In the previous post, we saw that, despite the casual fan’s obsession with it, athletic scores just do not predict how the production of an NFL wide receiver.
Using PFF data, we can look at how a WR produces in college and determine how that translates to NFL WR production. The best metric to use for NFL WR production is YPRR (yards per route run). This metric reduces the dependence of a WR on the offensive system – using yards, TDs, or other volume metrics can be heavily influenced by offensive system (teams who throw the ball more will have WRs who get more yards, TDs, etc).
Now, a WR’s production will always be tied to the QB (a receiver can’t throw the ball to himself). So, it would be good to find a metric that is independent of the QB – however, what I have found is that when trying to normalize WR production by QB statistics that the overall correlations don’t change much. Essentially, good WRs produce on a per route basis almost regardless of QB (I said, almost).
So, I have found using YPRR as the best metric for NFL WR production. Now, how does NFL YPRR correlate to College WR production?
College WR Production: Man vs Zone
In reviewing the data, there was something interesting I found in correlating NFL YPRR to college statistics using PFF stats. PFF does a good job of breaking out several statistics by the defensive coverage: man vs zone. You would think how a WR produces against man coverage is more predictive of NFL WR production. We have all seen the highlights of WRs beating their man one-on-one for the big plays – so that must be the top correlation.
But, it’s not – how a college WR produces against zone coverages actually matters more. And, reviewing how NFL defenses play, it does make sense.
NFL Defense Man Coverage % for each Down (from PFF)
NFL defenses run more zone coverage than man – especially on early downs (1st and 2nd downs). Approximately 75% of coverages on 1st and 2nd down are zone coverage. Now, 3rd and 4th down, those percentages get closer to 50% as defensive coordinators will call more blitzes on those downs. But, finding WRs that can beat zone coverage is very important for NFL teams because that is the defense you see the most.
If an NFL team can find a good WR against zone (and throw on early downs), then they will find more offensive success. And, you can see this in the correlation of NFL YPRR to College YPRR and Targeted QB rating against zone.
NFL YPRR vs College YPRR against Zone (all NFL Wide Receivers drafted since 2018 with >200 routes ran)
Now this correlation only has an R-squared value of 0.044 (much better than RAS but not overly correlated), however, you can see some interesting groupings within this correlation. For college WRs who had >3.75 YPRR against zone (19 total WRs since 2018), only 6 WRs have an NFL YPRR <1.25 (below average) and 13 WRs have NFL YPRR >1.75 (very good to elite). So, just looking at this single statistic, you can have a hit rate of almost 68% – far higher than the average hit rate for drafting WRs in the 1st round (which is just 27%).
You don’t see this same correlation to YPRR vs man coverage:
NFL YPRR vs College YPRR against Man (all NFL Wide Receivers drafted since 2018 with >200 routes ran)
This correlation has an R-squared of just 0.01 and you can see the greater spread in NFL YPRR at high YPRR vs Man, so the hit rate in finding good NFL WRs is not as high as it is when looking at YPRR vs Zone.
And, you see this similar issue when looking at Targeted Passer Rating vs Zone and Man (targeted passer rating is the passer rating of the QB when they target that specific WR). The correlation is much stronger for college WR production vs zone rather than man coverage.
NFL YPRR vs College Targeted Passer Rating against Zone and Man (all NFL Wide Receivers drafted since 2018 with >200 routes ran)
How a WR produces against zone coverages translates much more to NFL WR production than how the receiver produces against man coverage. However, I am not saying how a WR produces against man coverage doesn’t matter – it just matters much less.
Now that we see NFL WR production correlate differently to different college WR statistics, we can build a model to help predict NFL WR production. The caveat is that any model will have difficulty predicting NFL WR production – there are way too many variables (college offense/QB play, NFL offense/QB play, the human element of each draftee, etc).
But, if we can create a model to improve the hit rate of getting a good-to-elite WR then we should definitely use that – and, here is that model:
NFL YPRR vs Predicted NFL YPRR from College-to-NFL WR model (all NFL Wide Receivers drafted since 2018 with >200 routes ran)
This model has an R-squared value of 0.25 – not fantastic, but so much better than any individual statistic. The interesting thing about this model is for WRs who had Predicted NFL YPRR >1.5, only 3 out of those 26 WRs had NFL YPRR <1.5. So, we can have a hit rate of 88% of finding good-to-elite WRs by using this model, which is so much higher than the average hit rate for NFL WRs in the 1st round (just 27%).
There are some elite NFL WRs (>2.0 NFL YPRR) that did not meet that >1.5 Predicted NFL YPRR. Out of the truly elite NFL WRs, 9/15 met this model limit, or 60% (CeeDee Lamb was right on cusp though so the percentage may be closer to 66%), but when you start looking at WRs that had Predicted NFL YPPR <1.5, the variability in WR production gets very high.
NFL WR Tiers vs Predicted NFL WR Tiers (all NFL Wide Receivers drafted since 2018 with >200 routes ran)
For the College-to-NFL WR model, if a player is predicted as “Elite” (>1.5 Predicted NFL YPRR), then they have a 32% chance of being truly elite (>2.0 NFL YPRR) and 88% chance of being good-to-elite (>1.5 NFL YPRR) and only a 4% chance of being a bust (<1.0 NFL YPRR). For a predicted “Average-to-Good” WR, those percentages are just 9% for finding an elite WR, 22% for a good WR, and 43% for an average WR with a 26% chance of being a bust.
Now, remember that the hit rate for WRs in the 1st round is just 27% between 2000-2019. This model can increase that to 88% – it does this by focusing on the metrics that matter the most for NFL WR production, particularly how a WR produces against zone coverages.
Hit rate of 1st round draft picks 2000-2019
So, with this model – who are the top WRs in this year’s drafts? Which WRs have the best chance of being elite? There is one obvious player but then there are a few surprises – and why I think this draft may be deeper for WRs than the consensus. All of that will be in the next article.