Why the following meta strategy have very different # of trades and results for different runs? There's no NSF positions and capital is 50/50 split and separate for sub strategies. Shouldn't the results for different runs the same (or very similar)?
Rename
Different date ranges for the backtests?
Something's different. Compare the Positions list for clues.
Edit - just did it myself, and I see a difference too when running successive or Cloned runs. We'll look into it.
Something's different. Compare the Positions list for clues.
Edit - just did it myself, and I see a difference too when running successive or Cloned runs. We'll look into it.
You’ll see the nsf in the metrics report of the MetaStrategy under the Backtest Results tab, the values you underlined are for the individual component Strategies, not the combined results.
Different results simply because if both symbols trigger in one day one of them will be chosen at random since there’s no transaction weight.
Different results simply because if both symbols trigger in one day one of them will be chosen at random since there’s no transaction weight.
@Glitch: I thought that my configuration in the screenshot was for two isolated sub-strategies, each works on half of the capital. So if both symbols trigger both should be able to execute.
Did I miss anything here?
Did I miss anything here?
I'd still question the NSFs in the Meta run too. They're there, but these are limit orders. There are only 2 positions possible each assigned 50% of the equity.
I can't think of a reason why an NSF should occur.
They don't occur frequently and you have to load 6 years of data to see it happen even once. But doing repeated runs for the last 6 years, you'll eventually see this - no trades on TQQQ after 11/22/2021 -
And there is Cash -
I can't think of a reason why an NSF should occur.
They don't occur frequently and you have to load 6 years of data to see it happen even once. But doing repeated runs for the last 6 years, you'll eventually see this - no trades on TQQQ after 11/22/2021 -
And there is Cash -
As a quick workaround apply a little margin to the base Strategy, like 1.1 to 1.
Also, still trying to reproduce this locally, maybe I have to change date ranges? When I run the MS again and again the results are the same, and when I clone it and run the clone the results are always the same.
Gltich, at least 6 years backtest. 5 years or less are always the same for me too.
The scenario I showed above occurred after repeated runs with 6 years. Just keep clicking Run until it happens.
The scenario I showed above occurred after repeated runs with 6 years. Just keep clicking Run until it happens.
For Metastrategy where can one set range of data to use?
Strategy Settings tab.
I tried to recreate the problem with 7 years of data. The result was always the same. I must have run the backtest about 50 times in a row.
I noticed the problem happens when the margin factor is set to 1. With a slightly larger margin factor it goes away.
I'm not sure how, but this setting looks like it's the culprit. If I disable Use UST Yield, the problem doesn't occur.
(Interest on Cash and Margin Loan Rate are set at 0.)
(Interest on Cash and Margin Loan Rate are set at 0.)
Your Response
Post
Edit Post
Login is required