It only seems logical to divide the full-size speed by the scale. Think of this: at 25 knots, the full-size boat will travel her own length, 26.25 feet, in 0.622 seconds. Ideally, the 1/15 boat should travel her own length, 21", in the same amount of time to be at the scale speed of the 25 knot full-size boat. This comes out to 1.92 actual statute miles per hour for the 1/15 boat. Roughly, 2 mph.
Using this scale speed calculator jibes with my reasoning:
http://www.mcr5.org/articals/speed.php
Alas - model boats are not logical! Or rather, they are, but you are going about things the wrong way. Scale effects are very important as we move into model sizes, and they completely negate the simplistic example you have provided. Perhaps I can quote a similar example by the famous boat modeller Harvey Adam, from 1951. I
précis a bit:
"...If a full-scale ship of 300ft travelling at 20mph is modelled at 1/100th, making it 36", it should travel at 1/5th mph - which is inadequate to make headway. It will work better at 3-4mph, but it will look ridiculous if it goes any faster. If a 36ft speedboat travelling at 40mph is modeled at 1/12th, similarly making it 36", it will be able to make headway at the scale speed of 3.3mph, but will not sit on the plane like the prototype. To do this the model needs to travel at about 10mph...."
This is a well-recognised issue whenever we consider relative sizes - read J S B Haldane on the subject:
http://www.phys.ufl.edu/courses/phy3...eRightSize.pdf
Many modellers use brushless motors at the moment - these provide high performance. The EeZeBilt series of boats mentioned above - roughly similar to yours, have the fast boats designed around an EMax 2822 KV1200 brushless motor with a 35mm prop. This gives a reasonable fast plane.
The problem with brushless motors, particularly small cheap ones, is that they do not have flexible low speeds - they are a bit like an I/C engine in this regard. If you want flexible low speeds, a brushed 380 - 400, as recommended above, will suit.