Game Thread: March 1, 2008

For 2008 ...

• Grady's 75° projection (.315 EqA) is right around Granderon's 90° projection (.314).
  
• Grady's 50° projection (.298) is right around Granderon's 75° projection (.299).
  
• Grady's 25° projection (.287) is right around Granderson's 50° projection (.286).
  
• Grady's 10° projection (.271) is right around Granderson's 25° projection (.269).

Yes, the CI for multi-year projections is going to be huge, but perhaps not as huge as you think.  More to the point, there is still value to knowing the mean projection for each player as a point of comparison.

A related and interesting metric is the Attrition Rate, defined as the rate at which similar players end up losing half their playing time (due to injury or poor performance).  For 2013, Sizemore has a 23% attrition rate, Granderson's is 45%.  This, along with the mean projections, is indicative of the diversity of career paths reflected in each player's pool of similar players.  Granderson matches up with significantly more players who washed out after brief careers -- although still not all that well.

What should always be kept in mind here is that Granderson is 18 months older, and always will be.  So in looking at the attrition rates for 2013, we're talking about age 30 season for Sizemore and age 32 for Granderson, which makes a difference.

And that leads us back to my original point, or at least, the point I meant to imply but should have laid out better.  Sizemore is projected to be better in 2008, that is true.  But the more significant point is that, because of the age difference, the gap between the two players will be smallest in 2008 and 2009, when their ages are 25/27 and 26/28.  As Sizemore moves into what is likely his prime and Granderson moves into (a very mild) decline, the difference in value likely will increase.  

This isn't just a simple age curve, it's also reflective of the two players being different types of talents, again, despite how similar they seem at first blush.  Sizemore is fast, but Granderson is faster, and speed is a more significant part of his game both on the bases and in the field.  Speed is generally understood to peak at age 24, meaning that even Grady's speed is already in decline -- but Granderson's speed is a bit more in decline, and approaching more significant decline more quickly, and speed is a much bigger part of his game.

Grady is the fundamentally stronger hitter in every facet, and you see this everywhere in his stats.  Despite being closer to the ideal age of 27, Granderson has only had a higher batting average once, in 2007, and only with the aid of a .360 BABIP.  Granderson's ISO was also propped up by all those triples -- there's that speed again -- and yet Sizemore edged him out in SecAvg overall, because he draws nearly twice as many walks.

I point all this out only to show that we're looking at players that look less and less similar the deeper you look, and Sizemore's "typology" lends itself not only to better performance, but to a significantly better chance of sustaining a performance level over multiple years.  And all of these issues are incorporated in some way into PECOTA's similarity scoring.

As a side note, I was shocked to see the difference between the two players vs-LHP in 2007 -- 812 OPS for Grady, 494 for Granderson.  Granderson had a .211 BABIP vs-LHP, .390 vs-RHP.  We can safely assume that won't happen again.

I have to disagree with the comparison with scouting.  For one thing, I think it's just apples and oranges.  BUt if you're going to insist on a comparison, the fact is, we have far more solid data with which to evaluate any given version of PECOTA -- 20,000 player-seasons of results in excruciating detail.  Scouting has been around longer, but how can you really evaluate it?  Does it even make sense to apply an objective evaluation to it?

The other thing of course is that statistics in general have been used in player evaluation literally for over 100 years.  It's not like people just sat around and did whatever the scouts said, ignoring the numbers entirely.  So let's not pretend this is a scouts-vs-stats thing, that would be a ridiculous distortion.