In theory, a wireless signal loses power with the distance squared (due to the power distributing over the surface of a hypothetical, expanding sphere). Using at least three different measurements you could calculate the cross section of all three reception circles and find the source. (In three dimensions, you'd need to use reception spheres and at least four measurements. Most often the earth can replace the fourth sphere.)
In practice, this is much harder. Space isn't empty and anything inside the Fresnel zone between the transmitter and the receiver attenuates the signal - even the earth if you're not using masts. Additionally, other transmitters can disturb the signal and cause what might appear like attenuation.
So, unless you very carefully model the environment (nearly impossible) you're left estimating. With some experience, these estimates can be surprisingly accurate but then again they can be totally wrong.
Another, completely different aspect is the gain of a directional antenna. Depending on the antenna geometry, some directions will get much more power than others.