I think it was Partypooper that provided this commonsense explanation originally. It went something like this:

Lets say you have a strategy that loses 3% on turnover. And you play this strategy over a long period, say 20 years to eliminate the chance of variance in the results. So over the 20 year period it loses 3% on turnover.

Now, over this 20 year period, the bets you have 1 unit on will probably lose 3%, and the bets you have 2 units on will probably lose 3% as well, and the bets you have 3 units on will lose 3% as well.... on and on it goes. No matter how you work it, the results will even out over a long period and the same loss on turnover should apply across the board.

Makes sense to me.

You should ask Useless Bettor how his sports target betting approach ended up.