This presupposes that past performance predicts future results. In a world where we have incomplete information and a limited timeframe to assess relative to the universe of possibilities (the real world), a 60% chance is effectively indistinguishable from a coin flip. Once you introduce the possibility of an antagonistic opponent who might flip the odds at any given time, and account for the early mover's advantage of being first to adapt to new information, acting on what's always worked in the past starts to look naive.