Welcome to Regression Alert, a weekly column devoted to demystifying the most talked-about force in the fantasy football universe.
Two weeks ago, I discussed what regression to the mean was, what it wasn't, when it was useful, and when it was meaningless. Last week, I looked at our first candidates for regression of the season, starting with the ever-unpredictable yards per carry.
Because I'm a big fan of stating hypotheses and then testing them, I'm going to track how my predictions fare as weeks go on. Last week, I listed two groups of running backs. Group A had rushed for 81.8 yards per game through two weeks, while group B had rushed for 51.3. I then predicted that over the ensuing four weeks, group B would outrush group A.
Through one week, things don't look so hot for my prediction. Group A averaged 79.5 yards and 4.7 yards per carry, while Group B averaged 54.5 yards and 3.4 yards per carry. Alas, Rome wasn't built in a day and we have yet to gain the full benefit of the law of large numbers; with samples this small, group A is essentially kept afloat entirely by Kareem Hunt. Absent his huge day, the group would have averaged 61 yards and 3.6 yards per carry.
Undeterred by my slow start, it's on to week four.
Yards per Target
While intuition might suggest that yards per target is functionally the receiver's version of yards per carry, things are a bit trickier than that. Often mistaken for an efficiency stat, yards per target more clearly measures usage.
In the decade from 2007 to 2016, 180 different wide receivers received at least 100 targets. Among those 180 receivers, the correlation between yards per reception and yards per target was 0.70. This means that yards per reception alone explained about half of the variation in yards per target in the sample.
We could just look at players with high or low yards per target, but that's just going to give us a list of deep threats vs. slot receivers. Instead, we can account for role to find the true overperformers and underperformers in the sample.
The best-fit equation to predict a receiver's yards per target is 2.86 + 0.367 * (yards per reception). If we then compare to a receiver's actual yards per target, the outliers begin to stand out.
For instance, Brandin Cooks currently leads the NFL in yards per target with 14.2; however, he also leads the NFL with 25.6 yards per reception, which if it stood would be the highest single-season total since Flipper Anderson in 1989. (Since this is Regression Alert, forgive the spoilers but... it won't stand.)
Given his huge yards per reception figure, Cooks would be expected to average 12.3 yards per target. He is only outperforming that value by a relatively modest 1.9 yards per target.
Compare that to, say, Sammy Watkins. On the season, Watkins averages 13.9 yards per target, which doesn't seem quite as impressive as Cooks... until you realize that Watkins averages a much more modest 14.9 yards per reception. The result is that Watkins is outperforming his expected yards per target by a ludicrous 5.5 yards!
Through three weeks, there are 31 wide receivers with at least 150 receiving yards. Interestingly, of those 31 receivers, a full 28 are actually outperforming their expected yards per target.
(This isn't that mysterious; the players who are underperforming their estimated yards per target typically don't have enough yards to reach the 150-yard threshold. Anytime we look at the top performers, players who are overperforming will be overrepresented.)
If we sort those receivers based on how much they're overperforming, though, here are the groups that emerge.
- Group A: Sammy Watkins, Jordan Matthews, Adam Thielen, Michael Crabtree, Tyreek Hill, Stefon Diggs, Jermaine Kearse, T.Y. Hilton, Antonio Brown, Brandin Cooks
- Group B: DeVante Parker, A.J. Green, Mike Evans, Chris Hogan, Larry Fitzgerald, Rishard Matthews, Keenan Allen, DeAndre Hopkins, Emmanuel Sanders, Alshon Jeffery
As with last week, the groups are not cherrypicked. That is the top ten and bottom ten receivers on the list based on yards per target over expectation.
Through three weeks, Group A averages 6.8 targets, 5 receptions, and 79.8 yards per game. Group B averages 9.6 targets, 5.8 receptions, and 68.6 yards per game. Despite Group A averaging 16% more receiving yards per game to this point, I predict that the fluctuations from expected yards per target even out going forward and Group B's greater volume of targets wins out in the end. Over the next four weeks, I anticipate Group B will average more receiving yards per game than Group A.
As always, I'll be back again next week to track my progress and look at another statistic that's ripe for regression.