.

=== A Brief History of Time ===

From 1977 to about 1995, sabermetricians taught us to focus on component skills.  For example, Mat Olkin preached that "Olkin ERA" (OBP x SLG x 31) predicts ERA better than ERA predicts itself.  ... Slugging percentage predicts RBI better than RBI predicts itself.  And so on.  

I think it was Tony Robbins who went race driving with a sensei ... the sensei taught him that when there is a crash, you've got to look at the track line rather than the cars flying around, because the line will open up.  Maybe they were using a simulator.  Anyway, after two or three failures, here came another simulated crash and Robbins (?) was staring wide-eyed at the cars sailing around him ... sensei reached over and physically grabbed his head and twisted it to look at the line.  After that he was okay.

It's sooooooo hard to do this.  Remember Michael Pineda's first spring training game?  He gave up five runs -- but had a terrific K/BB and a 20% swing and miss rate.  Nobody could get past the five runs, which are nothing but ERA, of course.  You know what I'm saying, right?  If James Paxton gives up five runs, but has 7 whiffs and 1 BB, people are not going to be able to look past the ERA.

Early sabes taught us to try to look past the runs gained and runs lost.  It's hard.

..............

From about 1990 to 1999, sabes taught us to look at trendlines in the component skills.  Okay, Charlie Furbush fanned 7 batters and walked 3.5.  Great.  But what did he do the year before?  If the year before, he fanned 6 and walked 4, that's one thing.  If the year before, he fanned 8 and walked 2, that's a different thing.

This is a huge leap in sophistication, and it's still uncommon to see analysts focus on this properly.  When you see an analysis of Justin Upton, you won't routinely see a Hardball Times author say, "Hey, look at the trend in his contact rate over the last four years."

..............

Bill James, Ron Shandler take the above two ideas and merge them with the concept of pattern recognition.  James came up with the idea of "Hall of Fame Pitcher Families" and it was off to the races.  Still, in SSI's view this pattern recog is still underused.  In tournament chess, pattern recog is EVERYTHING.  As Bobby Fischer put it, "Whatever position you've got it's been played before."

Baseball-reference publishes most-comparable players, as does PECOTA*, but it is the intersection of pattern recog with 5-year trendlines that is the key.

Ron Shandler is the guy who stares at trendlines, and who has grok'ked the patterns these trendlines form for certain kinds of players, and it is why he is currently the most important projection analyst, in my view.  We don't say that he's always right, or that he has the only useful paradigm.  We only insist that you do not want to be ignorant of his assessment of a player.

What he says about Justin Upton, this year, is fascinating.

NEXT

.