Well, this message was supposed to be a PM but i am on vacation and over a slow internet connection and something got messed up.
So full disclosure, I sent the paper to folks for informal review and got no complaints. However, I realized that my dependent variables (3 and 7 day means) are correlated and I am changing this. They should be a 3 day mean centered around day 3 in the future and likewise for 7 but also using a 3 day mean. This will uncorrelate the variables. I dont think this will make difference in the result but I don't like having these loose ends.
Will fix upon return of vacation. Look for updates in the next week in github.