I think a lot of people use similar strategy without crunching the numbers to the level some of you guys do. It's possible we are losing out a little bit by not doing so. The bottom line is, if you hit most of your big plays, your ROI is going to look pretty good, but your ROI can't tell you which games to bet big. I would actually remove from the calculations any games where you bet a significantly larger amount than normal, because win or lose, it will influence your ROI disproportionately. I guess statisticians would call these outliers.
Of course you still have to factor the win or loss into your bankroll. If you felt so strongly about a play that you risked 10 units on it when your normal range is 1-5 units, and you win, you can't reliably expect that this will happen all the time. This brings up a separate discussion about money management, but every gambler has a different objective that determines how much of a risk they want to take with their picks. I'm guessing that a lot of the guys who employ ROI as a rule probably wouldn't make a play that is so much higher than their normal bet.
Of course you still have to factor the win or loss into your bankroll. If you felt so strongly about a play that you risked 10 units on it when your normal range is 1-5 units, and you win, you can't reliably expect that this will happen all the time. This brings up a separate discussion about money management, but every gambler has a different objective that determines how much of a risk they want to take with their picks. I'm guessing that a lot of the guys who employ ROI as a rule probably wouldn't make a play that is so much higher than their normal bet.