Tuesday, November 30, 2021

Let's Lie with Statistics

Stats Lie.

They do.  They can be manipulated, spun, crafted into whatever narrative the person presenting them wishes to tell.  They have no basis in reality insofar as reading tea leaves and making predictions go.  "Stats as Predictors" is something I absolutely abhor.  They tell a story of the past, not the future.  They tells us what we may not have seen, but not what we WILL see.    


So, screw it, why not use them to lie to myself and all of you lovely Dawgfans?  Here we go.  Let's get to lying.  

Georgia and Alabama face off in a few days time, and all this hullabaloo will be settled.  Either the Dawgs will win yet again, cement themselves as the #1 team for the remainder of 2021 and head into 2022 ready to capture their first National Title since 1980.  Or, BAMA will win (yet again), and setup a potential rematch with a freshly besmirched Dawg team, either in the Semi's or the Finals of the Playoff.  

Now, I can't tell you what the future holds.  But, I can tell you what's in the past.  I'm not interested in the 2017 past, though.  Not interested in 2018 either.  2020? Couldn't give a damn.  These two teams were not involved with any of those games. 

I'm interested only in the past of 2021.  You see, the most common argument being regurgitated by UGA fans regarding the two teams is that UGA has been completely demolishing teams and BAMA has been struggling with them.  They say the eyeball test against the best levels of competition shows UGA to be superior to Alabama.  They say that Bama simply hasn't seen a defense like ours, and that UGA doesn't care what offense it sees - they will all be chewed up and spit out. 

How in the world can folks be so confident when they spout fluff like that?

Well, the two teams played 4 Common Opponents this season.  So, let's take a look at how those games shook out.

So, these are the scores of the games, and the differential between actual and expected Margin of Victory for both UGA and BAMA.  In three of these four games, UGA's differential is vastly superior to Bama's... but Bama's ability to outdo us against the best offense we faced (TN is the #9 scoring Offense in the country, y'all) can't be ignored.  I mean, sure, Bama allowed TN to score 7 more points than they did against us.  And sure, the game was close until about 12:42 in the 4th, when Bama extended a 7 point lead to 14, and never looked back.  And sure, UGA dominated Tennessee for the entire game and allowed a second touchdown with 3 minutes to go in the game, in Knoxville (BAMA-TN was in Tuscaloosa).  Sure, all of that is true... but TN's differential IS higher than ours, so they definitely played a better game against TN than we did. 

Still, we're not done lying yet... because we haven't begun to use this data to PREDICT anything.  So let's do that shit.   

First, we have to determine some kind of trend in order to apply it.  And again, we can only use trend data from THESE four games.  We can't bring in the Wollymocker games against the likes of Charleston Southern or Mercer .. and we can't bring in Struggle Bus games like Clemson and LSU... those hardly seem like fair comparisons.  No, we will only lie with THIS data set.  So, here goes:

What THIS is showing, is the percentage of points allowed and scored by UGA and BAMA in comparison to the 12-game average points scored and allowed by these four common opponents.  You can see that, on average, UGA is allowing 25% of the opponents' average points scored, while BAMA is allowing 85% of the opponents' average points scored.  The offenses are much more similar, with UGA (national rank 6) scoring 146% of these opponents' average points allowed, while BAMA (national rank 5) is ripping off 147%. 

Now, this would lead some people to believe that these offenses are similar in their production, but anyone who watches the games knows that Alabama has a vastly superior offense to UGA's, and there's no way our guys can keep up with them.  They're Elite, and UGA just isn't, y'all.  But we're lying with statistics, so we're making it look like they are. 

So, what do these numbers tell us in a predictor fashion?  Well, for that, we'd compare these trends to the two teams' 12-game averages.  

So, what we have here, are two tiny data sets.  One, UGA's predicted points on offense, and allowed.  And BAMA's predicted points on offense and allowed.  

Oddly enough, those two data sets present some strikingly similar results.  


They are dirty, dirty bastards which have nothing to do with reality.  There is no way on God's Green Earth that the 12-0 Georgia Bulldogs who have barely played a football game into the fourth quarter this year could possibly be 21 points better than the World-Beater Crimson Tide who have, most recently, gone to 4 Overtimes with a 6-6 Auburn team led by a one-legged quarterback and a bunch of transfer portal defenders.  That is NOT possible.   

There is no way that the final score of the 2021 SEC Championship Game will in any way resemble a UGA victory, 31-10. 

Nope.  Not gonna happen.  But it's been fun lying about it. 

Go Dawgs.

No comments: