I do a good bit of financial/news scraping. I'd find this useful, as the OP is clearly a much cleaner read; so thanks for the demo! However I was always of the impression that large portions of what may be important data came in relatively unstructured and one needed to read the S1 with an eye for certain topics or inflection points? Do you find this to be true/do you think there are any heuristics to expose critical segments in a less supervised fashion?