Compared with the original parameter group of strategy, the profit of strategy may increase or decrease after the first sequential optimization, which is easy to understand. However, it is not easy to understand that the profit of the second sequential optimization will decrease, because the current parameters have met the sequential optimization, and the worst profit should be the current one.
Because I can't see the internal logic of sequential optimization calculation, I can only guess. For example, if the parameter range of sequential optimization is + 100%, - 100%, then the current algorithm logic may search from 0 to 100% and then from 0 to - 100%, so that the parameter near 0 itself is ignored as a parameter satisfying sequential optimization. If this is the reason, the search logic may change from - 100% to 100%.
Attached are examples of my tests, which are actually easy to reproduce.