Draw some sketches of familiar curves and note that the condition in the question can be interpreted geometrically as the function and the gradient having opposite signs at each point along the curve.

So, the question can be rephrased as: can the sign of a curve and the sign of its gradient be opposite everywhere?