On wkstats, I added an event each time the user navigates to the Projections page. You can use it like this:
//--------------------
wkof.on('wkstats.projections.loaded', your_startup_function);
function your_startup_function() {
// Your script's startup code here
}
//--------------------
[edit: oh, and I fixed the srs_stages issue, so the projections page’s calculations are now visible in the Javascript console again]
Yes, it is better, but still not quite right. It currently says that the fastest I can level up is NOW. But actually, I have a few kanji that I haven’t even done the lessons for yet, so even if I did them now, the fastest I could level up would be a few days from now.
And the right and left columns are still showing dates from last year.
Silly question, may have messed sth up but I seem to travel back in time with each level? 
Any reason why this may be happening?
Just pushed version 1.1.1 to add a debug current level view. If you would like to report problems with the projection for the fastest possible projection for your level + 1, please include a screenshot of this as well.
@Greti33 I don’t know. I can guess that this value for level 31 is in Central European Time and that this is the value of new Date(0) but I don’t have the slightest idea why that’s happening to you.
I’m a little concerned by your fastest as well: levels 31-39 shouldn’t be fast levels!
@rfindley Thank you for this. Unfortunately, I’ve played around with it, but I’m kind of stumped.
To anyone, this script does have a GitHub. I made this for personal convenience and hoped it might help others. If you know how to fix something wrong, please submit a pull request there! 
so happy this exists and sad at the same time…(realistic vs fastest)… thanks for putting this together…it’s not perfect but with my slowness it’s plenty accurate enough
2024… (maybe sooner with 50-60 but at least I’ve never reset)
Hahaha, that’s fine - I kinda like this power anyways 
I just fear that I may be unable to continue my WK journey once I get yote back into 1970, pre-WK launch.
But srsly now, thanks so much for putting this together (and unleashing my true potential) - it’s nice to be able to look at projections after so long! :333 I’ll try to have a look later myself, perhaps I can help out a lil :3
I updated the script to use @rfindley’s new callback to load the script once the projection page is loaded. I also found a bug where if the script had been active and another page on the site was loaded after it, some tables under Items would fail to load. I found out a modification of the Array prototype was causing it, so I changed the added methods to normal functions for now which seems to solve the issue. I made a merge request on the github for it 
@BIsTheAnswer Thank you. 1.1.2 is now live and your PR has been merged.
@rfindley Good catch, this has been fixed in 1.1.3.
The fast level mismatch is not specific to me though. If you look at shuly’s screenshot, fastest start date for 45 is Feb 17, and for 46 it’s Feb 20, only 3 days later.
Mine still claims I should have levelled up hours ago:
Interestingly, the kanji analysis looks okay as far as dates go (the times are a little off):
I also notice that 48 appears to be a fast level, when according to the autoupdated list of fast levels it isn’t. And 49 seems to be a slow level when it’s in fact fast. 52 also appears slow when it isn’t (always according to that list of fast levels).
Thank you for the analysis! I don’t know why I didn’t catch it in your console output yesterday, but I think I fixed this in 1.1.3.
1.1.3 is now live to fix undefined sort behavior, which could cause problems in the median and fastest possible projections.
@Greti33 @yatyisam Does this help with your issues by any chance?
It works perfectly now, thank you 
(post deleted by author)
Ah yes, apologies for my delayed response - it seems now I no longer timetravel backwards :33 Thanks for sorting it out so quickly!!!
Version 1.1.4 is now live to ensure the other projections are not before the fastest possible. This might occur if you spend a long time on the current level, or if you set the hypothetical to something impossible.
interesting…
with the updates seems to predict more accurate info w/o needing the hypothetical … if this was intentional…nice job…if not then … it’s still nice as slow as it is…it’s realistic (for me anyway)
I think that’s due to 1.1.3. I’m glad to see your output looking better!
can you make me learn all this faster though?
probably just asking a tad too much
creates userscript that accepts x as a correct answer for every item
I can use the ignore button for that hehe…
that’s the problem…haven’t been cheating
