Welcome to Regression Alert, your weekly guide to using regression to predict the future with uncanny accuracy.
Returning readers, you know how this works by now, but for new readers here's the deal. Every week I take a look at a specific statistic that is prone to regression and identify high and low outliers in that statistic, and then I wave my hands in the air and shout “regression!”
But since predictions aren't any fun without someone holding your feet to the fire afterward, I don't stop there. I lump all of the high outliers into Group A. I lump all of the low outliers into Group B. I verify that Group A is outperforming Group B. And then I predict that Group B will outperform Group A over the next four weeks.
I don't get to pick and choose my groups, beyond being free to pick and choose what statistics are especially prone to regression. If I'm tracking yards per target, and Antonio Brown is one of the high outliers in yards per target, then Antonio Brown goes into Group A and may the fantasy gods show mercy on my predictions.
And then, groups chosen and predictions made, I track my progress. That's this.
In Week 2, I outlined what regression was, what it wasn't, and how it worked. No prediction was made.
In Week 3, I listed running backs with exceptionally high and low yards per carry averages and predicted that the low-ypc cohort would outperform the high-ypc cohort over the next four weeks.
In Week 4, I looked at receivers who were overperforming and underperforming in yards per target and predicted that the underperformers would outperform the overperformers over the next four weeks.
In Week 5, I compared the predictive accuracy of in-season results to the predictive accuracy of preseason ADP. Outside of a general prediction that players would tend to regress in the direction of their preseason ADP, no specific prediction was made.
In Week 6, I looked at quarterbacks who were throwing too many or too few touchdowns given the amount of passing yards they were accumulating, then predicted that the underperformers would score more fantasy points than the overperformers going forward.
In Week 7, I looked at receivers who were catching too many or too few touchdowns based on their yardage total, then predicted that the underperformers would score more fantasy points than the overperformers going forward.
In Week 8, I revisited yards per carry, again predicting that the high-carry, low-ypc group would outrush the low-carry, high-ypc group going forward.
|Statistic for regression||Performance before prediction||Performance since prediction||Weeks remaining|
|yards per carry||Group A had 60% more rushing yards per game||Group B has 16% more rushing yards per game||None (Win!)|
|yards per target||Group A had 16% more receiving yards per game||Group B has 11% more receiving yards per game||None (Win!)|
|passing yards per touchdown||Group A had 13% more fantasy points per game||Group A has 16% more fantasy points per game||1|
|receiving yards per touchdown||Group A had 28% more fantasy points per game||Group A has 61% more fantasy points per game||2|
|yards per carry||Group A had 25% more fantasy points per game||Group A has 11% more fantasy points per game||3|
Alright, let's talk about the elephant in the room that is the two yard per touchdown predictions. Both are faring poorly and running out of time to turn things around.
The animating philosophy of this column is accountability, and accountability demands concrete predictions with concrete end-dates. To keep things fresh and moving, I track each prediction for four weeks. But in the grand scheme of things, four weeks is still a pretty small sample.
As I like to say, regression operates over longer timescales. Sometimes outliers can remain outliers for a couple weeks or even a month or two before reality catches up.
Which brings me to Deshaun Watson. After week 5, when I predicted he was likely to regress, he was throwing a touchdown on 8.3% of all pass attempts. That may or may not sound high to you, but consider: when Peyton Manning set the single-season passing touchdown record in 2013, he threw a touchdown on... yup, 8.3% of all pass attempts.
That kind of pace isn't sustainable. For his career, Peyton Manning had a touchdown rate of 5.5%. If you restrict yourself to his MVP-caliber peak, (from 2003 to 2014), he threw a touchdown on 6.3% of passes. If he'd maintained an 8.3% touchdown rate during that stretch, he would have jumped from 539 passing touchdowns, (the highest total in NFL history)... all the way to 664 touchdowns, a 125-touchdown increase.
So Watson was guaranteed to regress in the long run. But funny stuff can happen in the short run, and in his last two games, Watson managed to increase his touchdown rate up to 11.9%, bringing his season average to 9.3%.
Here's some perspective for you. Since the NFL/AFL merger in 1970, there have been three seasons where a player threw 200 passes and had a touchdown rate over 9%. Ken Stabler did it in 1976. Peyton Manning did it in 2004. And now Deshaun Watson did it in 2017.
3 players since the merger have had 9+% TD rate on 200+ attempts. 1976 Stabler, 2004 Manning, 2017 Watson.— Adam Harstad (@AdamHarstad) October 30, 2017
Zero chance he keeps this up.
Watson is joy itself and I would never speak ill of him, but someone needs to mention he currently has the 4th-best era-adjusted TD% ever. pic.twitter.com/EYfJsMTDzR— Adam Harstad (@AdamHarstad) October 30, 2017
Watson isn't by any means the only thing that's gone wrong for these predictions in the early going, (hello Carson Wentz and Juju Smith-Schuster), but in samples this small one outlier like Watson makes a big difference.
How much much of a difference? Removing Watson from the quarterback prediction doesn't salvage it, but it drops Group A's lead down to 11%. Removing Watson's receivers from the receiver prediction drops the lead from 60% down to 26%.
I feel confident that if he played out the season, there's no way Deshaun Watson could maintain a touchdown rate over 9% going forward. That's a feat that Drew Brees, Aaron Rodgers, Dan Marino, Tom Brady, Steve Young, Brett Favre, and nearly every other great quarterback in modern history never managed to accomplish.
Unfortunately for football fans anyway, the Fantasy Football Gods are fickle, and a torn ACL in practice robbed us all of the opportunity to see him try.
Anyway, on to the prediction.
Twice Bitten, Thrice... Still Not Shy
So, my yards per carry and yards per target predictions have paid pretty well so far this year, but my yard to touchdown ratio predictions have blown up in my face. Does that mean I'm swearing off of yard to touchdown ratio predictions?
Of course not. A Foolish consistency, as Mr. Ralph Waldo Emerson informs us, is the hobgoblin of little minds. My mind must be small indeed, as I tend to prefer Shakespeare, here: once more unto the breach, dear friends, once more; or close the wall up with our statistical dead.
Last week, we looked at 2nd- or 3rd-tier running back options. This week, let's deal with the stars at the position.
There are twelve running backs in the league right now who have topped 600 yards from scrimmage. Those twelve running backs have a yard-to-touchdown ratio ranging from 254 (LeSean McCoy) down to 95 (Melvin Gordon). In the long run, you'd expect everyone to settle more into the middle of that range.
Melvin Gordon, Leonard Fournette, Ezekiel Elliott, Todd Gurley, and Chris Thompson all average fewer than 140 yards for every touchdown scored. Collectively, they average 113.9 yards and 1.0 touchdowns per game, a ratio of 110:1, and score 17.6 fantasy points per game.
LeSean McCoy, Le'Veon Bell, Jordan Howard, Kareem Hunt, and Mark Ingram, on the other hand, all average more than 140 yards for every touchdown scored. Collectively, they average 110.6 yards and 0.6 touchdowns per game, a ratio of 191:1, and score 14.5 fantasy points per game.
This isn't a true regression scenario where we'd expect Group B to outperform Group A going forward; after all, Group A averages more yards per game. But Group A averages 21% more fantasy points per game to this point, and we'd expect Group B to close that gap substantially going forward.
At least, that's what we'd expect if the entire concept of yard to touchdown ratios is a meaningful one. Early-season returns to the contrary, I firmly believe that it is, though perhaps it has just been performing above its true predictive power in recent seasons and is therefore itself regressing. Tune in the rest of the year as we find out!
More articles from Adam HarstadSee all
More articles on: AnalysisSee all
Normalized SOS: Conference Championships - Lee
Beating the Odds: Week 19 - Zamichieli
Beating the Odds: Week 18 - Zamichieli