Adeniji-vs-Carman part II

[TecmoBengals]

(01-31-2022, 06:34 PM)Truck_1_0_1_ Wrote: They did not receive scores that low, despite playing significantly-worse.

Tecmo, as I've told the story dozens of times (many here can verify), I trained to work at PFF at the culmination of the 2012 season, as I was a MASSIVE booster of their grades and their wonderful stats that only they would track (even so, pass rush stats and running lane stats are only done by 1 or 2 other sites), from 2009 to 2015. The used to grade by starting at 0.0 and then going up or down, depending on the play; 0.1 or 0.2 for positive plays, up to 1.0 (and every number in-between), with negatives of those numbers for negative plays, thus you would have players grading +1.7 for a green game and -1.2 for a red game: -0.9 to +0.9 was considered a neutral game, just slightly-positive or negative.

This was SUCH a fantastic way to grade players, as the methodology was sound, the scale made sense and at the end of the year, you didn't have an, "average, weighted score," but instead a total score, culminating all of the games of the prior year. This method also allowed for non-finite grade totals (unlike now), as the grade could keep going up (or down), as good or as bad as the player played.

Once Collinsworth bought his stake in late 2014/early 2015, they went with the old system until week... 4? I think? of 2015 and then they went to the current one. The problem with the current grades are that:

- It is not told where they start the game, whether it is at 50, 0, 100, etc.; no context is given.
- Positive and negative plays can now be egregiously-high or low, again, with no context.
- You are seeing these ridiculous-low scores, for players that seem to play really good games.

There (seemingly!) appears to be no consistency or methodology and the grades are just left up to the opinion of the grader, thus with all of these reasons accounted for, I take 0 stock in their grades. I will view them, but I won't give them the time of day in a proper discussion on player's play. PFF is still a super-invaluable tool for pressure numbers and running lane numbers, but the grades are highly-speculative and (seemingly!) wildly-inaccurate, compared to how the used to be.

Why did I not work for them, you may ask? To get hired, you needed to show 95% success through the 3 rounds of grading, that you are able to identify the correct player. Once the correct players are identified, grading is just a matter of watching the play and grading accordingly. I made it through the first round with only 93.something% accuracy, but Khaled Elsayed (one of the current executives, based in the UK) allowed me to go the second round, as the quality of Gamepass at the time, was absolutely-dreadful. The second round had a different game and I was able to get up to 94.1% accuracy, but the difficulty of identifying the correct players, had me watching the game for about 17 hours, just to get the players right (and I still wasn't at 95%). Khaled wanted me to stick on the final round, but I pulled out because I knew I wouldn't be able to get good-enough quality video to do the job and it was taking precious time out of my life (I was in my final year of University as well) , so I dropped out of the running.

I know I sound curmudgeonly, but truly, the old way was the BEST way and it is significantly more-difficult to manipulate the old grade, because you had a finite amount of points you could grade a play on, unlike the wild west that is the current method of grading.

This was a fun and informative post to read, I'm going over it again. Thank you for sharing.