first, especially when we're hanging out with our own kind. But non-statheads won't even give the second one any consideration if they're turned off by the first.

If we're all going to get a long, with more fans appreciating the benefits of the saber point of view and statheads being seen as real baseball fans, then statheads need to realize that inexperience does not equate to stupidity and that speaking a non-numbers language does not equate to being dumb. Non-statheads need to realize that many statheads aren't actually being condescending and that they'd probably agree with much of the Saber Bible if it were explained in their own, non-techy language. I realize getting two groups with different styles to realize their goals are basically the same has never been an easy accomplishment -- dare I mention religious conflicts? -- but it really would be a whole lot more fun for everybody to just talk baseball instead of having annoying arguments.

Let me stop rambling and sum things up. Bad writing does not make the analysis bad. Incomplete analysis does not make the analysis wrong. And using a different language requires those who don't speak the language to translate what's being said, not slam the language being spoken. Sometimes we statheads have to meet the non-statheads half way, instead of expecting them to come to us while we hurl grenades into the path they're walking.

Tweet Comment 14 comments | 2 recs |

Do you like this story?

Comments

I think the problem with Reynolds' article wasn't the thesis

Which, as you mentioned, is that context matters.

It’s that most of his evidence either didn’t have much to do with the topic, and the examples were somewhat contradictory or otherwise flawed.

I think you’re potentially going too far in giving Reynolds the benefit of the doubt. Incomplete analysis should not refute an overall point, but it can definitely weaken it.

I thought Poz’s fisking hit the high points, and did so in a much more fair manner than FJM (or many other sites would have). I don’t think the intent was to be mean – I think he was just flabbergasted by Reynolds’ post.

That doesn’t mean the response was a good idea; just that I don’t think there was a desire to be mean.

On the meta-topic of tone, of course you’re right. No matter what you’re trying to explain, your listeners will turn off if the first thing you say is “You know that thing you’ve believed for a long time? Well it’s wrong. And for believing in it, you’re stupid.”

Too many people “explain” that way, when in reality, they should be trying to teach, not argue.

What does being confrontational really get you? Is the important thing how you’re thought of, or how your message is received? Because the tone of much Internet dialog suggests that’s it much more important to be thought right than it is to share your ideas and make sure they’re understood.

Of course, I’m much smarter than all of you, and my way is obviously right :-)

by Dan Turkenkopf on Jun 22, 2009 8:06 AM EDT reply actions

Well said

by Sokojoe on Jun 22, 2009 10:14 AM EDT up reply actions

"Teach" is a great verb, at least without any implied condescension.

When two non-statheads chat about baseball, they often argue about stuff, which usually doesn’t result in any ego bruising, just fun. When a stathead and a non-stathead chat, those arguments have more potential to go awry. Being aware of that fact is important.

I’ll fully admit I used to engage in debates with non-statheads just so I could have the feeling of “winning” and “being right”. Then I outgrew it. It was lame. There are non-statheads who take the opposite approach and purposely taunt/bait/troll statheads. Again, lame. Let’s just talk baseball and respect each other.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jun 22, 2009 10:21 AM EDT up reply actions

Excellent Article

I was once a non-stathead myself and remember how rudely statheads reacted when I felt that Juan Pierre was a good player because he consistently hit over .300 and scores bunches of runs and stole tons of bases (of course Pierre is having a half way respectable Saber type of year and he is still one of my favorite players). Nonetheless that rudeness prompted me to study Sabermetrics and learn about how enjoyable and better those types of numbers are in analysis. All in all it would have been much more pleasant if someone had explained it to me instead of using slurs and condescending language.

Its a wonderful game. Doesnt matter if you use numbers or argue rationally about given players, its still just a game and people love it in their own ways.

by backtocali on Jun 22, 2009 12:16 PM EDT reply actions

I think what annoys me about such articles as Reynold's

and why I enjoyed FJM so much, is the tone of anti-intelligence, and anti-rationality these articles usually take. When someone says “I don’t need OPS, or any fancypants stats because when I look into David Eckstein’s eys I know hes a grinder” sorta thing, it annoys me because its far more than a baseball thing, its a society thing, people willfully ignoring science and intelligence for silly reasons. So when I get angry at Bill Plaschke, in a way, Im getting angry at the Kansas Board Of Education.
That said, OPS is a flawed stat. But its the best of those understandable to your average person.

"Sometimes Joe (morgan) doesn't like facts to get in the way of his opinions."- billy beane
"That was a great pick...if this was 2002" Me, to guy who selected Barry Zito in a fantasy draft
www.27ClubPeak.blogspot.com

by harendaman365 on Jun 22, 2009 4:25 PM EDT reply actions

Well, part of your point might be that the average person can understand a lot more than OPS if they try.

I agree with your whole post. But instead of harassing those people for acting dumb (the fact that it’s true being ignored) we should take a path that’s more productive in the long run instead of relieving our short-term stress by venting.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jun 22, 2009 4:41 PM EDT up reply actions

You underestimate your fellow man...

“But its the best of those understandable to your average person.”

That’s the kind of attitude that creates the rift Sky’s talking about…
If people can understand BA (and the limitations that number creates), they can get OBP. And if they can get OBP, they can figure out SLG. You’re telling me combining those two numbers becomes mind boggling for people? Come on…

It’s not the stats themselves (or an inability to learn them) that bother non-SABR types, it’s the sabremetrician hype that THEIR numbers are the “end all be all” of stats that turns people off. A similarly one-sided view to someone who insists a guy is “Five Tool” despite never showing the numbers to justify it.

Are advanced stats “better” than “traditional” stats? Probably.
Can they find “underrated” players who’s value would have been overlooked in the past? Of course.
But anyone reasonably familiar with numbers should recognize that ANY stat has holes and is incomplete, from BA to OPS to UZR to VORP.

by BigGreenMonster on Jun 23, 2009 11:49 AM EDT up reply actions

to better explain myself

Im saying OPS is very understandable to your average person. Nowhere did I say the average person would be boggled by combining OBP and SLG. Quite the opposite in fact. Im saying that OPS is the best of those understandable because its something that is representative of things they can actually see. VORP is probably the better stat, but its too far removed from reality to really become accepted. People like to be able to count things, to see progress as its happening. You can say, he went 1-3 and that brings his average down to whatever, but no one will be able to offhandedly say what somebody’s VORP is. I know im not getting this across so really well, but hopefully you get what im saying…?
OPS’s main flaw lies in that its a crude combination, your putting combing two unlike terms in OBP (out of 1.000) and SLG (out of 4.000). But its the best easily understandable metric we have, and if you just wanna know who the best hittter in the league is, OPS is decent enough.

by harendaman365 on Jun 23, 2009 12:41 PM EDT up reply actions

I think OPS is easily understandable mathematically to everyone. (Hmm, I add? Yeah, I can do that.)

But logically? Why add the two? Both OBP and SLG are things you want to be good at, ok. But are they equally important? What does OPS measure? The answer is nothing, directly. I would think something like total average, (TB + BB) / PA , would make more logical sense to push, even if it correlates slightly less to run scoring. Bases per PA means something. And then you can hit the question, “well, are all bases created equal?” or “how do outs play into this, if two guys have the same bases but one makes more outs?”

Joe Poz wrote an article a while ago about how batting average actually doesn’t make a lot of logical sense, we’re just used to it. If we started from nothing, we’d probably use OBP. Sure, BBs are a bit less valuable than 1Bs, but the difference between a 1B and a 2B is larger, let alone the difference between the 1B and the HR.

And Wins for pitchers? The definition of a win is so convoluted if you think about. If we called it something else that wasn’t the same word as the word for the team being victorious, it wouldn’t be nearly as popular. It’s not really that hard to separate pitching performance from team performance. No, FIP/tRA/DIPS aren’t perfect and might only be a 7 out of 10. But Wins is like a 3/10.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jun 23, 2009 10:15 PM EDT up reply actions

I know OPS is not statistically sensible

and I just finished writing an article for my school paper about how all the stats we think are important aren’t. However, I think that there’s a certain point that a stat will become removed from what has actually happened. Yes, wOBA and FIP and stats like that are great for finding out who has actually been the best hitter or pitcher, and I use them for argument settling and fantasy baseball. But they’re too far removed from “reality”. Like, throwing out a name here, if Joel Pinero (guy’s been on my mind recently) throws a no-hitter, even if we know from our advanced stats he just got lucky, dude still threw a no-hitter. And while the non-advanced metrics are generally less accurate, no one wants to be plugging algorithims into a scoresheet. Don’t get me wrong, I love the advanced stats, and I think we’d all be better off if RBI and Wins didn’t matter at all. It’s just at some point stats have to be in reality.
This is not an anti-stats tirade, so please dont FJM me :-)

by harendaman365 on Jun 24, 2009 10:17 AM EDT up reply actions

Mr. Monster and I have talked about this before.

I whole-heartedly agree that the “average person” can handle a lot more than they currently do and than some people give them credit for. The first step is to be aware of that and always have the attitude that nobody’s a lost cause.

I also agree that statheads often don’t realize the limitation of the stats they use. We need to be aware they’re not perfect and do a better job of estimating or calculating error bars. I also encourage everyone read any stat presented with implied error bars even if one isn’t stated. Many authors DO realize the numbers aren’t exact. (And can we please use a lot fewer decimal places?)

I think one place Mr. Monster and I would disagree (and this probably models the larger picture) is the degree to which the advanced stats are better than the traditional ones. Using wOBA instead of AVG might have an average error of (pulling this out of my arse) .5 wins per player over 600 PAs. Not a big deal. But it’s the 1/10 (3/10?) players where AVG really misses the boat where teams can take advantage. If you can spend the same money to gain half a win per player and there are 15 positions (9 hitters, 5 starters, 1 reliever) that really matter, that’s 7.5 wins to the smart team right there.

Or take the Hall of Fame. Over a full career, run support, ballpark, luck, etc. tend to even out so that Wins are a decent measure of performance. You don’t win 250 games just by luck. But take Bert Blylevel, or Jim Rice from the other direction. If we’re judging HoF purely on career value and talent demonstrated (which might not be the definition we should using, true) then Blyleven should be a lock and Rice shouldn’t even be close. It’s tough, using solid, objective analysis to come to the opposite conclusion. Like Harold argued context is HUGE and needs to be accounted for.

Lastly, let me revisit the idea of “speaking French in China.” If you’re talking to someone unfamiliar with VORP, you can’t convince them it’s a good stat by showing them a correlation. You’re speaking statgeek when they don’t. You need to learn the other language, where numbers aren’t convincing. Figure out ways to make statements in English that any fan can agree with. For example, a fielder can’t earn an error on a ball they can’t get to. So you need to measure lack of range in addition to botched plays in order to have a good measure of fielding. And don’t try to the discussion from A to Z. Shoot for B. And remember you’re talking baseball. It should be fun, not recruiting for a cult.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jun 23, 2009 9:59 PM EDT up reply actions

Good article Sky

I have to admit I can’t get worked up about Reynolds-type stuff anymore, even with JoPo’s nudging… I don’t know what it is…

But you make good points that reflect some of the thinking I’ve been doing (as I think I’ve mentioned to you before) about how sabermetrics is supposed to open up the game to more people, not make it more abstruse.

I'm not a sabermetrician, but I do play one at Driveline Mechanics.

by Matt Klaassen on Jun 22, 2009 7:12 PM EDT reply actions

Bravo

Thanks for posting this, Sky. I’ve been mulling over a response with almost exactly these salient points since the Poz skewering of HR, but essentially chickened out. Not sure why s many are so resistant to advancing statistical analysis into “situational” analyses, but it’s clearly crucial for better understanding.

by BobbyMac on Jun 25, 2009 1:27 PM EDT reply actions

The most obnoxious statheads

are the ones that think they know what they’re talking about, when they actually have no idea. I’d rather read guys like Harold Reynolds than people who think down upon everyone else when they themselves don’t comprehend what they’re talking about.

Not naming names, but… (cough cough http://www.bucsdugout.com/2009/6/26/926468/lets-trade-all-our-bad-players-for cough cough)

I am like your Dan Aykroyd and biglow would be Jane, the ignorant slut. -Chad

Good ol' KO

by thecoolest on Jun 29, 2009 12:03 AM EDT reply actions