Tone sandhi is phonologically motivated, i.e., the tone changes arise from the pronunciation of the surrounding words, and are thus largely predictable. Cantonese changed tones, however, are generally lexically motivated, i.e., the changed tone is part of the realization of the word itself. Cantonese changed tones are thus more akin to Mandarin's erhua phenomenon (https://en.wikipedia.org/wiki/Erhua), which is also seemingly as random with regard to the words to which it applies.
> For Cantonese and Jyutping specifically, the tone sandhi is marked as, e.g., «faan4*2»
Note that the Jyutping standard actually doesn't specify how tone changes are marked. I believe the * convention originated at https://www.cantonese.sheik.co.uk/ to facilitate Cantonese learning. Wiktionary has adopted a similar convention but using a hyphen (-) instead (https://en.wiktionary.org/wiki/Wiktionary:About_Chinese/Cant...).