If you could consistently make money by shorting stocks that are about to fall off an index, the advantage would arbitraged to oblivion.

The question is whether you know that the stocks are about to fall off the index before other market participants. If your high frequency trading algorithm is the first to know that a stock is about to fall off an index, than you make money with it.

Using the effect to make money isn't easy because it requires having information before other market participants. That doesn't change anything about whether the index funds on average lose money on trades to update their portfolio to index changes.