According to a study from Morningstar, expense ratios are a better predictor of future performance than their star rating system. But it turns out this is not so helpful. It really only tells us that all funds together in the high cost category do worse than all the funds together in the low cost category. The problem here is that each category, high cost and low cost, has superior funds and inferior funds. Unless bad manager systematically charge more than good managers we would expect each category at the aggregate level to earn roughly the market index return before costs. Therefore, at the category average level, the higher cost category of funds should (by simple arithmetic) end up with lower net returns after subtracting costs.
We can make use of the fact that lower costs are better than higher costs but it’s not enough by itself. What we want to do is find a set of funds expected to produce above market returns and then choose the lower cost funds within this category