[OC] Win Pct Added (WPA): Spin Cycle Edition
Jun 30, 2022 21:32:18 GMT -6
Funkytown, Danchat, and 2 more like this
Post by Reignman on Jun 30, 2022 21:32:18 GMT -6
You've all seen those fancy graphs in the boxscores that show the win probabilty of both teams throughout a game.
Well they calculate that per play, using magic or something, I have no idea. All I know is I have the data, so just for the fun of it I used it to do some calculating, as you can see below in exhibit A.
Exhibit A:
That's average WPA per pass (not including sacks) for the entire 2021 season, sorted by the average rank of the 5 categories. The rank of each category is in parenthesis.
Some of the results are kinda shocking, I know. In fact it doesn't align with the narrative I set out to spin so I almost swept it under the rug like I normally do, but I did all the work and I hate wasting time, so I decided to try to come up with some excuses instead. Because what's the point of doing something if you can't confirm your bias?
First, I know what you're thinking, "but Reign, maybe you made a mistake?" ...
... oooook, now that all the silliness is out of the way, there are some serious flaws using this metric in this manner. For one, it's biased toward close games, especially in the 4th quarter. Take that chart from the Minneapolis Miracle. That one play accounted for 96.6 WPA, whereas a 60 yard TD on the final play of a game when you're up or down by 20 would add 0.
Having said that, the Vikings led the NFL in plays (76.6% NFL avg 63.1%) where the game was within one possession (down by 8 to up by 8) overall, and in the 4th quarter (21.2% NFL avg 12.9%).
It's also heavily biased in favor of the trailing team at a 42.5 to 15.6 ratio, and guess what the Vikings did a lot of in 2021? Haha j/k. They were 18th in trailing (41.5% of plays) and 19th at trailing in the 4th quarter (47.6%). So oh my, Trevor Manziel, err I mean Lawrence must be worse than we thought.
Well it's not enough to just be trailing, you still gotta make the plays.
So there you have it, bias confirmed. Cousins benefited from all the close games. Case closed.
If you want to know the WPA for any other scenario, give me the criteria.
Well they calculate that per play, using magic or something, I have no idea. All I know is I have the data, so just for the fun of it I used it to do some calculating, as you can see below in exhibit A.
Exhibit A:
ALL PASSES 3RD DOWN 4TH QUART LEADING TRAILING AVG RANK
J.Burrow 1.308 ( 1) 1.689 ( 2) 1.403 ( 3) 0.563 ( 4) 1.367 ( 2) 2.4
A.Rodgers 1.043 ( 2) 1.576 ( 3) 1.189 ( 4) 0.684 ( 3) 1.832 ( 1) 2.6
J.Garoppolo 1.015 ( 3) 1.704 ( 1) 1.439 ( 2) 0.332 (13) 1.366 ( 3) 4.4
M.Stafford 0.773 ( 7) 1.544 ( 4) 1.522 ( 1) 0.421 ( 9) 0.814 ( 9) 6.0
D.Carr 0.777 ( 6) 0.894 (11) 1.106 ( 5) 0.476 ( 6) 0.543 (13) 8.2
K.Cousins 0.841 ( 4) 1.253 ( 5) 0.881 ( 9) 0.180 (21) 1.072 ( 6) 9.0
P.Mahomes 0.675 (12) 0.826 (13) 0.922 ( 8) 0.372 (11) 0.894 ( 7) 10.2
T.Brady 0.680 (10) 1.146 ( 6) 0.650 (12) 0.234 (18) 0.690 (11) 11.4
J.Herbert 0.821 ( 5) 0.851 (12) 1.088 ( 6) -0.062 (32) 1.229 ( 4) 11.8
R.Tannehill 0.527 (14) 0.678 (14) 0.985 ( 7) 0.509 ( 5) 0.283 (23) 12.6
T.Tagovailoa 0.739 ( 8) 1.080 ( 7) 0.836 (10) 0.120 (26) 0.553 (12) 12.6
M.Ryan 0.520 (15) 0.978 ( 9) 0.492 (16) 0.426 ( 8) 0.431 (17) 13.0
D.Prescott 0.677 (11) 0.640 (16) 0.398 (20) 0.338 (12) 0.542 (14) 14.6
K.Murray 0.585 (13) 1.079 ( 8) 0.261 (25) 0.240 (17) 0.732 (10) 14.6
L.Jackson 0.701 ( 9) 0.152 (24) 0.763 (11) -0.031 (30) 1.140 ( 5) 15.8
T.Bridgewater 0.417 (18) 0.640 (15) 0.426 (19) 0.808 ( 1) 0.100 (28) 16.2
D.Jones 0.465 (17) 0.902 (10) -0.132 (30) 0.330 (14) 0.408 (19) 18.0
M.Jones 0.340 (21) 0.364 (18) 0.558 (15) 0.123 (25) 0.469 (15) 18.8
A.Dalton 0.414 (19) -0.567 (32) 0.581 (14) 0.135 (23) 0.879 ( 8) 19.2
R.Wilson 0.344 (20) 0.219 (21) -0.273 (31) 0.725 ( 2) 0.331 (22) 19.2
C.Wentz 0.214 (25) 0.591 (17) 0.336 (21) 0.382 (10) 0.149 (27) 20.0
J.Goff 0.515 (16) 0.105 (25) 0.434 (18) 0.192 (20) 0.333 (21) 20.0
J.Allen 0.276 (23) 0.175 (23) 0.451 (17) 0.208 (19) 0.422 (18) 20.0
B.Roethlisberger 0.235 (24) 0.272 (19) 0.643 (13) 0.077 (27) 0.361 (20) 20.6
D.Mills 0.156 (28) 0.178 (22) 0.013 (28) 0.435 ( 7) 0.009 (29) 22.8
T.Heinicke 0.292 (22) 0.262 (20) -0.106 (29) -0.034 (31) 0.444 (16) 23.6
J.Hurts 0.145 (29) -0.474 (31) 0.240 (26) 0.301 (15) 0.175 (26) 25.4
J.Fields 0.210 (26) -0.116 (28) 0.330 (22) -0.022 (29) 0.264 (24) 25.8
J.Brissett 0.180 (27) -0.211 (29) 0.263 (24) 0.060 (28) 0.208 (25) 26.6
S.Darnold 0.079 (30) 0.051 (27) 0.135 (27) 0.248 (16) -0.290 (33) 26.6
Z.Wilson 0.067 (31) 0.053 (26) 0.310 (23) 0.127 (24) -0.040 (30) 26.8
B.Mayfield 0.063 (32) -0.774 (33) -1.091 (33) 0.149 (22) -0.286 (32) 30.4
T.Lawrence -0.044 (33) -0.241 (30) -0.304 (32) -0.983 (33) -0.087 (31) 31.8
That's average WPA per pass (not including sacks) for the entire 2021 season, sorted by the average rank of the 5 categories. The rank of each category is in parenthesis.
Some of the results are kinda shocking, I know. In fact it doesn't align with the narrative I set out to spin so I almost swept it under the rug like I normally do, but I did all the work and I hate wasting time, so I decided to try to come up with some excuses instead. Because what's the point of doing something if you can't confirm your bias?
First, I know what you're thinking, "but Reign, maybe you made a mistake?" ...
... oooook, now that all the silliness is out of the way, there are some serious flaws using this metric in this manner. For one, it's biased toward close games, especially in the 4th quarter. Take that chart from the Minneapolis Miracle. That one play accounted for 96.6 WPA, whereas a 60 yard TD on the final play of a game when you're up or down by 20 would add 0.Having said that, the Vikings led the NFL in plays (76.6% NFL avg 63.1%) where the game was within one possession (down by 8 to up by 8) overall, and in the 4th quarter (21.2% NFL avg 12.9%).
It's also heavily biased in favor of the trailing team at a 42.5 to 15.6 ratio, and guess what the Vikings did a lot of in 2021? Haha j/k. They were 18th in trailing (41.5% of plays) and 19th at trailing in the 4th quarter (47.6%). So oh my, Trevor Manziel, err I mean Lawrence must be worse than we thought.
Well it's not enough to just be trailing, you still gotta make the plays.So there you have it, bias confirmed. Cousins benefited from all the close games. Case closed.
If you want to know the WPA for any other scenario, give me the criteria.
... and neither is this Chris guy.