Add new comment

I guess where I run into a sutmbling block regarding your post, Doc, is that it's not a minor thing you're sort of hand-waving off when you comment on RF/9.  You say "well we could argue about real chances bla bla bla...the 16/15 ratio is the real thing"...my response would be...we *have* to argue about real chances because RF is instantly flawed by the real-world limitation that there are 27 outs in every game...whether you're a bad fielding team or a good one.  It's not some insignificant theoretical adjustment the fielding metrics make to account for chances to make plays...it's a real and necessary one...and it's often very very large.  The reason Jones won the GG this year is that there are 30 voters in the AL who think RF/9 is a good enough measure of defensive prowess (or # of Web Gems on ESPN) to base their decision.  But we know that both of those things can be and routinely are severely biased.  Bad fielders make lots of great plays sometimes-Betancourt sure did.  If you have to make a heroic effort to get a ball that Ichi-san gets to at jogging speed, you're a bad outfielder compared to Ichiro but you'll look more spectacular.
And real chances...the Orioles *SURELY* had more gappers to run down than the Mariners.  Their pitching was TERRIBLE while ours was average-solid.  That means something REAL...it's not some imaginary math-geek tweak coming from nowhere-land, Doc.  The reason you can't use RF/9?  Because the real voters use it and it gives you the wrong answer...but a LOT...every year...routinely!  Because the range of RF/9 for full time players is smaller than the biases that throw if off (GB/FB distribution, K rate, park factors, pitching skill, hardness of batted balls).  You love James more than most...perhaps even more than I do...and he's one of the most important reasons I got into Sabermetrics.  Listen to James...he's calling to you to read his Win Shares book where he talks about why RF is a TERRIBLE way to evaluate fielders...it's not a piece of information...it's a piece of MIS-information.