QuantConnect Community

Colton Sellers Martin Molinero 

Maybe related to #5727 ?

Simon Koeman 

It would be helpful if you could attach a MWE, so the QC team can reproduce the issue. 

Hey Simon Koeman , if you can, would you mind sharing with us the details of the algorithm that triggered this bug? Arthur is correct that it is likely the change we just made to Lean on Friday but we are having trouble reproducing the issue. If you can't we totally understand, even a MRE like Arthur suggested that is stripped down version of your algorithm would be incredibly helpful.

Thanks again!

Colton

Just a follow up, we just merged a change that will provide more details to make it easier to reproduce for us in the event of future bug reports.

If you run your algorithm on the latest Master branch the details it will provide should be sufficient enough for me to reproduce the exact circumstances. Thanks again for the report.

Colton

I've since changed my algorithm and not had this particular issue again. I still don't know what exactly caused the issue but it probably was what you are saying. Next time if I have some time to spare I'll go though my code and try to make a MWE, but I don't think I can reproduce this error anymore.

No worries, thanks Simon.

As I mentioned above, now the error log should contain enough information for me to reproduce it without your code. At least in unit test form. The source is preferred but don't feel obliged to put in more work than you have time for 😊

Colton

Any update on this? 

I just got this error and here are the error logs:

Runtime Error: GetMaximumOrderQuantityForTargetBuyingPower failed to converge on the target margin: -14946.065217493140935875633023; the following information can be used to reproduce the issue. Total Portfolio Cash: 1013470.500; Leverage: 1; Order Fee: 139; Lot Size: 1; Per Unit Margin: 497.50000; Current Holdings: QuantConnect.Securities.Option.OptionHolding; Target Percentage: %-1.4962464078122763484654814700; Current Order Target Margin: 138229.03478250685906412436697; Current Order Margin: 138305.00000 in BuyingPowerModel.cs:line 467

I got a similar error regardless of if it was calls or puts I was using the emit insights and insight weighted portfolio construction to emit insights to sell

Hey Josh M !

Could you please provide an algorithm reproducing the issue? it will help us solve it ASAP, thanks!

Hey Martin, I'll post a block of code of a base algorithm as the next post in this series. (cant post backtest because it never completed!)

This isn't exactly my code, but a much reduced version just to replicate the error. Hope it helps! Let me know how I can further improve/assist <3

Best,

JM

from datetime import datetime

from collections import *

from System import *

from QuantConnect import *

from QuantConnect.Algorithm import *

from QuantConnect.Orders import *

from QuantConnect.Algorithm.Framework import *

from QuantConnect.Algorithm.Framework.Selection import *

from QuantConnect.Algorithm.Framework.Execution import *

from QuantConnect.Algorithm.Framework.Risk import *



from Execution.StandardDeviationExecutionModel import StandardDeviationExecutionModel

import decimal as d



class ETFSimple(QCAlgorithm):

    def Initialize(self):

        self.SetStartDate(2021,3,1) 

        self.SetCash(1000000) 

        self.SetPortfolioConstruction(InsightWeightingPortfolioConstructionModel()) #we want our portfolio to be made up of insights

        self.SetExecution(StandardDeviationExecutionModel(5, 1, Resolution.Daily))

        self.SetBrokerageModel(BrokerageName.AlphaStreams) #required brokerage model

        

        self.STRIKES = 30       # no of strikes around ATM => for uni selection

        self.MIN_EXPIRY = 75 # min num of days to expiration => for uni selection

        self.MAX_EXPIRY = 175 # max num of days to expiration => for uni selection

        self.SC_MIN_DELTA = d.Decimal(0.15)

        self.SC_MAX_DELTA = d.Decimal(0.35)

        self.MIN_PREMIUM = d.Decimal(0.05)

        

        

        equity = self.AddEquity("SPY", Resolution.Minute)

        option = self.AddOption("SPY", Resolution.Minute)

        self.optionSymbol = option.Symbol

        

        # set strike/expiry filter for this option chain

        option.SetFilter(-self.STRIKES, self.STRIKES, timedelta(self.MIN_EXPIRY), timedelta(self.MAX_EXPIRY))

        

        # for greeks and pricer (needs some warmup) - https://github.com/QuantConnect/Lean/blob/21cd972e99f70f007ce689bdaeeafe3cb4ea9c77/Common/Securities/Option/OptionPriceModels.cs#L81

        option.PriceModel = OptionPriceModels.CrankNicolsonFD()  # both European & American, automatically

        

        # this is needed for Greeks calcs 

        self.SetWarmUp(TimeSpan.FromDays(self.MAX_EXPIRY))





    def OnData(self, slice):

        

        if (self.IsWarmingUp): return

            

        self.shortCall(slice)





    def shortCall(self,slice):

        

        chain = slice.OptionChains.get(self.optionSymbol, None)

        

        if not chain:

            return

 

        # filter the put options contracts

        calls = [x for x in chain if x.Right == OptionRight.Call and abs(x.Greeks.Delta) > self.SC_MIN_DELTA and abs(x.Greeks.Delta) < self.SC_MAX_DELTA and x.BidPrice > self.MIN_PREMIUM] 



        # sorted the contracts according to the lowest price and the shortest expiry

        contracts = sorted(sorted(calls, key = lambda x: x.BidPrice), 

                                         key = lambda x: x.Expiry)



        if len(contracts) == 0: return



        scCall = contracts[0].Symbol

        self.EmitInsights(Insight.Price(scCall, timedelta(self.MIN_EXPIRY), InsightDirection.Down, None, None, None, .015))

        

Colton Sellers Martin Molinero 

Any feedback on what I can do to fix the error on my side via the minimal code I posted?

Thanks,

Josh

getting same error :

Hey everybody,

Thank you for your reports on this issue and also your patience. We worked out the problem in Lean and have created a PR for the fix here.

Josh M I reproduced this issue using your error message and also tested it with your supplied algorithm so thank you for all the info! We really appreciate it. They both succeed without any convergence issues on the branch.

We look to have this fix polished and merged tomorrow!

Thanks,

Colton

The PR has been merged!

Thanks again everyone!

Colton