Spend the last couple of weeks fine tuning some variables in the engine to try to restore sacks. In unit testing I also noticed that rushing stats have been low compared to NFL averages, so I made some tweaks to that as well. What I expect to see with this version:

1) Sacks should be back and should arrive around the NFL norm of 7% with normal play calling.
2) Rushing stats should in general grow from a league average of around 3 yards per carry up to the NFL average of around 4 yards per carry.
3) Completion percentage in my massive testing should be around 60%. I changed the logic a little to try to reduce the number of longer pass plays getting dumped to the outlet too quickly, and also tried to reduce the decisions for a QB to throw into tight coverage. Once we restore the DB play that is shelved, I imagine this logic will need to be revisited, but right now I just want to get a stable 0.4.6 out without those DB coverage updates.
4) Interception percentage should be around 2.5%, also in line with NFL stats.
5) Yards per attempt is low and I'm struggling to figure out how to bring it up.

To seth's comments in the previous version thread, yes, I do run lots of automated testing to verify that the stats come out reasonable. Depending on the goals of the changes, though, some stats (like YPA) may not be resolving to a satisfactory value just because I am not able to really understand why from such a macro view. That's where your film study is so helpful. Further, the best laid plans never survive the first interaction with actual humans running the engine... the stats with my unit testing never line up with what happens with humans running the teams. That's why I'm relying on you guys - my 'boots on the ground' - to help identify on a micro level things that are not working.

Oh, one more comment about this update - please someone continue to try to get the punt block exploit to work. I want to make sure the increasing of sacks is not also going to reinstate that exploit.