Hello fellows.
Since b37 I noted that Evolver sends strategies with a very small positions count to Apex. I am not sure it's a bug, because it strictly respects Sharpe Ratio, my favorite fitness function, but the way it does it is impractical due to there are one or two highly profitable trades in years. Before b37 Evolver was sending to Apex strategies that respected the fitness function (Sharpe) but with an enough number of trades.
¿It is just me or this is happening to someone else?
¿Which indicator do you evolve for?
¿Should I create customize the filter set?
¿Should I uncheck Apex collection?
Thank you in advance
Since b37 I noted that Evolver sends strategies with a very small positions count to Apex. I am not sure it's a bug, because it strictly respects Sharpe Ratio, my favorite fitness function, but the way it does it is impractical due to there are one or two highly profitable trades in years. Before b37 Evolver was sending to Apex strategies that respected the fitness function (Sharpe) but with an enough number of trades.
¿It is just me or this is happening to someone else?
¿Which indicator do you evolve for?
¿Should I create customize the filter set?
¿Should I uncheck Apex collection?
Thank you in advance
Rename
Why not change your Evolver filter to exclude these strategies?
I will try this. It is just that it did not happened before, even though I am using the same fitness function (Sharpe) and the same filter. The only thing I changed was not in the Evolver preferences but in the metrics columns, I removed APR, but it should not affect the Evolver (or yes?).
Nothing has changed in those builds which should have impacted the Evolver in this way. It could just be happenstance.
Maybe it is due to I removed APR from my metrics columns?
That should not effect things.
Hello again!
After running the Evolver tens of times I can confirm something is messed up at least with my local something.
After hours of running with all sort of common symbols such as QQQ, or datasets such as DOW30 and many more, the strategies showcased by the evolver produced a very few trades, or just one in 10 years. It did not happened some time ago, and I am still using the same fitness target of Highest Sharpe Ratio. I have tried with all sort of fitness targets, like APR (yes I putted it back) and even number of trades, same, same problem.
Evolver is a very useful component of WL. Your help here is much appreciated.
These are my Evolver Preferences, classic, almost default.
This is my Default Evolver Filter, Number of Trades is not set as I remember it wasn't set when it worked fine, months ago.
This is an example of what is happening, I am using SQQQ now but it happens with any symbol or dataset, I unchecked Apex but if I check it, the problem persists.
Again: your help is much needed here as Evolver is a great tool and very useful.
Thank you in advance.
After running the Evolver tens of times I can confirm something is messed up at least with my local something.
After hours of running with all sort of common symbols such as QQQ, or datasets such as DOW30 and many more, the strategies showcased by the evolver produced a very few trades, or just one in 10 years. It did not happened some time ago, and I am still using the same fitness target of Highest Sharpe Ratio. I have tried with all sort of fitness targets, like APR (yes I putted it back) and even number of trades, same, same problem.
Evolver is a very useful component of WL. Your help here is much appreciated.
These are my Evolver Preferences, classic, almost default.
This is my Default Evolver Filter, Number of Trades is not set as I remember it wasn't set when it worked fine, months ago.
This is an example of what is happening, I am using SQQQ now but it happens with any symbol or dataset, I unchecked Apex but if I check it, the problem persists.
Again: your help is much needed here as Evolver is a great tool and very useful.
Thank you in advance.
Click on the strategies and look at the rules, i.e., look what it's doing.
If you're only trading SQQQ, there's probably no strategy that you can imagine that would beat just shorting SQQQ once from just about any point in time.
If you're only trading SQQQ, there's probably no strategy that you can imagine that would beat just shorting SQQQ once from just about any point in time.
If you want the results to contain strategies with more than one position add that condition to your Evolver Filter.
SQQQ is just an example. The problem is witch whatever I use.
I will experiment adding number of trades as a filter, and I will also see how it performs in my spare computer so I can compare and find what is messed up in my main one, as it was working wonderfully some time ago.
I will let you know.
I will experiment adding number of trades as a filter, and I will also see how it performs in my spare computer so I can compare and find what is messed up in my main one, as it was working wonderfully some time ago.
I will let you know.
Hello team.
I compared both WL running on my main machine with the one on my spare computer and I confirmed there is a problem on my main installation.
On my spare computer Evolver strategies bring lots of positions, and on my main computer they only bring one or a few, same symbol, same range.
Default Evolver Filter is the same and the original by WL, unmodified, genes are the same.
What could possibly be the difference?
I am really needing clues to solve this. I need to make it work, as my spare computer is very old and slow, not suitable for the task.
If I re-install WL from the scratch I would be risking the profitable strategies I made with so much hard work, and many of them use neural networks, so I prefer to avoid this.
What could I do to? What could be messed up in my main WL instance if both filter and Evolver preferences are identical?
I compared both WL running on my main machine with the one on my spare computer and I confirmed there is a problem on my main installation.
On my spare computer Evolver strategies bring lots of positions, and on my main computer they only bring one or a few, same symbol, same range.
Default Evolver Filter is the same and the original by WL, unmodified, genes are the same.
What could possibly be the difference?
I am really needing clues to solve this. I need to make it work, as my spare computer is very old and slow, not suitable for the task.
If I re-install WL from the scratch I would be risking the profitable strategies I made with so much hard work, and many of them use neural networks, so I prefer to avoid this.
What could I do to? What could be messed up in my main WL instance if both filter and Evolver preferences are identical?
Compare all of the backtest preferences
There were some differences, mainly the cash interest, included in my main WL instance, and the flat $1 commission.
Should this influence the number of open positions from Evolver strategies?
A good new is that after adding the number of open positions as a new filter, despite it is not existent in the default Evolver filter, it started to bring a good output!
Thank you!!
Should this influence the number of open positions from Evolver strategies?
A good new is that after adding the number of open positions as a new filter, despite it is not existent in the default Evolver filter, it started to bring a good output!
Thank you!!
Yes of course those settings will influence what the Evolver outputs.
Your Response
Post
Edit Post
Login is required