Why RYPTPA matters for WR prospecting

RYPTPA stands for “receiving yards per team pass attempt”, and it is a stat that Peter Howard (@pahowdy) has been pushing lately, specifically to evaluate college wide receiver prospects. It’s a number that captures both volume and efficiency at the college level, not unlike dominator ratings and market share numbers. However, in this article I will argue that RYPTPA is better than yards market share specifically (and by extension dominator as well) because it inherently adjusts for teammate score. In other words, if a college WR was surrounded by significantly great or poor teammates, RYPTPA will naturally adjust for the situation and give a number that better captures the WR’s true talent.

I’ll mostly be comparing RYPTPA to yards market share because of how similar their formulae are. RYPTPA and yards MS both normalize raw receiving yard production by adjusting for team context (RYPTPA via team pass attempts, yards MS via team receiving yards) as shown in their derivations below.

As you can see from their formulae, RYPTPA essentially takes yards MS and divides by (read: normalizes by) team efficiency. I argue that viewed another way, RYPTPA, to an extent, corrects for specifically the level of teammate talent. My addition of LaTeX formulae perhaps makes this article out to be more quantitative than it is, but I will demonstrate my argument primarily with an easy example with numbers.

Example

Consider a team with three really good receivers and our goal is evaluate stud WR “Andy”. It’s likely that because the good receivers are vulturing each other’s yards and production such that each player’s yards MS is lower as a result. Let’s say here are the stat lines over the course of one season:

WR Prospect “Andy”: 60 rec, 1100 yards = 28% yards MS = 2.75 RYPTPA
Other WR “Bob”: 50 rec, 900 yards = 23% yards MS = 2.25 RYPTPA
Other WR “Chuck”: 60 rec, 1000 yards = 25% yards MS = 2.5 RYPTPA
Other players: 100 rec, 1000 yards
QB: 270/400, 4000 yards

All three of these receivers put up solid receiving yard production, and their yards MS are shown above. A 28% yards MS is good but is roughly average for a WR prospect expecting to get drafted (~50th percentile).

---

Let’s consider an alternate scenario where Andy plays at a team with similar QB play but an average WR corps. In this case, we should expect 1) Andy to have more raw production and 2) the QB to be less efficient with poorer talent. To put numbers to it:

WR Prospect “Andy”: 65 rec, 1200 yards = 32% yards MS = 2.67 RYPTPA
Other WR “Dilly II”: 50 rec, 700 yards = 19% yards MS = 1.56 RYPTPA
Other WR “Ernie”: 55 rec, 800 yards = 22% yards MS = 1.78 RYPTPA
Other players: 100 rec, 1000 yards
QB: 280/450, 3700 yards

By essentially “airdropping” Andy into an average offense, we see that market share vastly improved, which is to be expected. However, RYPTPA stayed relatively stable. This is because the formula for RYPTPA looks only at Andy’s production and the QB’s volume, two factors that are more inherently stable and neutralizing than the strength of the wide receiver corps, whose variability the yards MS is subject to.

---

Finally, let’s “airdrop” our stud prospect Andy into an offense with similar QB play but inefficient receivers, i.e. drag route and bubble screen specialists. In this scenario, we should expect 1) Andy to maintain decently strong production given that he’s the go-to guy and 2) the QB to be perhaps more accurate but less efficient as a whole.

WR Prospect “Andy”: 70 rec, 1100 yards = 37% yards MS = 2.62 RYPTPA
Other WR “Ferb”: 50 rec, 500 yards = 17% yards MS = 1.19 RYPTPA
Other WR “Gil”: 55 rec, 500 yards = 17% yards MS = 1.19 RYPTPA
Other players: 100 rec, 900 yards
QB: 275/420, 3000 yards

Once again, we see Andy’s yards MS increase as he’s the only good receiver in this offense and therefore takes up more yardage share. However, since Andy still put up ~1000 yards and the QB still threw the ball ~425 attempts, Andy’s RYPTPA remains a stable indication of his talent given this level of QB play. On the other hand with yards MS, we’ve seen how wildly it can fluctuate given the talent level of the wide receiver corps surrounding Andy.

Other considerations

1. The astute reader would recognize that in each of the three above scenarios, I had to specify that the QB play does not change. Differences in QB play across teams (think LSU vs. TCU 2019) is something that RYPTPA cannot account for! This aspect is where market share and dominator ratings excels and RYPTPA does not, and that makes sense: a great receiver regardless of QB talent, and all else equal (read: assuming wide receiver corps are equally talented across the league), will have a great market share or dominator.However, RYPTPA can be reasonably expected to be affected by QB performance. It makes intuitive sense on a macro level: the team’s total receiving yards per pass attempt is lower because of poor and inefficient QB play. On a micro level: great receivers can still post great stats with a poor QB. However as a QB’s completion percentage goes down, we can reasonably expect Andy’s receiving yard total goes down and possibly expect team pass attempts to even go up — both of which will lower the receiver’s RYPTPA.
2. Another detail that I want to mention is why I chose yards market share specifically to compare against RYPTPA. Simply put, they are very similar in how they are calculated. One of the most often used stat is simply called “college dominator” and this is calculated by averaging TD dominator (itself an average of the best season’s and last season’s TD market share) with yards dominator (itself an average of the best season’s and last season’s yards market share). At the core of dominator ratings, regardless of how which dominator, are market share calculations, and therefore are subject to the same pitfalls as market share — that it can change depending strength of play by teammates.  It is well documented that college dominator is affected by teammate play, and it has led to analysts coming up with specific metrics to account for this phenomenon (for example, the Breakout Finder’s Teammate Score). Therefore, the benefits of RYPTPA over yards MS applies to dominator rating as well, making RYPTPA a useful tool to account for noise caused by teammate production.
3. In my opening paragraph, I declared that RYPTPA is a better metric than yards MS and dominator by extension. However, you may argue that all I’ve shown you is that it’s better at certain things than dominator. What if I tell you that RYPTPA is better at predicting future NFL production? Stay tuned…