I appreciate the effort, but it’s a terrible metric.
You have all this data and you decide that 3 factors are enough to rank receivers? Low volume doesn’t seem to bother you at all. Beckham at 10? That’s a hot take. But even within your own metric and data, it’s incomprehensible how you generate your scores.
For example lets compare AJ Brown and Tyreek Hill:
Hill (ngs)
Hill (your score)
A.J.Brown (ngs)
A.J. (your score)
SEP / Open
3.3
86
2.5
86
Catchrate %
73.33
59
70.83
99
YAC
6.2
46
4.8
61
You may have used xYac/R +/- as yac data, which makes some more sense.
Hill (ngs)
Hill (your score)
A.J.Brown (ngs)
A.J. (your score)
YACOverExpted
1.1
46
2.1
61
But even if you chose xYAC/r +/- that’s 1 out of 3. Catchrate and Openness are still incomprehensible.
Why didn’t you use, for example, yds/route run as an additional input, that’s a nice handy score to work with. Instead you’ve gone with separation, which is a terrible indicator of WR skill because it depends on how a receiver is covered, so there’s always a second person influencing the data. Bad receivers -> bad coverage -> easy separation. Good receivers -> better coverage -> harder separation. Elite receiver -> best coverage + game plan adjustment -> hardest separation. Not all seperations are created equal.
I wouldn’t mind if your metric produced some reasonable results, but if you compare your results to dvoa or dyar for example, you’ll see how far off you are from those stats.
I appreciate the effort, but it’s a terrible metric.
You have all this data and you decide that 3 factors are enough to rank receivers? Low volume doesn’t seem to bother you at all. Beckham at 10? That’s a hot take. But even within your own metric and data, it’s incomprehensible how you generate your scores.
For example lets compare AJ Brown and Tyreek Hill:
You may have used xYac/R +/- as yac data, which makes some more sense.
But even if you chose xYAC/r +/- that’s 1 out of 3. Catchrate and Openness are still incomprehensible.
Why didn’t you use, for example, yds/route run as an additional input, that’s a nice handy score to work with. Instead you’ve gone with separation, which is a terrible indicator of WR skill because it depends on how a receiver is covered, so there’s always a second person influencing the data. Bad receivers -> bad coverage -> easy separation. Good receivers -> better coverage -> harder separation. Elite receiver -> best coverage + game plan adjustment -> hardest separation. Not all seperations are created equal.
I wouldn’t mind if your metric produced some reasonable results, but if you compare your results to dvoa or dyar for example, you’ll see how far off you are from those stats.