Wednesday, January 13, 2016

Joe Thomas and More Analytics Corrections

In this Pat McNanoman article on the ramifications of Joe Thomas leaving, Pat illustrates more misconceptions about analytics. 

Pat cites Joe Banner, who said that analytics will always favor draft picks over aging veterans.  Joe misspoke, and perhaps didn't fully grasp analytics himself.

Let's substitute the word analysis for analytics.  Is it becoming clearer now?  Analysis, correctly done, separates bias/prejudice/emotion/subjectivity from the equation.  

As a general rule, analytics will favor draft picks over aging veterans.  So does Bill Belichick.

Let's stick with Bill here.  Any Patriot (except Brady) who crowds 30 better have his bags packed.  Well...except offensive linemen, too.  Because offensive linemen, who don't rely on explosive speed and quickness, and aren't subjected to high-speed impacts as much, tend to remain viable longer.

And then, Bill will often bring in 29 or 30 year old free agents on inexpensive, short-term contracts to fill gaps.

Bill is analytical.  This is how it works.  It's whatever makes the most sense.

Now, Joe Thomas is an offensive lineman.  He has never been seriously injured.  He hasn't been run into the ground in training by stupid coaches.  His remaining contract is a bargain for what he provides, and has three years left.

A true analysis considers all these factors, then integrates them with situational and contextual factors:

1: A probable rookie quarterback starting.  Remember Tim Couch?  
2: The fact that quarterbacks usually develop slowly.
3: Other free agents, like Schwartze and Mack.  As Pat noted in his article, these are Joe's friends, and he is a big part of what gives them any pause in leaving for a couple extra shekels.
4: The potential of the team to contend before the end of his contract.

I know, I just heard Black Cloud spit his coffee out, but he's not analytical.  He sees everything through excrement-colored glasses.  He lumps Joe Haden in with Gilbert, Gordon in with Manziel, and Crowell in with West.  The new coach will be lumped in with the worst of the previous coaches, as well.  Black Cloud will have a final verdict on the 2016 draft after the first pre-season game, if not during training camp.  (Except the rookie quarterback, because they don't count).

In his article, Pat repeated that Big Joe was "almost traded" last year.  What really happened was Farmer got an offer and said no.  He might have been laughing when he said it.  Pat isn't as emotionally detached or objective as he should be.

Pat said that if analytics valued Joe as worth a second round pick (excuse me....no wait....ok I...ok I think I can stop laughing now) that they'd trade him for one.

It's true that DePodesta (with the help of the personnel people) will assign values to each player, and consider trading them for those values if offers are made.  It's true that younger is better, until you are on the brink of a Super Bowl.  (Younger and cheaper btw).

But Pat doesn't seem aware of just how deep these analyses go.  You can't do them on calculators, Pat!

Joe is worth two first round picks, for starters.  That's for the Browns.  That is, the Browns would need that much to have any chance to redeem him for equal value.  What he's worth to the other team is irrelevant.  The Browns won't actively seek a trade unless Joe asks for it, or the new coach wants it.

Speaking of "almost traded", real analysts analyze word-choices.  I know from how Pat writes that he sometimes makes assumptions.

This type of analysis can be applied to recorded interviews.  It includes what the eyes do as they answer, blinks, and other tells.  Hesitations, struggling with unfamiliar words they've rehearsed.

This is interrogation science.  It's how Israeli security agents become human lie detectors, and the best detectives nail criminals...that's analysis too.

Probably, DePodesta will seek to quantify Joe Thomas's emotional value to the offensive line and quarterback in particular, and the team in general.

That won't be as hard as it sounds.  If his model includes all players, Joe Thomas would be a 100.  Joe Haden in the nineties...etc.

Reports of Joe Thomas's demise are greatly exaggerated...and not analytical.

No comments: