Hypothetical example would be selling ES today 1939 (previous reaction high) with a 27 point stop at 1966, (1939 - 1803)/5=27 points. Assuming probability @ 20% and 5:1 reward we could allocate 4% of trading account to each trade. Let's consider that being equal to 1 lot based on current range based target and stop distance.
It's not even wrong. In the sense that it's far from being concrete, applied. So in theory that sounds good. But it is too abstract for being an edge. Theory without practice is worthless. I think it's bad to bet everything on one peace of a puzzle. I've done that before. Focusing on asymmetries and convexity. But I wasn't improving my trading. Because it's always gonna be as weak as the weakest link. So a good thing to do is to avoid being fooled. Exemple, in believing an edge is about one and only one thing. Especially when risk management rely on outside information, conditionals. It's not self sufficient.
This call travelled around 94% to target yesterday and broke down to 2 points from stop loss, today we are ringing the register with +20% @ 2013 ES. Tally: -4%, +20%, +20%
I'd say ... Evaluate the P(G) of your system. And if the average R:R is less than 1/P(G) Then DO NOT trade your system. The ratio may be your only edge, But it's not the only variable that matters.