THE BOOK--Playing The Percentages In Baseball

Hmmm… there was no mention of Matt, who is also tremendous.  As well as Tim, who is pretty much brand-new to me.

If you click on the Russell Carleton link, there’s an interesting article on home field advantage.

I was a little disappointed that Matt wasn’t highlighted a bit.  I’ve enjoyed most of Matt’s articles.

For me, I’m only getting to an article/month, so no need to shout-out that infrequent of a contributor.

By the way, the one major flaw to QERA is to try to mix in differing denominators.

http://www.baseballprospectus.com/glossary/index.php?search=QERA

This is the equation:
QERA =(2.69+K%*(-3.4)+BB%*3.88+GB%*(-0.66))^2

K% and BB% are per PA, while GB% is per BIP.  Imagine, therefore, if BIP = 1 and PA = 100.  Clearly, it is irrelevant if that BIP was a GB or FB.  It will have very little impact to the final ERA.  Virtually the entire ERA will be driven by the K and BB numbers.

I had proposed my own bipERA version, along these lines:
(K-BB) + (GB-FB)/4 + (Pop-LD)

It was on the idea that the spread in run values between K and BB is 4 times larger than the spread in run values between GB and FB.  And Pops are akin to K and LD are akin to BB.

Some variation of that.  One day, I should really sit down and work it out what it should be.

Also, from the commenters:

Kevin: I’m sorry, but you buried the lead.

Based on Kevin’s response, he did not realize that the reader is suggesting that the lead story is that Joe Sheehan left.  It would be like MGL is leaving, but PizzaCutter and Colin are coming on board here: in no way would I put those two events in the same article.  Each deserved its own prominence.  Joe’s readers deserved far better from their investment in him than the eulogy he got.

This reader made it more clear:

Kevin,
That was a lame sendoff to Joe Sheehan, who is the best baseball writer around in my opinion. There are other writers at BP I like but like a number of other subscribers have already said, a lot of articles now read like homework.
The content ain’t as fun as it should be.

Thanks, Nick

Agreed on the lame sendoff. 

As for the homework aspect, I like the analogy.  I think this is a feature, not a bug.

I was this || close to not renewing my subscription to BP. Their stat-cred was in a real decline.

But now I’m glad I kept it. It seems like they have been listening to the criticisms and have taken bold steps to correct it. Their new stat heavy lineup is something to be excited about. With multiple stat guys they are not so vulnerable to a MLB poaching and being left with no hard core saberists.

Looking forward to ‘10 baseball season.

I’m in Matt’s boat. I was only renewing it because I love Kevin’s work. The rest could have been found elsewhere. But now that they have Eric, Colin, Tommy, Matt, etc. etc. I’m excited for the new BPro.

This year, in addition to THT’s annual I bought BPro’s annual for the first time. It sounds like next year’s annual could be historic for BPro with these great minds and Hit F/X & Field F/X coming around.

In the column, he wrote that BP is working on new, better metrics.  I’m just wondering if they’ll come up with something better than what’s out there already.  For example, will SIERA be better than xFIP?  I know they like to promote their metrics ("deadly accurate PECOTA projections").  I’ll be curious to see what they come up with.

I wrote this at Primer:
http://www.baseballthinkfactory.org/files/newsstand/discussion/bpro_goldstein_state_of_the_prospectus/

Colin: dude, I can’t wait for your followup to this article.

***

I’m sure there are tons of Primates pulling for you.

***

And PECOTA, for instance, is no more or less “black boxy” than any projection system outside of Marcels that I’m aware of. It’s certainly well more open than several other projection systems

There are two objections that I have:
1 - Not a single person has ever shown any evidence for the PECOTA percentile ranges being accurate.  For example, a rookie and a veteran in his late 20s both have similar kind of range expectation, even though the range expectation should be tighter for the guy you have more PA or IP for.  Reliever ERAs fluctuate far more than starter ERAs, but you wouldn’t think that looking at PECOTAs.

PizzaCutter, Matt, Colin?  That’s a challenge.

2 - The “deadly accurate” claim on their (soon to be your) book every year is baseless. The way I describe it, standard systems are 81-81, Marcel is like being 82-80, better systems are like an 83-win team, maybe 84.  That’s as far as it goes.  There’s no 90-win or 95-win team out there.

Other than these two points, there’s no issue with PECOTA.  If you hold PECOTA up to be more than that, then that’s where the takedowns are coming from.  If I’m given MGL, Chone, ZiPS, and PECOTA, I’m pretty much weighting them 25% each.  I might go as high as 30% for one of them, but no more.

I think criticising them for something on the cover of a book is weak sauce. Reading book covers is going to sound more like Billy Mays than Steven Jay Gould and likely done by the publisher.

It’s not just on the cover of the annual that they talk up PECOTA.  In the articles on the site they often refer to what a great projection system PECOTA is. 

I realize they are in the business of making money and PECOTA is one of their big draws, but PECOTA just hasn’t been shown to be any better than the others that are out there.

LJ, I think it’s perfectly fair for Tango to direct a comment like that at me, the guy who once subheaded an article with “A ‘deadly accurate’ look at why you shouldn’t believe PECOTA when it comes to Matt Wieters.”

If I dish it out, I should be able to take it, right? And if I argue that things can be done better, well when I get the chance to get in there and do it myself I damn well ought to do better, right?

I don’t really see it as “us versus them” sort of thing here, mostly because I could hardly be expected to remember who’s supposed to be “us” and who’s supposed to be “them” at this point. Everyone’s an “us” as far as I’m concerned.

Let’s be honest - I’ve racked up over 900 comments on this site. I’m no stranger to ANY of the regulars here. So they’re going to be straight with me about what they think and why. And that’s fine. If I do what I’m supposed to, when I’m right I’ll defend my position and when I’m wrong I’ll correct my position.

And, yes, the percentile forecasts. That’s something I’d like to study. (Although frankly the seven-year forecasts interest me more.) I can’t promise you when this will happen - in addition to all of the other stuff I’m working on for BP, I have things like a day job and a family to tend to as well

Awww, Tom, you’re so sweet.

And, yes, the percentile forecasts. That’s something I’d like to study. (Although frankly the seven-year forecasts interest me more.)

Me too.  Even just a system to archive the old PECOTA cards without search functionality would be an improvement. 

I’d love to see the same for the old minor league Davenport Translations including his Peal projections.

I guess I have a different take on the changes at BPro. I don’t want to see it recommit itself to sabermetrics. Instead I want to see it fully embrace what it had become...a place for semi-mainstream “statistically informed” baseball writing and analysis.

It’s been quite awhile since I made the switch to HBT, Fangraphs, and other free sites for the kinds of things that BPro seems to be trying to do now. I don’t want cutting edge stuff going behind a pay wall. (Even though I am a subscriber, I still like to keep as much of the good stuff free as possible!) I would have been thrilled if Goldstein had come out and said that they were leaving the research to others and would be abandoning the homework style pieces in lieu of more easily digested bits of baseball analysis (based of course on the excellent work being done at the other sites.)

Admittedly this is a guess and I would hope that the BPro crew would have a greater feel for this than I, but isn’t it a mistake to think the silent majority of their subscriber base wants more homework style pieces? Doesn’t that limit their potential audience?

Brian, you seem to definitely echo the feeling of the commenters of that thread.  Take for example this thoughtful post from Tarakas:

I want to note that I am sad to see Joe go. I’m a longtime subscriber, and increasingly, there seem to be mainly two sorts of articles at BP--articles focusing on increasingly marginal statistics using highly complex math, that don’t really argue anything or come to some major conclusion, but instead spend paragraphs giving complex methodology to measure something, followed by a list. The list is sometimes merely a short list of leaders and trailers. If the teams/players you are interested in are neither, you never even see where they rank or rate. I think there is a place for work like this, but it is losing my interest. For example, reading pages of math on a computer screen to find out the Colorado Rockies were three runs better at something than the Reds last season, well, I’ve forgotten it an hour later.

The other type of article is the statistically informed opinion piece, filled with attitude and wit. I love these--this is mainly why I subscribe--and Joe and Christina are the main sources of these, and with Joe gone, that worries me.

This is a gross overgeneralization--Will’s pieces don’t fit into these category, and are great, and there are other exceptions as well. But overall, I’m worried.

He got a rating of “+122”.  That means 122 more people agree with him than disagree.  If you look at other highly rated posts, they all follow along the same lines.

***

So, it’s quite the challenge to not turn off the people who have come to accept BPro for what it was, even if it was different than what BPro made a name for itself in the beginning.

Brian/16 and Tango/17, here are a couple comments from Kevin Goldstein’s chat yesterday:

Tex Premium Lager (NJ): Clearly you don’t like to address comments that are critical, but the State of BP comment by Tarakas is almost at a +50 rating!! Please address something that so many clearly agree with!

Kevin Goldstein: I’m happy to. We think the new content will address a lot of your needs, but at the same time, for every email I get saying what Tarakas does, I get one saying the exact opposite, and how we need more hardcore statistical stuff. We’re trying to make everyone happy, and I absolutely love our plan for 2010 to do so.

Randy (Ann Arbor, MI): Is the level of interest/chagrin at Joe’s departure actually flattering on some levels? I’m thinking of the sense that “boy, if people are taking our website this personally, we must be doing something right?”

Kevin Goldstein: Absolutely. I think one of the best things about BP is just how passionate our following is. I expected nothing less than a lot of people with strong opinions, and my email inbox is filled with things you haven’t seen. And it’s all really important to us—we want to know what you want from BP. Of course everyone is going to miss Joe, including all of us, but now we have to focus on moving forward and I truly believe our subscribers are going to be thrilled with the end results.

I agree that Brian’s take is an interesting one.  I guess one concern would be that even though their stats are not cutting edge, that they nevertheless seem to feel like they need to be able to say that to differentiate themselves from the more mainstream baseball writing that they would be leaning towards.

The comments have certainly been interesting, but I’m not sure that I’d read too much into the direction.  It’s possible that if they had dumped the stats work, then all the people who were upset about that would be the ones commenting.

We’re really only looking at 122 (or whatever) people out of a 20k subscriber base.  Even if the hardcore stats people are a minority of the subscriber base it wouldn’t be too hard to drum up 122.

And as many posters here well know, statheads who feel slighted take a backseat in their zealous protests to no one.

Randy Brown at Bpro said:

“Now, my impression of Tango is that he is unafraid to be critical of front offices, other saberists, farm animals, and little old ladies if he feels that the criticism is warranted. So for him to endorse the new authors means something to me.”

Such nice words, I got a good chuckle out of it

Mike: interesting.

The best way I would think is to change the layout of the site so that there’s more of a separation.

Here’s what I would do: ask each reader to fill out their “profile” for the kinds of articles they like, as well as authors.  Then, you can set up a personalized feed so that the articles that are most interest to them bubble to the top.

This way, if someone is really anti-Matt Swartz, then it’ll appear at the bottom of his list.

Personalized home pages (like My Yahoo homepage).  Everyone wins.

I’m not sure BPro can uphold their subscription fee if they don’t have the hardcore saber articles and stats.  Would anyone pay for what Fangraphs provides on the blog front right now?  Probably not many, even though it’s awesome.

But you can do both, for sure.  And BPro’s mention of more blogs and more authors contributing short snippets of commentary to the blogs addresses the “saber-slanted mainstream writing” category.

I’ve subscribed to BP for several years and am super encouraged by the changes that seem to be happening.  One of the things they mentioned in the “State of the Prospectus” article was about a new interface.  I have high hopes for a new stat interface at BP but that site, more than any other, inundates me with stats that are outdated, misleading or unverified. The only stats I use there right now with any reliability are the baserunning metrics that Dan Fox created.

Personally, I’d like the writers to not just stop using things like FRAA, which is irreparably broken/obsolete but get it off the site—stop publishing it altogether.  There’s no use for it.  It (and other equally useless stats) cause unnecessary clutter and distraction when I’m trying to navigate their statistics.

When reading the Mike Silva question about stat saturation, BPro immediately comes to my mind.  The interface and the volume of stats that I really don’t even use anymore creates a sense of “saturation” for me.

Tango,

BPro is being overly sensitive to criticism coming from the sabermetric community. It’s a bit puzzling considering Kahrl and now Goldstein have been in charge recently.

I mostly agree with the comment you posted from Tarakas, though I personally enjoy reading articles focusing on “increasingly marginal statistics using complex math” even if I don’t understand all of them. I would rather BPro concentrate on writing and analysis, and let the sabermetric community at large perform research and develop metrics. Once the community has separated wheat from chaff, incorporate the research into BPro lexicon. My fear is that BPro resources are going to go into metrics and analysis that are no better than what already exist (anyone up for WARP 4?), or worse, a bunch of black box pieces that may actually be worth my $40 if only I was able to tell what they were actually doing. BPro has shown that they are not interested in peer review.

I don’t see how this doesn’t erode their market share. I also don’t see how this benefits the baseball community. The people they are adding are terrific. I just fear what constraints they will have to work under and what will happen to their work behind a pay wall.

@Sky #22:

“And BPro’s mention of more blogs and more authors contributing short snippets of commentary to the blogs addresses the “saber-slanted mainstream writing” category.”

This is an interesting development to me. However, I will remain skeptical that these blogs will be updated regularly. The last 3 BP Unfiltered posts were from 12/30, 12/29, and 12/24. Taking a quick look back, they seem to average 4-5 posts per week. Additionally, BPro scheduled chats are becoming less frequent...a not insignificant percentage of the time the author doesn’t even bother to show up.

Still, even if the new blogs are executed well and updated frequently, I’m paying $40/yr for feature length analysis and opinion articles. 200 word blog entries ain’t going to open my wallet.

I’ve got to second what Sky mentioned above. I don’t think BP could charge for content that’s more akin to what you’d see on BtB, THT, or FanGraphs. There needs to be something unique, and before the flood of available resources that we’ve seen recently, the unique part of BP has been its statistics, which at the time were on the cutting edge.

My favorite part of this new BP change is that I feel a bit better about the vetting process for their work. After many of the guys in the forefront left (Woolner, Silver, etc.), BP had Clay as the only guy with a strong stats background, and his voice was more behind the scenes than anything. Now they have guys like Seidman, Dr. Cutter, and Colin to explain to the readers what they are doing. It’s significantly better than having people who were more on the analysis front try to explain how and why their stats work.

Not sure what you mean.  The statistics section on BP have always been free.  Other than PECOTA, that is.  What they’ve charged for is the writing.

There are a few other key fantasy features - like the depth charts or the PFM - that are pay-only as well. But yes, stuff like EqA, VORP, EqBRR and so on have always been outside the pay wall.

Unless you know something I don’t, nobody’s suggested that will change.

I don’t think I’ve lost my membership card in the overall sabermetric community, BrianK. I still plan on doing a lot of the same kind of work I’ve been doing at THT, and before that StatSpeak, and before that my own little blog.

Now maybe I (or Russell, or Eric, or Matt, or all of us - and more - collaboratively) can’t do anything better than what anyone else is doing. But it won’t be because we’re not trying or because the suits have Harrison Bergeroned us all and shackled us to keep us from doing good work.

Colin,

Nobody ever accused me of knowing anything, lol.

I’m sure you (and the others) will continue to do the good work I have come to expect. I just see it becoming less accessible. I’d hate to see you guys working in a silo.

Good luck and congratulations of course!

Good for BP. Let a thousand flowers bloom.

Colin, What is PFM?

PFM is Player Forecast Manager. Lets you produce dollar values for auction based upon your league settings and PECOTA projections/depth charts. I think it also works for straight drafting as well.

PFM = Player Forecast Manager

It’s a fantasy draft tool that assigns dollar values to players based on the PECOTA projections and league parameters.

FWIW, I would love to see PFM be adapted to limiting the pool of players. For instance, my league does free agency. So the only people up for auction are the free agents. But I can’t do that with the current set up (I don’t think).

Ah.  Similar to what Mays Copeland does:
http://www.lastplayerpicked.com/priceguide/

JD/34, with PFM you can designate players as being unavailable for auction, e.g., on keeper lists, and it will adjust dollar values accordingly.

Hey Pizza, you didn’t tell me you and pbconnection were best friends?!

LJ:

I think criticising them for something on the cover of a book is weak sauce

http://baseballprospectus.com/subscriptions/

You’ll receive:

* Complete depth charts and forecasts for AL and NL pitchers and hitters using Baseball Prospectus’ deadly-accurate PECOTA projection system--the same one used in MLB front offices.

Can someone tell me who exactly is responsible for that bullsh!t, and why is no one involved at BPro bothered by “Fair and Balanced” type of claims?

Remember, bullsh!t = summary opinion without evidence

My projections are also used in MLB front offices, but I haven’t yet claimed deadly accuracy.

I know at of at least one team that looks at mine.  Probably others do as well but I don’t have contacts with.

No money made out of it, hard to charge teams when I put them on the web for fantasy owners to use for free, but if they are at all useful to multi-million dollar businesses, a donation would be appreciated.