What if I told you that we could highlight a smaller group of players who are more likely to hit in the NFL right now than whoever is drafted in the first round in any given draft? What if I told you I didn’t need to know the results of the combine, the rumors about what coaches think, or the potential of any landing spot to do it?
What if you could know that right now?
Using the past to predict the future is a fool’s game. There is always a Tyreek Hill, Michael Thomas or Calvin Ridley waiting to make a fool out of anyone that tries. However, if we’re going to try, we may as well know the odds.
The 2019 NFL draft is coming and I’ve broken down past wide receiver classes looking for patterns in production from 2012-2017. In other words, we have been grinding production, instead of tape, and crunching names instead of numbers.
We now have a six-year sample of what the best wide receiver production looked like in that time. And we didn’t even have to look at any graphs. (Though I did slip one in, for funsies.)
We made a model with words, and now it’s time to see the results.
I’m about to geek out a little. I’ve been holding it in for six years’ worth of draft prospects now. If you aren’t interested in “geeking” along with me, you can skip to the “What do I need to know?” section.
I think there has been value in walking through each step one by one in this series. Hopefully, you agree. But while you haven’t had to count anything, I have been keeping track of what we’ve seen in our time machine.
The first thing I want to know is how well production whittled down the potential player pool. There’s no sense in saying “90% of players who produced above Y had a top 24 season” if everyone since 2012 has produced over that threshold.
[am4show have=’g1;’ guest_error=’sub_message’ user_error=’sub_message’ ]
The production numbers we found most relevant from 2012 to 2017 cut half (or thereabouts) of the players out of the pool. In other words, it highlighted half the group and discounted everyone else.
But, can we do better? I don’t know about you but I think there are players every year drafted in rounds one-three who seem easier to “throw out” my draft board than others. How about if we combine some of these numbers?
So, I looked at the percentage of players who were above average in three categories:
- Combining all of these production statistics
- Just MS yards and yards over average (YOa) with average Dominator Rating
- Just yards, YOa and breakout age
What I found is: in the players we looked at draft rounds one-three from 2012-2017, combining all of these metrics limited the 84-player sample down to just 18 players. Just removing College Dominator and average Dominator limited it to 25 players, removing College Dominator and Average Dominator limited it to 34 players and removing everything but breakout age limited the potential player pool to 48 players.
Using all of our statistics, we can look at all 84 players drafted in our sample size and whittle it down to 18 players who stand out as the most productive, or 21.4% of all players drafted.
First-round draft capital only limits the available players to 32.1% (27/84). So combining these numbers definitely makes it easier to identify a tighter group of players.
Now we need to know the hit rates for these groups of players.
61.1% of players who were above average in all five of our statistics had at least one top-24 PPR season in the NFL. All three groupings beat the “First Round” hit rate if you wanted to be less aggressive in slimming the player pool.
By comparison, 40% of players drafted in the first round met the same mark.
And remember: our group of players is smaller (consider that a “fish in a barrel” barometer”). It was a smaller pool, with a greater chance of hitting the right player.
“Wait a minute! Did that jumped-up little number guys just say production could be a better indicator of success than draft capital? Should we get the torches?”
Well, hold that thought, but… kinda?
Ranking Players and What I’m really saying
Remember we have persistently seen that production struggles to rank players. Also note that 33% of players who were not above average in all five statistics, *but were drafted in the first round* still hit in the NFL (with at least 1 top 24 seasons).
Also, our sample size of six seasons and 84 players is small. If we’re to expand out, the numbers drop. Honestly, But even if we cut 10% of the top and sticking it onto the “none- above average in every metric” group it still beats draft capital.
When I increase the threshold for “Hit” to a top 12 season the pattern persists. In that, it identified a smaller group of players with a higher hit rate in each round and all three rounds together, then draft capital alone.
But to be clear we are dealing with so few hits at this NFL threshold that I’m not willing to suggest it is likely or trustable for evaluating 2019. We’ll stick to our goal of looking for players that can/should/will have at least one top 24 PPR season.
So being above average in all five statistics gives you a “better than coin flip” chance that players will hit in the NFL. Players who are not above average in these statistics had a one in four shot, to give some leeway to the numbers based on our smaller sample. That’s a healthy, optimism soaked 25% shot at being productive in the NFL.
That’s not very “analytical” of you Mr. Howard.
True, but it’s also a lot more reasonable based on what we know of the variance between drafts overall. Being slightly better than a coin flip would still be a big improvement than just relying on draft capital. And we draft names, not numbers, so being a little pessimistic about the potential for players to hit is never a bad idea.
What do I need to know?
In general, players drafted since 2012 with:
- An Average Dominator rating over 26%
- A positive Market Share of receiving yards compared to successful NFL players (YOa)
- An MS receiving yards over 25%
- A College Dominator Rating over 35%
- A breakout age equal to or under 20
- Drafted in the first three rounds of the NFL draft
… Have had at least one top-24 season 60% of the time between 2012 and 2017.
Our player sample that fit that definition is smaller and has a higher hit rate than those drafted in the first round.
We can put more fish in a smaller barrel, in other words.
And they said numbers weren’t exciting.
Out of Sample
So, we didn’t break down the 2018 class – mainly because we have only seen one year from them so we can obviously only be less certain who is hitting and who is missing so far. However, it also gives us an opportunity to test our numbers against a class we didn’t use to create them (that’s what “out of sample” means).
After all, what good is a model if you only ask it to tell you who’s good out of the players you have taught it about?
You: Hey model, TY Hilton is good
You: Hey model, is TY Hilton Good?
Model: Are you kidding me right now? He’s great.
So let’s look at players from the 2018 class that our model doesn’t know.
Of the ten players drafted round one-three in 2018, our threshold model identified four players more likely to have a top-24 season.
No real surprises. Calvin Ridley, (the only player to have a top-24 season already from this class, is missing. This is where we can mock and ridicule models, numbers and all things geeky that require wearing glasses. Obvi. Dante Pettis is also missing. Unsurprisingly, I was lower on these players than most in rookie drafts last year.
That brings us to the execution of our model.
You didn’t have to fade Calvin Ridley
I really shouldn’t have. He productive from age 20 onwards compared to average.
All I think/thought this threshold model means is that DJ Moore looked “more likely” to have a top-24 season within his first three years. It didn’t mean he’s due more of them, or to do them more quickly.
That would be ranking.
I don’t think we should be afraid of Ridley’s breakout age. He is clearly the rare “productive every year at an older age” outlier. He’s also still behind one of the best wide receivers of our generation and was touchdown-dependent in 2018, just sayin’. But he’s now just good. Screw the rookie profile at this point.
However, I think we are seeing the threshold model’s value persist out of sample. All four of our wide receivers identified, for example, were good in their rookie season. To borrow another one of our patterns, its noticeable all four came out in the top ten from this class in receiving yards.
Stats from Pro Football Reference.
Antonio Callaway and Marquez Valdes-Scantling are two other rookies the thresholds “missed.” But neither was drafted in the first three rounds. On top of that, Callaway also had an age-18 breakout which our production process could have identified as valuable if we looked past round three. Valdes-Scantling was also still available in round three and four if you found something else you liked (landing spot, combine, tape evaluation and so on).
I really liked Anthony Miller and Tre’Quan Smith and James Washington from 2018 based on a closer look at their profiles (the same look we just did for every class since 2012).
This meant I could try to target them using DLF ADP data after Moore, Kirk, and Sutton were gone. True, I’d target Gallup over them, but you could also lean Smith or Miller based on landing spot based on their draft capital. If I couldn’t get them later and Gallup doesn’t hit, then he’ll fall into that 40% of players in this group who don’t.
In other words. The model should never stop at round three. ADP is important. So is all other information. But in the end, two out of three players to get over 600 receiving yards (actually 700 in this case) in their rookie season in 2018, are on our very small list. I’m going to call that a win.
The Big Moment: 2019
This is not the be-all and end-all of my rookie evaluation, obviously. I’m interested in a lot of rookie wide receivers in 2019, especially if draft pick values continue to fall due to overall pessimism about the class. Something else we have also seen is that most classes, even the “bad ones”, have players worth rostering in dynasty. Some of the best players come from the worse classes and there’s a 50-60% chance that they met these thresholds in college.
We don’t have draft capital. Which will also help.
Okay, here we go. This is the list of players who met the Threshold Model from the 2019 class.
Any surprises? Would you rather argue about agility drills and 40 times? No problem, we can add those things as we go.
- Ashton Dulin was playing at Malone University, so some adjustment for his competition is fair
- Keelan Doss saw a significant drop in his production in his final year, often used to evaluate players
- Andy Isabella saw a huge jump in MS Yards in his final year. Not sure if it’s relevant but it was noticeable
- N’Keal Harry also has that positive age-18 breakout
You can add all the names you want for whatever reason you want. But I think we should remember these ones, and that they stand out in rarified company since 2012.
One last time, thank you for checking out my series breaking down past draft classes and looking for production patterns. Hopefully, you’ve found it interesting and of value for your evaluations. For now, it’s time to stop looking back and start looking forward. Who is an athlete, who is getting first-round draft capital buzz? Who do we think is going to get stuck in Baltimore?
Fortunately for us, we have a team of DLF writers already cranking out articles on exactly these and other topics of interest. I can’t wait to read them and add it to my knowledge of this class.
See you in the comment sections!
- 2022 Dynasty Fantasy Football Rookie Prospect: Tyler Badie, RB Missouri - April 5, 2022
- 2022 Dynasty Fantasy Football Rookie Prospect: Calvin Austin, WR Memphis - March 28, 2022
- 20/20: Jahan Dotson, WR Penn State - February 21, 2022