Algorithm Aversion Let Data Take the Wheel on Henry Ruggs

By Kyle Dvorchak
Kyle Dvorchak

Breakout Age, the age at which a receiver reached a significant share of his team's offensive production in college, is one of the most useful metrics the fantasy community and NFL teams have for predicting wide receiver success. Players who account for a large portion of their offense at a young age tend to age like a fine Cabernet Sauvignon, which the internet says is a wine that ages well. This concept was first popularized by Jon Moore, Shawn Siegele, and others has entered the mainstream of football analysis.

Ignoring this metric has been a path to ruin but many are choosing to do just that this offseason with Henry Ruggs. Even people who respect the value of Breakout Age are choosing to see Ruggs' lack of a breakout and make a manual override in his specific case. He played with Jerry Jeudy who is likely to be the first or second receiver off the board come draft day. Ruggs also played alongside Devonta Smith, a possible day two pick in the 2021 NFL Draft. He should be the exception to the breakout rule...right?

Algorithm Aversion

A University of Pennsylvania study tasked subjects to predict the success of MBA students using data points including undergraduate scores, measures of interview quality, and work experience. Subjects were also given a model that predicted MBA success and the model had previously been shown to be more accurate than the subjects. Subjects largely chose to tie their earnings for participating to their own performance instead of the model.

Subjects were especially likely to side with themselves when they watched the model make an error, even if the model was more accurate on the whole. This phenomenon, coined algorithm aversion, cost the subjects money and continues to rack up costs in the real world.

The Next Victim - You

Breakout Age isn't a model itself but it is a dataset used for predictions that will beat scouts, fantasy ADP, and most metrics outside of actual NFL draft position, which begets future volume and Breakout Age already plays a factor in deciding. Not having a breakout season is damning. According to Peter Howard's MS Database, there have been ten NFL receiver prospects since 2003 to post a top-24 season without a college breakout (using 20% of their teams' receiving yards as the threshold). The other 145 failed.

To like Ruggs, you have to see yourself as the model-beater. The Penn study shows (and then the results were replicated in similar studies) that model-beaters aren't real. The way to beat a model is the build a better model, not to cherry-pick when the model might be wrong.

But What About…?

There are a number of things to like on Ruggs' profile coming out of Alabama. He posted 17.9 yards per reception in college backed that up with a stunning 4.27 Forty-Yard Dash at the combine. He's also projected to be taken in the first round of the draft. That makes him one of the fastest players to ever participate in the combine and guarantees the team that selects him will be giving him ample opportunities. The outlook still isn't pretty.

Three receivers were taken in the first round with the latest being Cordarelle Patterson. None of the three ever produced a WR2 season.

Three receivers ran faster in the 4.3s or faster and there is one hit in the group: Tyreek Hill

Using Ruggs' two biggest selling points generate one similar player with NFL success.

Why People Override the Model

There's no concrete explanation for why people are so inclined to side with themselves when data proves itself a better judge. One theory is that people believe they are able to sing things the model can't. The model doesn't learn after it's created so it won't understand that Ruggs is an easy exception to the rule. But there have been players with Ruggs speed or draft capital to fail. Craig David busted as a first-rounder for the Chargers while Marquise Goodwin recorded an identical 40 time and also failed.

Even the NFL success stories with stout competition in college found a way. Odell Beckham Jr. and Jarvis Landry both broke out. Last year, Ole Miss teammates A.J. Brown and D.K Metcalf entered the league with breakouts on their resumes. Ruggs wasn't able to do what these players did, and make a name for themselves in college while playing with other NFL-caliber players. Here's a secret I'll let you in on: everyone he'll be playing alongside in the NFL will be NFL talent. It's the NFL.

Taking it to the Draft

The Ruggs equation is simple. The biggest characteristics going in his favor haven't proven to be a consistent method in beating the non-breakout. He is currently going in the back half of the first round in rookie drafts. Let someone else take Ruggs' and his lack of successful comps.

Ruggs isn't the only player in this draft class who lacks a breakout. K.J. Hill of Ohio State and Van Jefferson of Florida both failed to break out. They'll cost less in rookie drafts but will also get less support inherent in their draft capital.

This isn't an attack on a single player either. Ruggs can be fast and provide dynamic value to an NFL team without ever being fantasy relevant. Cordarelle Patterson has been a dynamic threat on the NFL field without ascending to fantasy stardom. There will also be more players like Ruggs in the future. They too will be worth fading in fantasy leagues. Some of them will even hit but they have not succeeded at a rate that makes them worth passing up other receivers for. Drafting players without breakouts is a blind pull at the slot machine. Even if you win this time, the house will take your money in the long run.