This comes up from time to time. It has to do with how decimal numbers are represented in binary digital computers. There is an IEEE standard that Apple follows strictly in Numbers, versus doing some "massaging" of the results to better meet customer expectations. I believe it is the strict adherence to the standard that is the problem you see. Decimal numbers often cannot be represented exactly in binary and it results in small errors. As a test, do the following
B2 = .1
B3= -.3
B4= .2
B5 = SUM(B2:B4)
The result will not be zero
The answer may also change depending on the order of operations
501.3*0.15 -75 - 0.195 is equal to -6.8E-15
501.3*0.15 - 0.195 - 75 is equal to 0
Whether Apple is right or wrong in sticking strictly to the standard is debatable. There are cases in Excel where the "massaging" creates a false zero when the true result is supposed to be a very small number. My personal preference is for the numbers to be massaged to meet my expectations but that's not how it is in Numbers.