We extend to stochastic integral processes 
with deterministic integrands  the results previously
proved by Graversen, Peskir and Shiryaev, 
on stopping Brownian motion as close as possible 
to the ultimate value
of a functional.