Myc*seshub has its problems for sure but from what i can tell (I do these sorts of integrations for my day job) it's probably looping through all possible receipt numbers within a certain time frame. If it gets a real receipt number, it's able to pull info on it, if not a real number, then the site gets an error.
The insight is to go to the /insights page and go to the Today's Lucky Ones widget. Put I-821D and IOE. This gives you a rolling list of approved DACA renewals that the app is pulling (there is probably more, but it at least can be used as a proxy i.e. 0-few listed means it was a slow day for USCIS renewals but dozens-100s means they went ham), I am unsure how often they are being pulled. I don't think it's daily, it might be every few hours.
The discovery i made is that while I-821D's in MCH do not have file dates 99% of the time, I-765's do like 50% of the time. ALSO, generally, the I-765 Receipt Number is one number higher than the I-821D's
Example: I-821D is IOE000000001 and the corresponding # for I-765 is IOE000000002. This is because most of the time these forms are filed concurrently so their system assigns them numbers as such.
THEREFORE, you can go to Today's Lucky Ones for a given day, click on the I-821Ds (it takes you to a url ending like /analysis/IOE0000000001) and check their corresponding I-765 by incrementing the last numbers of the # in the URL. And 50% of the time ish you will be able to see the corresponding file dates.
In this way, we can get a decent idea of what was being worked on that day (usually takes like 2 days to get as good a picture as possible)
So for example, yesterday 5/5, there were some Octobers being finished up, Nov19s (big date for filing) and Dec5s (3,4,5 is another hump it seems).
sorry this is kinda schizo and i only found this out cuz ive developed an anxious obsession over it, also i like puzzles, but i havent seen anyone else post about this and we've really only had other people's posts to use as a proxy for estimating approvals. This is another proxy I think, and less prone to bad faith posters. If you found this valuable, and find the time to go through this repetitive manual process (or create a AI agent or something to do it for you but BE WARY OF RATE LIMITS or having your IP blocked), then it'd be helpful if you could analyze it and post insights. I've been doing this for like 2 weeks now and the approach seems to be pretty inline with the anecdotal evidence (people making posts about their timelines/approvals).
That's all for now kthx bye