Working off the assumption that the random crashes are caused by possible divide by zero errors, we have made some changes.
Areas that could possibly result in a divide by zero, but have not in any of our testing, have had checks put in place to prevent even the possibility.
Hopefully this resolves any future random crashes.
We are continuing to monitor and attempting to recreate the crashes.