I've been colour contrast checking UI's since before the Internet caught on; since I worked on and with the UI standard that was co-developed by IBM, M$, and Apple for both Windows and Mac in the late '80's... which itself was derived from the VGA specifications RGB to greyscale conversion. (back when monochrome VGA was actually more popular with publishers than colour as you had more depth... the conversion on most cards was analog via resistors, resulting in all 18 bits converted to around 13 bits actual usable depth)
Mix that with a decade of working in accessibility, I can just look at colours and go "uhm.. no"
The math for it -- at least the WCAG's version -- is too complex for its own damned good (and based on the wrong conversion formula, they overvalue red)... but at least we have tools to check it like the one at WebAIM.
Building my own tool for this is on my to-do list, particularly since I want to implement my own separate formula that's based on YCrCb and therefor doesn't overvalue red. It's possible with the WCAG version to have colours that fail the deuteranomoly test.
I also want to make it so you can point it at fonts to check how rendering effects the colour. A LOT of webfonts even at colour declarations that should deliver AAA compliance actually render thanks to font-smoothing far below even AA large minimums.
It's a bigger -- and more important -- topic than most people think.