Genetic Optimizer aggregated iteration count is off by 26x

[nathanfranke]

I'm using genetic optimizer on 38 instruments aggregated, 5 generations, 25 each generation. 38*5*25=4750. *26=123500. I don't know what 26 means.

This is a screenshot right before the optimization completed successfully.



Edit: I ran it again with 38 instruments, 10 generations, 35 each generation, and 13300 total, it displays 532000, which is 40x.

EDIT 2: BUG CAUSE

Okay, I think it is caused by Optimize Data Series. The total count is incorrectly multiplied by the number of permutations of Data Series. The optimization still ends early after 38*5*25.

Finishing example:

[QuantKey_Bruce]

There are X iterations per generation, where X is the number of permutations of the variables. For instance, if you optimize an indicator length from 1 to 10, that is 10 iterations. 38 * 5 * 25 in your example above doesn't include any factor for how many iterations there are per generation.

[nathanfranke]

There are X iterations per generation, where X is the number of permutations of the variables. For instance, if you optimize an indicator length from 1 to 10, that is 10 iterations. 38 * 5 * 25 in your example above doesn't include any factor for how many iterations there are per generation.

Genetic Optimizer uses random samples of each parameter, so adding more possibilities has no effect on the total count. The only exception is this bug, where the DataSeries permutations is being multiplied incorrectly. I have hundreds of millions if not billions of combinations (subset below), this would be impossible with the default optimizer.

[QuantKey_Bruce]

I see you later posted that this was caused by optimize data series so this may be representative of the data series possibilities. Possibly the data series optimization is treated differently than another parameter because the data series must be loaded in advance before the iterations are run. If that is the case, it might be handled as if each data series possibility is its own separate genetic optimization.