We present a quantum implementation of Parrondo's game with randomly switched strategies using 1) a quantum walk as a source of "randomness" and 2) a completely positive (CP) map as a randomized evolution. The game exhibits the same paradox as in the classical setting where a combination of two losing strategies might result in a winning strategy. We show that the CP-map scheme leads to significantly lower net gain than the quantum walk scheme.