DislikedJust a thought here regarding any system that requires 99.9% accuracy:
You will never know the true accuracy of the system in practice. When using this extreme confidence level you must be certain what will happen if your accuracy rate is not right.
In the original example you have a 40% stop loss, and let's use 10,000 trials in our sample.
At 99.9% you would expect to have your stop loss hit 10 times. According to your model it would still be a profitable strategy.
But then consider maybe your model is wrong and you are only accurate 99.8%...Ignored