This is simply wrong. First of all, when people say bell curve, or Gaussian distribution or whatever, they never mean it literally. Nothing can ever have a perfect Gaussian distribution, as you would need an infinite sample size. However, loads of things are described to an excellent approximation by a bell curve. That is what people mean when they say something is described by a bell curve.

Secondly, there is no reason to assume the average is at half the maximum value. You can very well have a distribution which is symmetric, aside from being truncated at zero. In that case, the average strength would be slightly shifted away from the most likely strength (the top of the curve). However, it could still be perfectly reasonable to call it a bell curve.