snip~
Let it go, bro!... this is already boring.
Sorry, but you make it sound like "I already proved you whatever, let's move on".
But you are in a huge, humongous error.
Again: if I have 1 win with method A in 1000 simulations, with 50 million total ops.
And 999 wins with method B in 1000 simulations, with 50 million total ops.
Then on average, they both have the exact same performance:
50 mil ops / 1000 sim = 50 thousand ops / simulation.
Your formula, on the other hand, is absurd, showing that method A runs in 50 million ops / win, and method B runs in ~ 50000 ops / win, which is totally false. Method A ran 50 million ops for ALL the sims, not just for the win.
Get it now?
The goal is to evaluate which method is better, so it makes sense to calculate the average per victory, as victories are the indicator of how effective each method is. This is ideal for identifying which method is more efficient in achieving its objective. When you globalize everything, you move away from the main objective by calculating the average across simulations.
Whenn the goal is to determine which method has a better success rate.
Or does winning not matter? If we include simulations where a method did not succeed, we would be mixing factors that do not reflect performance under conditions of success.
You must not lose sight of the main objective when interpreting the results.
I see you anxious to dissent without foundation; everyone knowledgeable who reads you sees it. This is the most famous forum in the world about Bitcoin. Do not ridicule yourself by changing the subject every three publications, searching for a corner to hold onto.