Simple answer: Yes.
More complex answer: No.
Radio propagation is complex. In particular, a desired signal may be received by multiple paths. That means there are multiple directions in which the antenna may point and get improved reception. Some paths are better than others, so there's probably just one direction which is best, but without trying all the directions, you can't know for sure if you've found the absolute best direction, or just a local peak of a sub-optimal path.
Fortunately, it's usually pretty easy to try all the directions, and in very many situations there may not be significant paths in significantly different directions from the best one. Thus the simple answer, "yes".
There's also the additional problem that in some situations (HF, especially) the path is constantly changing. What the best path is now is different from what it will be in 30 seconds, so the best we can manage with a mechanically rotated antenna is to orient it in the direction that's pretty good most of the time.
Further complicating the issue is that orienting the antenna for maximum received signal power is not necessarily optimal for communication. What we are actually trying to optimize is the signal-to-noise ratio, so we also want to minimize received noise. That is, in some situations it may be more advantageous to aim a null at the noise, rather than aim a peak at the signal.
Very complicated answer: Yes.
Interestingly, for electronically phased antennas (that is, an array of antennas where each element is fed an independent signal), there exists an algorithm which can find the best (electronically selected) orientation of the antenna array if we know what the signal should be. Frequently this is achieved by periodically transmitting a synchronization signal.
This method is called minimum mean square error weighting. If we know what the signal should be, then we can look at what was actually received by each antenna and calculate weights for each antenna that most closely approximate the expected signal. This is very much like regression analysis.