QuantConnect Community

Hey Mitch Christow,

I changed self.Time with datetime.now() and deleted "self" from EquitiesFundamentalSelectionModel(self). I attached the backtest, hope it works.

HI Nico,

Thank you so much for taking a crack at this. I had done that before as well, but the date information is not populated at that point (which is why there are no purchases taking place. I was trying to refactor the code from a fellow Quantconnector. Louis had written this code, and I tried to move the Universe selection into its own class. I had ended up at point with the same code that you are shoeing, but there should be transactions. Here is the thread with the original code and my refactored code.

I don't know why the DateTime object is not propagating properly from QCAlgorithm, but I am hoping that smarter people than me can figure out what coding mistake I've made. If you look at Loius original code (cloned below) my refactored code should yield the same result. If you have any ideas what I am doing wrong I would be eternally grateful.

Cheers,
Mitch

Hi, Mitch, I have fixed time bug here (it was also wrong in fine filter condition). However, this fine filter returns nothing. It seems there are no securities that match given sectors filter.

Hey Yuri Lopukhov,

nice work! If I change the filtered_fine so that it is correct:

[x.Symbol for x in cast(Iterable[FineFundamental], fine)

                         if x.SecurityReference.IPODate + timedelta(5*365) < self.algo.Time and 

                         x.AssetClassification.MorningstarSectorCode in sectors and

                         x.OperationRatios.ROE.Value > 0 and

                         x.OperationRatios.NetMargin.Value > 0 and

                         x.ValuationRatios.PERatio > 0]

I receive a runtime error:

Runtime Error: Trying to dynamically access a method that does not exist throws a TypeError exception. To prevent the exception, ensure each parameter type matches those required by the 'QuantConnect.Data.UniverseSelection.SecurityChanges'>) method. Please checkout the API documentation.

So somewhere there might still be an error.

Yeah I am very perplexed about this one. It should be trivial to refactor the code from the cloned algorithm from Louis. I know that the refactoring worked once the result is the same as Louis' code. But somehow this is not working. In Louis' version, the code works since the main.py file inherits QCAlgorithm, which includes the Time component that is needed for back testing. 

When one moves the three methods from main.py into their own universe selection class, then you lose access to Time. I tried passing in QCAlgorithm, which I am obviously doing wrong since that's the error that I got stuck on. In general it seems that it should be a whole lot simpler to access the methods from QCAlgorithm throughout the framework. 

I really appreciate your help in trying to fix this.

Cheers,
Mitch 

I have been able to clean up the refactored code to the point where the universe selection is returning the right data as part of the fine selection filters. However I am now stuck at an error which I believe is located in my alphamodel class. I am getting the error:

Trying to dynamically access a method that does not exist throws a TypeError exception. To prevent the exception, ensure each parameter type matches those required by the 'QuantConnect.Data.UniverseSelection.SecurityChanges'>) method. Please checkout the API documentation.

Which so far I have been unable to debug. Any helpful ideas would be greatly appreciated. 

Hi, Mitch Christow can you attach the updated code? Since its not possible to attach backtest with errors, just add self.Quit() somewhere before the error to have the backtest complete without error.