Question. I read a magazine article that said that most commercial optimisers do not produce predictable results - is this true?
Answer. NO, definitely not. If you analyse your results over a period of time you will find that on average the results will be consistently better than manual methods.
What the article is referring to is that you may see occasionally a result that is not improved or even worse. This is usually caused by a very awkward combination of sizes and it is certainly difficult to predict when or if these unexpected results will occur. You will also find that the range of improvements for each set of data will vary quite a lot from marginal to dramatic and once again this very much depends on the sizes and parameter settings.