THE BOOK--Playing The Percentages In Baseball

It’s behind the scenes where it gets complex, and why Davenport Translations have never been seriously contested. Unlike attempts at the “One True Stat” like VORP, or Runs Created, or WARP, or Win Shares, all with their various degrees success or failure, translations seldom raise any significant argument among serious statheads, and no one has developed a competing system.

I think they’re kidding themselves a bit here.  First, it’s extremely hard to “seriously contest” a method that’s not public; sure, they have written vague articles about how their translations work, but the nitty-gritty is tucked away in a black box.

Second, I can only speak for myself, but I am not particularly interested in the question of how many home runs Ty Cobb would hit today.  To the extent that I am interested in it, I think that there are too many factors in play to take any methodology too seriously, and so I have no reason to examine closely or object to the DTs, or the Willie Davis method, etc.

They also say that translating stats has been a bedrock of the BP, or something to that effect.  This may be true, but in the annual BP book, these translations are between parks or between levels of pro ball, not between the deadball era and today.  I can accept their translation of how Ryan Braun would have hit in the majors last year as fairly reliable without carrying that faith over to cross-era translations.

I wanted to write what Patriot wrote, but I couldn’t say it very well.  Since he did such a fine job, I’ll simply sign my name to the bottom of his post.

What he said.  Also, it seems a mite suspicious to use a standardized park effect translation in the case of say, Mel Ott.  Did they somehow neutralize his home/road split in his own era and then translate that number to today’s game of smaller parks and power hitting, or did they just take his career totals and increase them?

When I read the part of the article that Patriot quoted, I immediately did a double take. This is exactly the kind of thing that, whether fairly or not, gives BP a reputation for arrogance and ignoring outside research (for what it’s worth, I think that BP has mostly done a good job of acknowledging outside work recently). Michael Schell released a whole book on translating hitter statistics into the modern age and Sean Foreman has neutralized stats up on B-R...and by the way, even if no one has ever developed a competing system, how does it improve this article to spend a paragraph bragging about it?

I find Carroll to be very arrogant all around; much more so than BP as a whole. As for the DT’s, I’m with Patriot.

I’m really not sure what these translations are trying to do.  Let’s forget about the absolute numbers for a second.  Are they trying to ascertain what Ruth, Gehrig,, Aaron, etc., would do, relative to one another, if they all lived in the same era and faced the same pitchers in the same park, etc.?  Are we to believe that Ruth would outhomer Bonds?  I have a hard time believing that?  Or are they trying to figure out how much better each player is than their competition?  If that is the case, then I have no problem believing that Ruth was much better than his competition than was Bonds.  And they really lost me with the Gehgig doubles and triples thing.  That sounds more like a projection methodology which is a lot different than what they are trying to do in terms of translations - I think.

Bottom line is that I don’t think I give a rat’s ass about all this stuff.  And it is almost pointless to discuss their methodology until they, one, explain exactly what it is they are trying to accomplish or what question they are trying to answer, and two, their exact methodology.

Tango, when you link to a BP article, how do you get such a nice page as opposed to the page that would come up if you went directly to their site?  And how do you get to link to their premium content?  You have permission?

There isn’t much to Davenport translations.  They are a collection of mindless formuli that tell you George Hall, a 142 pound player in 1876 “translates” to 60 homeruns.

If that’s the answer then I don’t want to know what the question is.

MGL:

That particular article was not flagged as “premium”. 

And there’s a “printer-friendly” link at the top right of all these pages (same applies to Hardball Times, NY Times, Washington Post, etc).

***

We’ve discussed the different points-of-view in the past about living-in-same-era.  You have to choose:
- do you make Babe Ruth born 70 years later, and allow him to get the benefit of whatever environment his peers got
- do you magically transport Babe Ruth 70 years into the future as an adult, so that the 30yr old Ruth in 1925 is now playing in 1995

On the flip-side, rather than breaking the space/time continuum for every player, I break it for exactly one player.  I take some guy, say Hubie Bloomquist, a completely average player in every aspect for 2007, and magically transport him to 2006, 2005, 2004, 2003… 1919, 1918… Then I ask: how would this player do against Ruth’s competition in 1921 in Ruth’s parks?  This player being born in say 1980, has the benefit of our science.

The reason I prefer this method is that it’s far easier to speculate how such an ordinary player would do anywhere, than to speculate how even Manny Ramirez himself would do at Coors, before even considering the timeline adjustment.  With the timeline adjustment (meaning completely different types of pitchers and parks), that’s realyl tough, as clearly Manny and Juan Pierre couldn’t possibly be affected the same way.

(Note: Text in italics edited in from original post.)

***

DSG/4: Sean’s translation is the Bill James Willie Davis method.

I think the DTs represent a strong effort at trying to put every player into the identical environment. Obviously, there are tons of hazards that render a high degree of accuracy impossible, but true men can but try. There’s not a ton of accuracy that can be attained through forecasting, yet I’d argue it’s still a worthwhile exercise. Unlike forecasting, this type of exercise doesn’t have any bearing on analyzing baseball today, but I certainly would argue that exploring issues of league difficulty over time is interesting, if nothing else.

That being said, the portion of the article quoted by Patriot reeks of arrogance and there’s a big difference between a good-faith effort and being above reproach and/or impregnable. Attitudes like this give sabermetrics and its devotees undesired and undeserved reputations as know-it-alls. And yes, the DTs should be open source for the reasons stated above and also because BP does not charge anyone to view them.

It’s notable that A-Rod leaving Yankee Stadium for his home games might actually hurt his chances at the translated career title. That’s because park effects are a big part of translating performance, and help the numbers of those players who have to hit in pitcher’s parks.

This part confuses me. Let’s say A-Rod hits 50 HR this year in Yankee Stadium, which DT’s say translates to 55 HR in a neutral environment. Shouldn’t the system then be confident that, if A-Rod were transported to that neutral environment, that he would then hit 55 HR, in which case his move has cost him nothing in translated HR?

Good catch.