<<Another thing that I stubbed out for Big Data that I may want to come back to some day.>>
I don't have time to write out a full post on this yet. But I wanted to start stubbing out some notes on accuracy, which like speed, is also parabolic.
A few years I wrote this post about how passing accuracy as an attribute provides diminishing returns based on the QB/team experience with a play:
https://bigdata.myfootballnow.com/community/1/2290?page=1#15119In 4.5, when accuracy was transitioned to parabolic, JDB made this comment about it:
jdavidbakr wrote:
It's not the only attribute that matters, but it is very difficult to overcome a low accuracy rating. Accuracy is exponential, so ratings less than 50 are probably going to be almost unserviceable. I might need to look at the player generation algorithm to try to make sure that QBs are not generated with such low accuracy.
Accuracy is described as exponential here, but is a downward opening parabola like speed. From this thread:
https://mfn1.myfootballnow.com/community/3/2443?page=9I hadn't put much thought into it before, but JDB's comment about ratings less than 50 (bolded above) now makes me wonder whether the X intercept of the parabola was set to (50, 0) so that with a display of 50, a QB functionally has 0 accuracy. This seems to line up somewhat with the crazy results you see when a QB has < 50 accuracy.
I'll actually put down a few examples of this later, but thinking about accuracy in this way at least helps visualize what good minimum attributes a QB needs to be successful (which I think is somewhere between 70 and 80 as long as they have high Intelligence.
