Using work from someone else's talk for oneself

Give the impression that the preprint is almost ready so that big-wig sees no chance to be first.

Assuming you have results that are developed enough that there is something to steal, why are you giving talks before writing it up? Write it up and put a preprint on arXiv or similar. It doesn't have to be in 100% ready to submit form, just include the important parts that you are trying to establish precedence for.

Alternately, give a big-picture overview that does not give enough small details that your results could be recreated easily. Presumably your work would take even a big-wig months of compiling data, running computations, or SOMETHING before they could write a paper and submit it before you.