Kemp Posted June 15, 2020 Posted June 15, 2020 Gen Stats tracks a metric called completion percentage above expectation. It looks at the probability of a completion on every throw, based on factors like how far the throw is, how open the receiver is and how much pressure the quarterback is under. It then comes up with an expected completion percentage and compares that number to the quarterback’s actual completion percentage. Allen ranked ranked 35th out of 39 quarterbacks.
YattaOkasan Posted June 15, 2020 Posted June 15, 2020 (edited) 2 hours ago, Kemp said: Gen Stats tracks a metric called completion percentage above expectation. It looks at the probability of a completion on every throw, based on factors like how far the throw is, how open the receiver is and how much pressure the quarterback is under. It then comes up with an expected completion percentage and compares that number to the quarterback’s actual completion percentage. Allen ranked ranked 35th out of 39 quarterbacks. I'm reading this differently. Josh is low on the expected completion percentage (29th of 39), but was one of the most negatively effected by those parameters as demonstrated in +/- (this is where he ranks 35th) which is just completion percentage - expected completion percentage (per glossary). So this means he had below average help with respect to line and separation. This better aligns with my eyes that while our line got better it was still only average and that while better our receivers didnt seem to get a ton of separation (also drops). Someone else let me know if I'm reading the link wrong. Edited June 15, 2020 by YattaOkasan
Billl Posted June 15, 2020 Posted June 15, 2020 2 hours ago, YattaOkasan said: I'm reading this differently. Josh is low on the expected completion percentage (29th of 39), but was one of the most negatively effected by those parameters as demonstrated in +/- (this is where he ranks 35th) which is just completion percentage - expected completion percentage (per glossary). So this means he had below average help with respect to line and separation. This better aligns with my eyes that while our line got better it was still only average and that while better our receivers didnt seem to get a ton of separation (also drops). Someone else let me know if I'm reading the link wrong. Unless David Blough, Gardner Minshew, Dwayne Haskins, and Josh Allen blew the rest of the QBs out of the water, I think it’s safe to say you’re reading it backwards. Those 4 had the biggest negative discrepancy between expected completion percentages and their actual completion percentages.
JESSEFEFFER Posted June 15, 2020 Posted June 15, 2020 2 hours ago, YattaOkasan said: I'm reading this differently. Josh is low on the expected completion percentage (29th of 39), but was one of the most negatively effected by those parameters as demonstrated in +/- (this is where he ranks 35th) which is just completion percentage - expected completion percentage (per glossary). So this means he had below average help with respect to line and separation. This better aligns with my eyes that while our line got better it was still only average and that while better our receivers didnt seem to get a ton of separation (also drops). Someone else let me know if I'm reading the link wrong. Supposedly protection and separation are accounted for in the XCOMP%. Some factors not accounted for would be drops, tipped passes and the weather, mostly wind effects. At least it is not stated that they are but maybe they are under the "and more" designation. Also, totally out of the mix is the decision made about where the ball went. So, there is a hodgepodge of factors that would effect both the comp% and the XCOMP%, including the accuracy of the throw. Completion Probability The probability of a pass completion, based on numerous factors such as receiver separation from the nearest defender, where the receiver is on the field, the separation the passer had at time of throw from the nearest pass rusher, and more. Expected Completion Percentage (xCOMP) Using a passer’s Completion Probability on every play, determine what a passer’s completion percentage is expected to be.
BigBuff423 Posted June 15, 2020 Posted June 15, 2020 I think Simms' placement of JA in the rankings is right....and I also think he absolutely "could" be top 5 QB in a year or two and I also don't agree that Mahomes' arm is stronger, but no way of knowing definitively. To me this is right where JA is at the current time: presently NFL average, but has potential to be Favre-esque or even better if all of the lights go on and he stays healthy with his trajectory continuing upward. But, he also could regress and that means the 18th rank stays in the unknown realm. Fair IMO all the way around.
Billl Posted June 15, 2020 Posted June 15, 2020 3 minutes ago, JESSEFEFFER said: Supposedly protection and separation are accounted for in the XCOMP%. Some factors not accounted for would be drops, tipped passes and the weather, mostly wind effects. At least it is not stated that they are but maybe they are under the "and more" designation. Also, totally out of the mix is the decision made about where the ball went. So, there is a hodgepodge of factors that would effect both the comp% and the XCOMP%, including the accuracy of the throw. Completion Probability The probability of a pass completion, based on numerous factors such as receiver separation from the nearest defender, where the receiver is on the field, the separation the passer had at time of throw from the nearest pass rusher, and more. Expected Completion Percentage (xCOMP) Using a passer’s Completion Probability on every play, determine what a passer’s completion percentage is expected to be. Interestingly enough, Josh and Dak were both expected to complete exactly 62.6% of their passes.
YattaOkasan Posted June 15, 2020 Posted June 15, 2020 10 minutes ago, Billl said: Unless David Blough, Gardner Minshew, Dwayne Haskins, and Josh Allen blew the rest of the QBs out of the water, I think it’s safe to say you’re reading it backwards. Those 4 had the biggest negative discrepancy between expected completion percentages and their actual completion percentages. Guess I’m confused. It says Josh should’ve had a higher completion percentage based on the other factors. For the QBs with a positive factor it says that their expected completion percentage was lower than actual. Pretty sure I got it correct. Other players with negative +/- (and closer to Allen than the ones you listed) are Mahomes and Jackson. Pretty sure that +/- is how mich your team helped or hindered.
Billl Posted June 15, 2020 Posted June 15, 2020 2 minutes ago, YattaOkasan said: Guess I’m confused. It says Josh should’ve had a higher completion percentage based on the other factors. For the QBs with a positive factor it says that their expected completion percentage was lower than actual. Pretty sure I got it correct. Other players with negative +/- (and closer to Allen than the ones you listed) are Mahomes and Jackson. Pretty sure that +/- is how mich your team helped or hindered. The last 3 columns show the actual completion percentage, the expected completion percentage, and how much better or worse than expected each QB did. Josh and Dak were expected to complete 62.6% of their passes when adjusting for things like drops, pressure, etc. Dak completed 65.1% (+2.5%) and Josh completed 58.9% (-3.7%).
JESSEFEFFER Posted June 15, 2020 Posted June 15, 2020 3 minutes ago, YattaOkasan said: Guess I’m confused. It says Josh should’ve had a higher completion percentage based on the other factors. For the QBs with a positive factor it says that their expected completion percentage was lower than actual. Pretty sure I got it correct. Other players with negative +/- (and closer to Allen than the ones you listed) are Mahomes and Jackson. Pretty sure that +/- is how mich your team helped or hindered. Well receiver separation and defender separation are accounted for. Receiver not making the catch is not, maybe. Nor is if the throw is made from a stationary position or while on the move, maybe. These could be covered under the "and more" factors in completion probability.
YattaOkasan Posted June 15, 2020 Posted June 15, 2020 (edited) 6 minutes ago, Billl said: The last 3 columns show the actual completion percentage, the expected completion percentage, and how much better or worse than expected each QB did. Josh and Dak were expected to complete 62.6% of their passes when adjusting for things like drops, pressure, etc. Dak completed 65.1% (+2.5%) and Josh completed 58.9% (-3.7%). Yes! So doesn’t that mean that Dak was helped by his line and WR And Josh was held back by his. To me based on numbers and eyes it was the drops that really hurt Joshs numbers. Someone might have the drop % relative to average but about 3% seems right. 3 minutes ago, JESSEFEFFER said: Well receiver separation and defender separation are accounted for. Receiver not making the catch is not, maybe. Nor is if the throw is made from a stationary position or while on the move, maybe. These could be covered under the "and more" factors in completion probability. Yeah I couldn’t make sense of what did and don’t go into it exactly. I think average line and separation and bad drop numbers makes sense to me if that’s what they are saying. Edited June 15, 2020 by YattaOkasan
JESSEFEFFER Posted June 15, 2020 Posted June 15, 2020 3 minutes ago, Billl said: The last 3 columns show the actual completion percentage, the expected completion percentage, and how much better or worse than expected each QB did. Josh and Dak were expected to complete 62.6% of their passes when adjusting for things like drops, pressure, etc. Dak completed 65.1% (+2.5%) and Josh completed 58.9% (-3.7%). It does not say that drops are factored in. They do not even list drop% among the WR/TE stats.
Billl Posted June 15, 2020 Posted June 15, 2020 3 minutes ago, YattaOkasan said: Yes! So doesn’t that mean that Dak was helped by his line and WR And Josh was held back by his. To me based on numbers and eyes it was the drops that really hurt Joshs numbers. Someone might have the drop % relative to average but about 3% seems right. Yeah I couldn’t make sense of what did and don’t go into it exactly. I think average line and separation and bad drop numbers makes sense to me if that’s what they are saying. No. It means that, when controlling for all of those factors, both should have been able to complete 62.6% of their passes. Dak was especially accurate allowing his receivers to catch more passes. Josh...wasn’t. He was one of the very worst in the league even when grading him on a curve that includes drops. Had his receivers dropped fewer passes, his expected completion percentage would have been higher than 62.6%. 1
JESSEFEFFER Posted June 15, 2020 Posted June 15, 2020 1 hour ago, Billl said: No. It means that, when controlling for all of those factors, both should have been able to complete 62.6% of their passes. Dak was especially accurate allowing his receivers to catch more passes. Josh...wasn’t. He was one of the very worst in the league even when grading him on a curve that includes drops. Had his receivers dropped fewer passes, his expected completion percentage would have been higher than 62.6%. Show me anywhere drops are mentioned. No sign that it was even tracked. 1
YattaOkasan Posted June 15, 2020 Posted June 15, 2020 2 hours ago, Billl said: No. It means that, when controlling for all of those factors, both should have been able to complete 62.6% of their passes. Dak was especially accurate allowing his receivers to catch more passes. Josh...wasn’t. He was one of the very worst in the league even when grading him on a curve that includes drops. Had his receivers dropped fewer passes, his expected completion percentage would have been higher than 62.6%. I see it now. Thanks. That diff is almost certainly drop differential between Josh and the average then as I expected before but was thinking of it wrong. Thanks again. Lots of signs pointing to Josh is average which is happily where I have him.
JESSEFEFFER Posted June 16, 2020 Posted June 16, 2020 Actually, from what I recall, Dak had a pretty high drop% as well. It is the result of accuracy in part. But, does the "and more" include any weather adjustment? Dak played 10 games in a climate controlled venue, 8 at home, NO and Detroit. Josh played 1, in Dallas. Does the "and more" adjust for batted balls and spikes/throwaways? Maybe. It would be logical to do so. Probability = 0 for those would make sense. How about tipped passes? Those are tough to identify.
Billl Posted June 16, 2020 Posted June 16, 2020 5 hours ago, JESSEFEFFER said: Show me anywhere drops are mentioned. No sign that it was even tracked. The variables that factor into the formula don’t matter for the sake of this question. What matters is that having a completion percentage below the expected level is bad and vice versa.
JESSEFEFFER Posted June 16, 2020 Posted June 16, 2020 6 minutes ago, Billl said: The variables that factor into the formula don’t matter for the sake of this question. What matters is that having a completion percentage below the expected level is bad and vice versa. Garbage in garbage out. Incomplete consideration of the inputs means that the outputs are questionable. 1
Recommended Posts