bicknyers
No way 10/11 for a 10/23 sign up, is it linear regression or did you fit a quadratic/exponential? I think more weight to later dates, since there was a lot of variation in the constrained supply early on.
That was my reaction too when I first punched in the date. After some thought though, the drops have not been consistent. Sometimes drops a couple of weeks in a row, then nothing at all for a couple of weeks. Typically the drops cover a couple of days at most. Under what we've been seeing historically, 10/11 for a 10/23 sign up is not actually crazy.
It's a linear regression plot - quadratic/linear had some crazy results for some reason. I had also plotted only more recent points to try seeing how that affected the outcome. Funny thing was, the latest iteration had both plots give me the same predicted date (sheer coincidence - I think). In the end it doesn't really matter. Trying to accurately predict these drop dates considering more than logical systems are at play here is just lunacy. In my opinion, with something so complex, it's just better to keep it simple and accept it's flawed.
As mentioned before, if things have actually changed, then yes, it would break a linear formula. That also means you can expect your card sooner than 3 months from now. At least that's a ballpark window.