^^ nice, however that was explicitly forbidden by FD, you naughty boy!!! Tut Tut!!!
(Why do you think we're not all ripping the Companion App API to shreds for our tools?).
Ok, so the way I saw it was this. The companion app is out there, and is likely getting next to no use whatsoever, as it's not particularly brilliant at the moment. I figured a script which made one call to the JSON data every five minutes was very unlikely to cause a load issue.
Now for what you're doing with the (excellent, btw) trade tool, you'd need to facilitate everyone who uses your tool to hammer the API, rather than just for you alone, and this is why Frontier didn't want it to become widespread. The moment you shink-wrap this for anyone to use, THEN it would cause a problem, and by the API's nature, you need many calls to aggregate useful data.
Anyway, better to do something and seek forgiveness than ask permission in my book (sorry Michael!!)
My point with posting a basic graph to this thread is that there's a lot of supposition of what would make a good API, and having some material exploration of the data would be hugely beneficial in sorting out what would and wouldn't work.
My friend Tom and I are knocking around with the code at the moment, and it's up on Github at https://github.com/infovore/ed_investigation , minus the scraper, which is being kept closed for the reasons above.