The Outliers

Peter Howard

Let there be outliers! I love outliers.

At this time of year the dynasty community is awash with rookie intrigue and excitement. As the solitary member of Team #UDFAsMatter, and someone who enjoys nothing more than hitting on an unlikely player in my rookie draft, I am a very big fan. But “outliers” is a word thrown around so much right now that it can start to lose its meaning and glory, so some context is needed.

Is an “outlier” just anyone someone else didn’t like, or what?

This is especially good timing for an exploration of the topic, considering the release of DLF’s new market share tool, a tutorial for which may be found here.

Outliers

In and of itself an outlier is just a player who is less like previous good players in the NFL in some way. Of course, good NFL players don’t look like each other very much at all. From size to speed, and draft capital to BMI, they are all, in fact, outliers.

I have a database of 361 drafted wide receiver prospects dating between 2003 and 2017, and 512 in total (UDFAs!). In that time 49 of those drafted players have had two top-24 seasons. (I’m defining “good” based on Jacob Rickrode’s work showing that players who pass the threshold therein described are much more likely to continue to produce.)

For those keeping track, that is a 15.5% overall hit rate. Good players aren’t just outliers, they are unicorns. Significantly, though, we have noted several solid groups that are more likely than not to be in that select group.

The most obvious and most consistent indicator is draft capital in the first three rounds. Interestingly, draft round is better than draft pick for wide receiver. In other words, N’Keal Harry being drafted late in round one isn’t as important as the fact he was, in fact, drafted in round one.

The next, and almost (so very almost) as predictive indicator is college production. I say that like it’s a single number, but it’s not. Some people prefer college dominator (yuk), and some prefer per-game college dominator (double yuk). You can see what I think of those stats here.

I prefer age-adjusted production. And, yes, there are also several different stats I like and others prefer. Mine is called YOA (yards over average), and I make graphs using it. These are the same kind of graphs you can make using DLF’s new MS tool, by the way.

word image

However, a good single number proxy for age-adjusted production, is Breakout Age (BOA) – the age at which a player produces 20% of more of his team’s offence in yards and touchdowns.

Even then, there is disagreement. Some like using just yards (like me in my graphs), some think a decimal place (“DJ Moore broke out at 19.2, Damn it!”) is better than a rounded age (18, 19, 20, etc.).

The long and short of it is this: good player’s, by and large, produce well in college. They are all unicorns – miracles, from a certain perspective. (Keep that in mind when you’re turning your nose up at Cole Beasley, by the way. The man’s had a fantasy career to die for compared to the majority of the very best football players to ever strap on pads. And he’s tiny. And he’s kind of slow.) Most good NFL players, though, do have college production in common.

Draft Capital Outlier

“Outlier” as I have defined it is Stefon Diggs, TY Hilton, Adam Thielen, Doug Baldwin and Steve Johnson. Are they rare? You bet your bippy they are. They are outliers, but ones I prefer chasing because they do something similar – something I can put in context, track, and search for.

But this article isn’t about them. I’ve written a lot about them if you want to dig through my back catalog here at DLF.

So, what do the other guys look like? Who are these “other” outliers, who don’t produce well, but still break out in the NFL?

Production Outliers

Up front, let me say, yes, if Terry McLaurin ever hits top-24, twice, he will be a production outlier. So who else is on that list? Below is a table of all the players who have a breakout age higher then 20, or who didn’t breakout in college at all since 2003.

word image 1

Why don’t I like searching for these players? I mean, they are about as rare as the TY Hilton’s of the world, so, like, chill, right? But I have no chill. I have the grind, and I will grind this bad argument into dust.

For a start, while I hate being the “I liked him when” guy, I don’t believe anyone pointed out Kenny Golladay significantly before me in 2017. I was and remain a big fan. How is that possible if these are the “bad” outliers? Well, because he produced in college people! Ha, I switched it up on you.

Golladay was amazingly productive at an older age, having moved up in conference level (and produced well at the lower conference, obviously, at a younger age). I liked that he had maintained his production after moving up in competition level, and I liked how well he produced in college at the age he played.

Who else?

word image 2

Was I as high on Michael Thomas as I would like to have been? No. But with second round draft capital, Drew Brees as his quarterback, and the fact that he broke out eventually, I didn’t ignore him. Phew.

I would have had very strong questions about Anquan Boldin, and Brandon Marshall as well, I imagine. But looking into them more you find that Boldin played one of his years outside of my data window (2000-2019), he didn’t play in his age 20 season, and when he came back he blew up for over 40% of he’s teams offence.

Brandon Marshall’s mental health challenges are well known, but it’s fair to say I wouldn’t have known what to do with that. I probably would have missed. But he did break out at 21 for over 40% of his team’s production. Sure, I’ll let it go. I’d have missed. I’d have liked Nate Burleson (yuk) and Andre Johnson (yay!) more.

So let’s dig into these other names, shall we?

Julian Edelman was a quarterback.

Wes Welker was a running back.

Tyreek Hill had only one year (twelve games) of college stats before his off-field problems earned him a lower conference. In that one year he was used more as a running back. Ultimately, he barely played in college.

Let’s recap. Of the six players left on the “bad” outlier list, three barely played or didn’t play wide receiver in college. Two had a great final college dominator season (over 40%). One of them is Michael Thomas.

None of them failed to breakout in college if they played wide receiver and didn’t miss time.

Thomas is the only clean example, in my opinion, of a production outlier hit. But you could include Marshall and Boldin if you want to.

That’s it. If a rookie player didn’t produce, has an older break out age, but at least put up a 40% dominator, that’s something I have to admit has a slim chance. But its slimmer then Draft Capital Outliers. And the number of potential players on that list is large. There have been 97 players drafted since 2003 who broke out after 20 years old or didn’t breakout at all – 25 of them were drafted between rounds one and three.

UDFAs hit as well, and their hit rate isn’t much lower. Plus, the cost to roster them is lower then say, Hakeem Butler. That, and the fact I can find, contextualize and search for positive production, means I prefer looking for the other (or the “good”) outliers I mentioned earlier.

The 2020 Class

There are fifteen players I know about in the 2020 class right now who somewhat fit the “production outlier” criteria.

Only five produced a college dominator over 18% (a limit I chose because I know people like Henry Ruggs.) Only three produced a decent college dominator in their final year (not Ruggs). None of them bridged 40%. Only one seems to have missed a significant amount of time (George Campbell). None of them played running back or quarterback, as far as I know.

word image 3

I recommend you check out DLF’s new MS tool I mentioned at the start of the article to explore these players further.

What are we really saying here?

Don’t try and hit on Terry McLaurin? Give up on finding Michael Thomas? Personally there is such a small potential that I’m normally comfortable giving up that variance. However, there’s no reason you have to. I just prefer to adjust them down while valuing their draft capital, tape and physical metrics accordingly.

We only have so many examples to look at. The fact that outliers in both draft and production categories exists, means there has been more, and there will be more. There’s hope.

Draft capital and production are the only two things, though, that most good NFL players have in common. Without anything we can put in context, everyone else has to be on the “bad” outlier list. All I try to do is find the kind of pool I’m fishing in when looking at a player’s production. Is it a well-stocked barrel, or an ocean with only two goldfish left?

Remember we have really only looked at three things: draft capital, breakout age and college dominator. And we have not looked that deeply. I don’t think college dominator is the ultimate, only or best stat to do it. Models, evaluations and the articles here at DLF all use more data to break them down and try to see the trees in this forest.

In the end life, being life, is very difficult to predict. Unless you think having Jordan Matthews over Odell Beckham would have worked out, there’s no defending the idea that analytics can rank prospects by itself better – right now. But it can show you who has a better and worse chance of actually having whatever it takes (no matter how they are built or what their personal situation is) to be a unicorn.

So, by all means, disagree. Buy the jersey and draft Ruggs or Hakeem Butler 1.01 overall. Have fun. Take a victory lap. Let’s play some fantasy football!

Then again look who I’m talking to; a bunch of dynasty players. You already know everything, in the end, comes down to value. Outliers, the draft, college production are just the best way of establishing odds.

Now I’m going to get back to putting this wide receiver class in the context of the ones that went before it.

Thanks for reading, see you next time.

You can find @Pahowdy’s entire database, for free, by following his pinned tweet here.

peter howard
Follow me