Welcome to Regression Alert, your weekly guide to using regression to predict the future with uncanny accuracy.
For those who are new to the feature, here's the deal: every week, I dive into the topic of regression to the mean. Sometimes I'll explain what it really is, why you hear so much about it, and how you can harness its power for yourself. Sometimes I'll give some practical examples of regression at work.
In weeks where I'm giving practical examples, I will select a metric to focus on. I'll rank all players in the league according to that metric, and separate the top players into Group A and the bottom players into Group B. I will verify that the players in Group A have outscored the players in Group B to that point in the season. And then I will predict that, by the magic of regression, Group B will outscore Group A going forward.
Crucially, I don't get to pick my samples (other than choosing which metric to focus on). If the metric I'm focusing on is touchdown rate, and Christian McCaffrey is one of the high outliers in touchdown rate, then Christian McCaffrey goes into Group A, and may the fantasy gods show mercy on my predictions.
Most importantly, because predictions mean nothing without accountability, I track the results of my predictions over the course of the season and highlight when they prove correct and also when they prove incorrect. Here's a list of my predictions from 2019 and their final results, here's the list from 2018, and here's the list from 2017.
THE SCORECARD
In Week 2, I opened with a primer on what regression to the mean was, how it worked, and how we would use it to our advantage. No specific prediction was made.
In Week 3, I dove into the reasons why yards per carry is almost entirely noise, shared some research to that effect, and predicted that the sample of backs with lots of carries but a poor per-carry average would outrush the sample with fewer carries but more yards per carry.
In Week 4, I talked about how the ability to convert yards into touchdowns was most certainly a skill, but it was a skill that operated within a fairly narrow and clearly-defined range, and any values outside of that range were probably just random noise and therefore due to regress. I predicted that high-yardage, low-touchdown receivers would outscore low-yardage, high-touchdown receivers going forward.
In Week 5, I talked about how historical patterns suggested we had just reached the informational tipping point, the time when performance to this point in the season carried as much predictive power as ADP. In general, I predicted that players whose early performance differed substantially from their ADP would tend to move toward a point between their early performance and their draft position, but no specific prediction was made.
In Week 6, I talked about simple ways to tell whether a statistic was especially likely to regress or not. No specific prediction was made.
| Statistic for regression | Performance before prediction | Performance since prediction | Weeks remaining |
|---|---|---|---|
| Yards per Carry | Group A had 3% more rushing yards per game | Group B has 36% more rushing yards per game | Success! |
| Yard to Touchdown Ratio | Group A averaged 2% more fantasy points per game | Group B averages 40% more fantasy points per game | 1 |
When I made my yards per carry regression prediction, I said that Group B would need to outrush Group A by at least 15% for me to count it as a win. I assumed that Group B would maintain their volume advantage, and they did— Group B backs averaged nearly 19% more carries per game than Group A backs. If both groups had averaged the same yards per carry going forward, I would have met the threshold necessary to call my prediction a success. As a result, I believe the underlying process behind the prediction was sound.
But both groups did not average the same ypc going forward. At the time of the prediction, Group A backs were outgaining Group B backs by more than two yards per carry (5.84 vs. 3.78). Since then, Group B has outgained Group A by more than half a yard per carry (4.45 vs. 3.86). This is not the result of a single outlier performance. Only 15% of Group A backs averaged more than 5 yards per carry over the last two weeks (and both of the backs in question, Austin Ekeler and Nick Chubb, played just 1.5 games during this span). 18% of Group B backs managed the feat. Only 31% of Group A backs topped 4.0 yards per carry, compared to 55% of Group B backs. Only 62% of Group A backs even topped 3.5 yards per carry, compared to 82% of Group B backs.
Group B averaged more yards per carry across the board than Group A. They had more high-ypc performances and fewer low-ypc performances. As a result, they didn't just beat the prediction, they smashed it. But this fact isn't especially meaningful; because yards per carry is so random, Group B is just as likely to average fewer yards per carry over the next four weeks. Yards per carry is not a meaningful indicator. Always bet on the underlying volume, instead.
There's one more week left on our yards per touchdown prediction, but Group B has staked a monster edge and will very likely be closing out our second win. Our Group B receivers— the high-yardage guys who struggled to convert yards to touchdowns— have not only been averaging more yards than our Group A receivers, but they've continued converting those yards into touchdowns at a higher rate, too. To this point, Group B averages a touchdown for every 180 yards (vs. 213 yards for Group A) and 0.36 touchdowns per game (vs. 0.22 for Group A).
Kickers are People, Too
In addition to Regression Alert, I write a column for Footballguys in which I identify the best kickers who might be available on waivers in any given week. In a bit of a "worlds collide" moment, my model has been having issues with Jason Myers of the Seattle Seahawks. Including this week (but ignoring his bye), I've recommended Myers as a streaming option in five out of six possible weeks, which is a problem because Jason Myers ranks last among all kickers with at least four starts in fantasy points (though he does rank 28th in points per game).
The issue is that my model likes kickers on teams that score a lot of points, and it has identified the Seattle Seahawks as one such team. So far, so good; Seattle leads the NFL in scoring offense.
But to this point, Seattle's offense has been too good. The team has scored 23 touchdowns against just 2 field goal attempts. This ratio is, to put it bluntly, insane. Seattle has scored a touchdown on 89% of its trips to the red-zone. From 2015-2019 the team converted for a touchdown on just 58% of its red-zone trips. Over the last decade, only one team has converted at a rate better than 75% in any season (the 2019 Tennessee Titans, who scored a touchdown on 77% of their red-zone trips), and only seven other teams even topped 70%.
Because obviously-unsustainable rates are obviously unsustainable, I continue to recommend Myers despite the terrible results to date. (When he's a top recommendation, Myers averages 4.6 points; all other top recommendations average 7.9 points per game.) But it occurred to me that I've explicitly predicted regression for quarterbacks in this column. I've predicted regression for wide receivers and tight ends. I've predicted regression for entire offenses and defenses, and I've predicted regression for incoming draft classes. But I've never predicted regression for kickers before, and that's a shame, because kickers are people, too, and they shouldn't be spared the implacable gaze of our regression overlords.
So here's what we'll do. I've added together all field goal attempts and touchdowns to calculate an offense's "scoring opportunities". Sixteen teams average 5 or more scoring opportunities per game, sixteen teams average fewer than five per game. Of the 16 "low-opportunity" teams, the five teams with the most points from their kickers are the Bears, Texans, Giants, Broncos, and Bengals. These five teams average two touchdowns per game, but 2.4 field goal attempts per game. They're our Group A.
Of the 16 "high-opportunity" teams, the five teams with the fewest points from their kickers are the Seahawks, Cardinals, Cowboys, Chiefs, and Steelers. These five teams average 3.6 touchdowns per game, but just 1.4 field goal attempts. They're our Group B.
Group A is scoring "too many" field goals, while Group B is scoring "too few", and because field goals are three times as valuable as extra points for fantasy, kickers in Group A average 8.2 points per game in fantasy vs. just 6.8 for kickers in Group B. Despite a 20% edge in fantasy points per game through six weeks, however, I predict that through the magic of regression, kickers from Group B will outscore kickers from Group A over the next four weeks.
