Ok, got it. But if you're more likely to win after an initial win than to lose after an initial win (since I'm assuming that a streak of 3-5 is more likely than a streak of 1-2), wouldn't you want to bet more after your first win than after your first loss?
The only (non-contrived) way I can imagine for such a method to work is that it would have some kind of memory; for example, it incorporates new information after every win and it incorporates this information better/faster than the sports books. But if that's the case, then a win should not only be more likely after another win but also more likely than after a loss, so you wouldn't have streaks of losses. This is one reason why I'm having trouble imagining that such a method could exist.
I'm skeptical when I hear mention of methods that generate streaks of wins/losses because people (myself included) tend to see patterns in random processes where there aren't patterns. The statistician Robert Abelson puts it well: "
Chance is lumpy".