Classical Conditioning vis-a-vis Casey Kotchman, 1

I/O Malcontent:  LA, Atlanta, and Boston punished Kotchman's fly balls to right.  Safeco will reward them.  Since 2007, Kotchman has hit 22 of his 32 homers in road stadiums.


CRUNCH:  This idea -- as it applies to the generic ballplayer -- is underappreciated.

Time and again, Dr. D cringes when analysts use the syntax "Ichiro hit .352 as a Mariner; playing in a neutral park, Ichiro would have hit .368/.402/.513."


=== Canadian Exchange Dept. ===

No way no how.  That is NOT what park translations imply. 

What park translations imply is that Ichiro's .352 average, in Safeco, was worth a .368 average in a neutral park -- that Ichiro's .352 average won as many games as a .368 AVG did in (say) Cleveland.*

You can say, with absolute certainty, that Ichiro's batting line in 2009 was "worth" 368/402/513.  (Um, wow.)


=== Changes of Scenery Dept. ===

But!  What Ichiro might hypothetically do, in some other park with its alternate grass, and fences, and lighting, and lineup, is strictly for the sci-fi writers.  That's an alternate reality scenario.  Spock might be captain of the Enterprise if one fanged butterfly had flapped its wings differently on Wulcan. 

In another neutral park, Ichiro might have a fence that was just within his critical perimeter and he might hit 30 homers.  Or he might have the opposite effect.

You'll see the very finest analysts say stuff like "remember what we do here.  We figure out what players would have done under neutral circumstances."

No, you'd need a gypsy to figure that out, and the costs associated are not acceptable when consulting familiar spirits and mirrors and pentagrams and stuff.  :- ) What you can figure out, is what Kotchman's performances were worth in the Canadian Exchange dollar of 1907 baseball, or 1984 baseball, or 2009 baseball in Candlestick Park.


James warned us about this a million times in the 1980s when he invented park translations:  we're not saying that Wade Boggs would hit only .308 in a neutral park.  We're saying that his .350 AVG's are only worth .308 right now.

This is a very important point! 

The assumption is that by normalizing stats, you can tell MLB teams what Joe Shlabotnik will do next year, for their team.  You can't!  Their park, their coaches and their team will change the player, making him worse or making him better.

The Mariners do not believe that Kotchman is about to hit his normalized batting line -- Safeco-adjusted.  They believe that under different circumstances, he's about to do something very different from his normalized batting line.

Different parks, different coaches, different contexts can and do change ballplayers.  We don't get to simply run their stats through a filter and say, "this is what they'd do in an alternate reality."  We need to keep an eye out for who would do better with a change of scenery.


Part 2



Nice job, Doc.
I do like the growing tendency to break down WHY park effects influence hitters (hurts lefties - helps righties, etc.), which is obviously much more valuable when examining player acquisition.
Using Ichiro as your example is hilarious.  For his career, (ya reckon 3300 PAs both home and away is sufficient sample size?), Ichiro hits .820 home - .804 on the road.  For 2009 specifically?  Home: .908 -- Road: .796.    :-)


IMO - of every baseball player on the entire planet, Ichiro is the perhaps the one I'd peg as the absolute least likely to be impacted by ANY baseball park - (even Coors).  He's gotta be one of (if not THE) poster-boy for "hit it where they ain't" in modern baseball history.
I personally find the use of Ichiro when discussing park effects sort of like referencing Ryan Leaf when discussing Hall of Fame quarterbacks. 
What I was attempting to get at in regards to Park Effects is that saying ANY player was "worth" something other than what they actually hit in a park is (IMHO) a fools errand from the start.  Parks have NEVER impacted every hitter equally.
If Raul Ibanez hits .293/.354/.479 (.837) in 2008 - my personal opinion it is absurd in the extreme to attempt to say "In a neutral park" he would've hit .300/.360/.500, because of the aggregate Safeco Park effect -- ESPECIALLY when you know precisely what he did hit in Safeco in 2008 (.276/.353/.497 .850). 
I mean - the logic (or lack thereof) just makes me take head to desk.  He actually posts an .850 in Safeco (26 EXTRA points of OPS) -- but because of the "generic" park effect, we're going to say the PARK held him back in some way?!?
Don't get me wrong.  Parks do influence outcomes.  They exist.  But, I've yet to see any case where applying them to INDIVIDUAL hitters is actually very helpful.  Seattle fans understand they have a pitcher's park.  They also understand right-hand batters take a much more severe beating from the park than lefties.  But, every righty and lefty hitter aren't identical.  A righty version of Ichiro is likely going to be just as effective.  A righty version of Branyan, however, might suffer a bit - but not by some predetermined number assigned to the park as a whole.


1.  You're aware that my point was that Ichiro would NOT hit his translation? 
And that his road numbers underline my assertion?  The entire article said only one thing: you CANNOT use translations to project road performance.
2.  We're trying to stay away from ridicule at SSI.  I don't do it to you, right?


In re-reading, now I get it.  I was intending to agree with your position and expand on it.  For me it was hilariously because you picked the absolute perfect example to make your point.  (I obviously did a really bad job of making mine).


The reason we normalize stats for the aggragate park factor and appraise a player's value against that normalization is because that player is being compared TO THE GENERIC REPLACEMENT LEVEL PLAYER. :)  The whole reason for showing the normalized value is not to figure out what Ichiro would hit in a neutral park...or Ibanez for that's to show how much Ibanez was worth specifically ot the Mariners.  Pick a random replacement player from the MLB selection pool and Ibanez' production - though not directly depressed by Safeco will compare more favorably to the production of the other guy's numbers which WILL be depressed by the aggragate average since he's an aggragate average replacement player.
That's the point...and you missed it.


Then again - I'm biased against the entire "replacement" paradigm, anyway.
IMO - While I understand the statisticians arguments for replacement level - my position is replacment level players don't exist - and really aren't any better than 'average' player for most purposes, (other than obscuring the machinery to the layperson). 
The reality is that the deep thinkers cannot even agree on precisely where the replacement level should be set - (Is there any disagreement on where average falls?)
I mean - everybody can go to bbref and see that the average RF posted a .278/.353/.451 (.804) line in 2009.  And pretty much anyone over the age of 10 can understand what is being talked about.
What would the "average" RF hit in Safeco?  Well - the 'average' result for RFs in Safeco is a combo of righties - lefties - switchers - (and of course, includes Ichiro).  And all those results would be based on only Seattle RFs on one side - and having only faced Seattle pitchers (and its defense), on the other. 
My point is that the "average" and "replacement" hitters -- neither exist.  The closest thing to an 'amalgum' hitter would be a switch hitter -- who sometime benefits, sometimes is hurt by - the park.  Utilizing the "generically assembled" park effect - and applying it to ANY individual doesn't actually reveal anything.
Ichiro isn't worth more (or less) than an .850 hitter in Safeco (if he hits .850).  He's worth more than Willie Bloomquist - and he's worth less than say J.D. Drew (in 2009).  But, J.D. Drew and Jason Bay would almost certainly have vastly different results if moved to Safeco, (though each hit over .900 in 2009 for Boston). 
The reality is that there is zero evidence that Safeco had any impact on Ichiro's production.  Therefore, the Safeco park effect becomes irrelevant when discussing Ichiro.
Of course, if you're looking at Beltre (for example), in 2009 - hitting 70 points worse at home -- okay - the park hurts him.  But, it hurts him because he's a righty.  Is EVERY righty going to lose 70 points of OPS in Safeco?  No. 
Honestly - the park effects - applied on INDIVIDUALS - means that when talking movement, the reality gets further and further and further from the data.  The generic Boston park effect doesn't fully capture the impact on lefty or righty individual hitters.  And, if moving to Seattle, the Ms park effect will be different depending on righty-lefty (or pulls hitter - or spray hitter - or speedster, etc.)
For me - the discussion of "Player X was actually worth -- when compared to our fearless wonder Park Neutral Replacement Level MAN! (da - duh duh - DA!) ... is about as useful as saying -- well, if you compare each player to what Barry Bonds would've done here, this is what they were worth.


And the above commented adds are valuable on their own ...
Still am on my own crusade to discourage the idea that "370/410/510 is WHAT ICHIRO WOULD HIT UNDER NORMAL CIRCUMSTANCES."
That is what guys say, and it's what they believe, and it's very unhelpful. 
This mistaken paradigm rules out the possibility that, say, Casey Kotchman's performance would change if you matched him to the right circumstances.
Real-life GM's deal in these possibilities all the time.  It's what we're seeing with Kotch.


If it's true, as Sandy believes, that Ichiro would hit the same in neutral parks as he does in Safeco.
We *can* say that Ichiro was "worth" 370/410/510 last year without fear of contradiction, and that's a dream season by Rickey Henderson's standards, even.


The problem with average is that you lose the ability to place an absolute value on the average player.  Average players aren't worth zero....they're worth need a lower baseline to properly evaluate average play.  But I *agree* with you that the concept of replacement level is...well not all that useful...since it's so nebulous and hard to define precisely.  I avoid this problem by using a MATHEMATICALLY DEFINED lower benchmark...the ZERO-VALUE MARGIN...which is based on the levels of production at the team level which lead to zero additional wins in the record.  That's mathematically definite...the repalcement level is not.


And this is why I understand the reasons for Replacement Level.  I think it isn't completely unreasonable when thinking "specifically" in terms of trying to compare value of two specific players.  If you're trading Ichiro for Chase Utley - having 'some' kind of system for balancing position, age, production, defense, etc. is absolutely called for.  It's helpful - (even if imperfect).  My gripe is that the $ values computed are treated by way too many people as gospel for EVERY possible comparison. 
Catcher is my favorite nightmare in that regard.  Because of how catchers are PAID, they are given a massive adjustment for defensive value.  The value is solely and completely based on what they are paid for being catchers - and has essentially no link whatsoever to how well they actually play the position.  So, the system for examining trades gets warped into trying to support a claim that catcher X "saves" some ungodly number of runs per season. 


Especially as it pertains to catchers, the position adjustment Tom Tango invented is ludicrous and laughable IMHO.  In strict value terms, catcher is the most overrated position on the diamond.

Add comment

Filtered HTML

  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd><p><br>
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.


  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.
  • Web page addresses and e-mail addresses turn into links automatically.