I've now been running the strategy evolver over night on my two computers and have been finding several iterations creating strategies that do not result in any positions when opened in the strategy runner and backtesting them individually. It does not happen every time but I have had several iterations have this happen so far.
Rename
Has anyone else been experiencing this issue?
To start with, have you made sure that the position size and data loading range are both sufficient to have a signal?
When you go from the evolver to open it in strategy menu, all settings are identical to that of the evolver preferences. I did not change any setting and it is not reporting a single position in some instances.
OK. It may be helpful to have more info and screenshots of such strategies. Currently there's no details.
I am not sure if what I am experiencing is the same as @arflouro82, but I will try to explain the problem below.
When the Strategy Evolver first came out in Build 44, I immediately started playing with it and getting quite good results:
- DataSet: basket of 4 continuous back-adjusted contracts: ES, NQ, YM and RTY;
- Scale ='Yearly';
- Range: 2017 to 2020;
- Evolve for Highest 'Sharpe Ratio'; and
- Use Filter Set 'Default Evolver Filter'.
With Build 44, this was resulting in some very nice results after some minutes; and even better, after leaving it overnight.
However, since around Build 45 or 46 (but I admit that it can be just a coincidence), I am getting basically nothing: not good nor bad strategies; it looks that all the found strategies have the 'Sharpe Ratio" column empty, probably because of not enough trades to calculate it.
I already checked the the historical data and everything seemed fine. Also went thought all the options that I could think of and also looked good.
I have no idea of what is the problem. Any suggestion to point me back to the right direction, please?
When the Strategy Evolver first came out in Build 44, I immediately started playing with it and getting quite good results:
- DataSet: basket of 4 continuous back-adjusted contracts: ES, NQ, YM and RTY;
- Scale ='Yearly';
- Range: 2017 to 2020;
- Evolve for Highest 'Sharpe Ratio'; and
- Use Filter Set 'Default Evolver Filter'.
With Build 44, this was resulting in some very nice results after some minutes; and even better, after leaving it overnight.
However, since around Build 45 or 46 (but I admit that it can be just a coincidence), I am getting basically nothing: not good nor bad strategies; it looks that all the found strategies have the 'Sharpe Ratio" column empty, probably because of not enough trades to calculate it.
I already checked the the historical data and everything seemed fine. Also went thought all the options that I could think of and also looked good.
I have no idea of what is the problem. Any suggestion to point me back to the right direction, please?
Did you disable the Filter Set or use a different Filter Set? It's still operating normally at my end.
I have the default filter activated. I also tried with different filters and no filters and still have the same problem.
It’s very weird. I don’t know what else to try…
It’s very weird. I don’t know what else to try…
alkimist, did you ever find any more details related to this? Do the evolved results still seem less impressive compared to the original release around Build 44?
It's purely anecdotal, but my experience seems to be the same. I'm tempted to try an old installation of WL7 and compare it with the WL8 evolver results now. Perhaps some change had unintended consequences? Or it could just be the datasets that I've been focused on lately.
It's purely anecdotal, but my experience seems to be the same. I'm tempted to try an old installation of WL7 and compare it with the WL8 evolver results now. Perhaps some change had unintended consequences? Or it could just be the datasets that I've been focused on lately.
Hi msdosey,
This was a long time ago but, if my memory doesn't fail me, it was human error (mine that is)... Somehow, I messed up with the configuration and that was not giving the expected results. I sorted it out (don't remember how) and that was it. Since then, I have been using the Evolver less, but I did find some interesting ideas to try back then,
This was a long time ago but, if my memory doesn't fail me, it was human error (mine that is)... Somehow, I messed up with the configuration and that was not giving the expected results. I sorted it out (don't remember how) and that was it. Since then, I have been using the Evolver less, but I did find some interesting ideas to try back then,
Your Response
Post
Edit Post
Login is required