I think it would be less strong than a wavy wall of similar brick count, but still more efficient than an equivalent strength wall built in a straight line.
My mental reasoning for this is that a (pseudo) sinusoid spends a lot more of its path further away from the centre. Thinking of it as a point moving along the path through time, it will dwell and the peaks, and cruise through he centre. The contribution of each brick to wall stiffness will be related to the cube of the distance from the centre line (neutral axis), so more 'time' spent at the peaks is best. This holds true on the macro scale, but could vary on the scale of a half 'wavelength' as the lack of inversion of curvature could be beneficial there.
Everything moderately reasonable seems to be better than a straight line in this instance. In the limit, two much thinner walls, far apart, is the optimal solution, but that becomes unreasonable as those walls must be coupled together to provide strength.