Like many RotoViz writers, I take a heavily numbers-based approach to fantasy football. In dynasty, whenever I get new ADP, is to compare current prices to expected prices, which I estimate using projections for next year and age. Large discrepancies between those two give me a set of players to buy or sell.
Last month, I followed my usual process and discovered this strange result:
Player B is projected1 for only 15 fewer points than player A next year (and partly compensates for that by being half a year younger), yet player A was priced at 2.01, while player B was priced at 5.07. That basic fact tells me that either (1) player A is overvalued, (2) player B is undervalued, or (3) there is some difference between these players other than production and age that is really important to other drafters.
Those who have read the title of this article will guess that player B is Tyreek Hill. Player A might surprise you.
Player A is Allen Robinson. While I do hear some arguing nowadays that Robinson is overvalued, my simple analysis thinks his ADP is entirely fair based on his age. The mistake, it believes, is the price of Tyreek Hill.
I should mention that Hill’s price did move up in the last month. He is now being taken at the fourth and fifth round turn. But whatever momentum he gained around the time Jeremy Maclin was cut has now disappeared.
The reason for that seems clear: the fantasy community remains deeply skeptical of Hill. In a survey of 1500 recent tweets about Hill from football accounts, there was a roughly 60/40 split of positive to negative comments. However, the negative comments were nearly all from fantasy football accounts, while the positive comments were mainly from “real football” accounts. I believe this negative sentiment is the main reason that drafters are not taking Hill much earlier in drafts.
What is especially odd is that this skepticism is much stronger in dynasty than in redraft. According to the RotoViz Best Ball App, Hill is the WR22. This redraft ADP is actually lower (more expensive) than his dynasty ADP, which seems extremely odd for a player that is 23 years old. In fact, a linear model predicting dynasty ADP from redraft ADP and age alone identifies Hill as the most undervalued receiver on the board by a good margin. Thus, whether we estimate Hill’s near-term prospects using projections or redraft ADP, his dynasty ADP just does not make sense.
I don’t mean to say that drafters are wrong to be skeptical about Hill. He is a unique player — certainly no clone of Allen Robinson — with a wide range of possible outcomes. But many of the analysts I have read are convinced that they know with certainty that Hill will fail (or, in a few cases, that he will succeed). As I stated in a recent article, my view is that neither fantasy managers nor stock pickers are likely to be successful thinking in absolutes like this.
Hill’s outcome is an unknown, a mystery. To find the smart play, in my view, we need to consider a range of possible outcomes, estimate their likelihoods, and determine whether Hill is priced closer to his downside (making him a buy) or his upside (a sell).
Here are the most useful comparisons that I can find for Tyreek Hill as we consider who he might become in his NFL career.
Is He Cordarrelle Patterson?
Hill is often grouped together with “gadget players” like Percy Harvin, Tavon Austin, and Cordarrelle Patterson. Of those three, Patterson is the one whose early career looks, to me, most similar to Hill’s.
When trying to compare receivers, I find measures of how their targets are distributed, like aDOT, to be the most useful since they are the most sticky year-to-year. Hill, Harvin, Austin, and Patterson appear similar in terms of how many targets they received at or behind the line of scrimmage: these made up roughly 30 percent or more of each of their targets early in their careers. In terms of deep targets (20+ yards), though, there are more substantial differences. Harvin and Austin saw 10 percent or less, while Hill and Patterson saw 14 percent or more. (Austin’s deep targets increased in his third year, but still never reached 14 percent.)
Overall, the target distributions of Hill and Patterson look the most similar:
|Player||Year||<= 0 yds||1-9 yds||10-19 yds||>= 20 yds|
Differences between Hill and Patterson appear, however, when we look at what the two did with those deep targets. Patterson caught only 25 percent of his targets, while Hill caught 42 percent (about league average). As a result of his poor performance on deep throws, Patterson saw fewer of them each year after, quickly dropping to 10 percent or fewer deep targets like Harvin and Austin.
This is not the only evidence that Hill is simply a better receiver than Patterson. Hill scored well in Matt Harmon’s Reception Perception project. He also had elite numbers in Pro Football Focus’s yards per route run metric. The declaration from the Chiefs, that he will be their WR1, also strongly indicates that the Chiefs believe he is a good receiver.
Altogether, the evidence suggests to me that Hill is not likely to fail in the same manner as Patterson.
Finally, it’s worth noting the difference in price between Patterson and Hill. After the season shown in the table above (2013), Patterson’s was a mid-second round pick. Hill is still available at the end of the fourth round, so he is both more likely to be successful and substantially cheaper than Patterson.
Is He Jamison Crowder… or Brandin Cooks?
While targets behind the line of scrimmage are often associated with gadget players, they are certainly not the only receivers to see them. For example, this offseason, we hear rumors of the Broncos offense planning to throw more bubble screens to Demaryius Thomas, a play on which he was especially successful early in his career.
More generally, we should not be surprised to see targets behind the line of scrimmage for young players, as some of Broncos fans noted after Thomas’s rookie season: the bubble screen is an easy play to learn, so it’s an easy way to get targets to a play maker (like Thomas or Hill) while they’re still adapting to the complexity of NFL offenses.
Here are two other receivers who saw a large percentage of such targets in their rookie season but are not considered gadget players:
|Player||Year||<= 0 yds||1-9 yds||10-19 yds||>= 20 yds|
Crowder saw nearly as many targets behind the line of scrimmage as Hill in his rookie season. In season two, however, this number dropped to just eight percent. The missing targets mainly converted into short (1-9 yard) targets, but Crowder also saw an increase in intermediate (10-19 yards) and deep targets.
Cooks saw a similar transition, with his 25 percent of targets behind the line of scrimmage reducing to 19 percent in season two and then nine percent in season three. Unlike Crowder, far more of the missing targets converted to deep targets. While Cooks saw only nine percent deep targets his rookie season, he saw around 22 percent in both of the subsequent years.
The same path taken by Crowder and Cooks — seeing many targets behind the line of scrimmage as a rookie, while still learning to work in an NFL offense, and then converting to a more normal target distribution in years two and three — seems available to Hill.
Of the two comparisons, Cooks seems most similar to Hill. Both are elite athletes, and both are deep threats. Crowder is a sub-par NFL athlete with 4.56 40 yard dash and a 19th percentile SPARQ score. His deep targets have also never reached 14 percent. Hill and Cooks also have a big play ability that allows virtually any reception to turn into a touchdown.
Hill is currently being priced the same as Crowder. Cooks, the more similar player, in my opinion, is a second round pick. When Cooks was entering his second year, as Hill is now, he was also priced in the mid-second round.
Is He Jeremy Maclin… Plus Rushing?
As mentioned above, the Chiefs have declared Hill their WR1, replacing Jeremy Maclin. Hence, another natural comparison is to Jeremy Maclin or at least to an average WR1 in Reid’s offense.
From 2007 to 2015,2 Reid’s WR1 averaged 115 targets. That would be nearly a 40 percent increase for Hill. Reid’s average WR1 had 68 receptions for 944 yards and 5.9 touchdowns, fairly similar to the projected yardage for Hill I gave at the beginning of the article (but more on that below). It’s worth noting that, over this range of time, Reid’s WR1s were not just Jeremy Maclin and DeSean Jackson but also Dwayne Bowe and Kevin Curtis.
Of course, this projection includes only receiving yards, while Hill also gets carries. If I project him for 180 yards rushing and 1.3 touchdowns, then his overall projection is over 1100 yards from scrimmage and over seven TDs. In other words, if Hill is an average Reid WR1 but adds in some rushing, his projected fantasy point total would be the same as Allen Robinson’s.
Rather than comparing Hill to an average Reid WR1, it might make more sense to compare him directly to Maclin, the player he is replacing. Here is how the two compare in their target distribution:
|Player||Year||<= 0 yds||1-9 yds||10-19 yds||>= 20 yds|
Despite being seen as fairly different players, Maclin in 2015 also saw a substantial number of targets behind the line of scrimmage. He also saw a rate of deep targets almost identical to Hill’s. Thus, for Hill to become Maclin, in terms of target distribution, he does not need to eliminate targets behind the line of scrimmage. Instead, he needs to convert about half of these targets into short and intermediate targets. As we have seen from Crowder and Cooks, this would not be an unusual transition for a receiver after their first year in the NFL.
The numbers above are from Maclin’s 2015 season, where he turned 124 targets into 87 receptions, 1088 yards, and eight touchdowns, for a total of 244 fantasy points. It is worth noting that Maclin’s ADP, after that season, was around 3.01, and that Maclin was over 27 years old at that time, whereas Hill is still 23. Hence, if Hill were to put together a season like Maclin’s in 2015, an ADP in the mid-second round would not be unreasonable for Hill next year. Oh, and those numbers do not account for Hill’s rushing yardage.
We’ve looked at a range of possible outcomes for Hill this season. On the one end, he could become the next Patterson. If that happens, his ADP will undoubtedly drop. He could also become the next Crowder. In that case, his ADP is about right where it is now. On the other end, he could become the next Cooks, Maclin, or just an average Reid WR1. In any of those cases, an ADP in the second round next year would be possible.
If this is a good representation of the set of possible outcomes, then the question is whether we think the odds that he is Patterson are higher than the odds that he is a good receiver like Cooks, Maclin, or an average Reid WR1. If you do, then fade Hill this year. In my case, Hill’s success on deep targets, in Reception Perception, and in yards per route run make me think is more likely to be one of the upside options. That’s why I’m buying Tyreek Hill.