In MC 5.5 Beta 3 there is a bug with the Genetic Optimizer with "VERY MUCH" combinations. When having to many combinations, the Genetic Optimizer doesn´t optimize correctly anymore.
In MC 5.5 Beta 2 this bug was not present, just Beta 3 did introduce it.
As you can see on Screenshot 1, my strategy has a lot of possible combinations to calculate.
Now in Screenshot 2 I have switched to the Genetic Optimizer. Strangely it here suggests a Population Size of only "2". When I optimize with combinations, it does suggest something like "100000" there, but it seems when the combinations amount is getting to high,there is a kind of "buffer overflow" and it just suggests "2" for the population size.
Anyway, I ignore that and do change the population size to 100 and the max generations to 1000 and now run the Genetic Optimization. It runs through the optimizing process fine, but the outcome (Screenshot 3) doesn´t find any results, just "0" two times.
When reducing the number of combinations the Genetic Optimization works fine with exactly the same strategy (just less parameters optimized).
So I guess the Genetic Optimizer defintely has a kind of "buffer overflow" with to many combinations.
As I said, in Beta 2 this was working fine.
[Bug Report] MC 5.5 Beta 3, Genetic Optimizer "overload
Questions about MultiCharts and user contributed studies.
2 posts • Page 1 of 1
- tomanycombinations3.GIF (87.67 KiB) Viewed 93 times
- tomanycombinations2.GIF (88.55 KiB) Viewed 92 times
- tomanycombinations1.GIF (84.3 KiB) Viewed 103 times