Examples of how to use OptionStrategyPositionGroupBuyingPowerModel?

When I attempt to use it before each trade, like this: 

self.equity.SetBuyingPowerModel(OptionStrategyPositionGroupBuyingPowerModel(strategy))
self.Buy(strategy, qtyToPurchase)

This is the error i am currently seeing:

NotImplementedException : TInterface must be fully implemented. 

Please implement these missing methods on <class 'QuantConnect.Securities.Option.OptionStrategyPositionGroupBuyingPowerModel'>: 

GetLeverage, SetLeverage, GetMaximumOrderQuantityForTargetBuyingPower, GetMaximumOrderQuantityForDeltaBuyingPower, GetReservedBuyingPowerForPosition, GetBuyingPower
   
   at QuantConnect.Python.PythonWrapper.ValidateImplementationOf[TInterface](PyObject model) in /LeanCloud/CI.Builder/bin/Debug/src/QuantConnect/Lean/Common/Python/PythonWrapper.cs:line 13

   at QuantConnect.Python.BuyingPowerModelPythonWrapper..ctor(PyObject model) in /LeanCloud/CI.Builder/bin/Debug/src/QuantConnect/Lean/Common/Python/BuyingPowerModelPythonWrapper.cs:line -9

   at QuantConnect.Securities.Security.SetBuyingPowerModel(PyObject pyObject) in /LeanCloud/CI.Builder/bin/Debug/src/QuantConnect/Lean/Common/Securities/Security.cs:line 715

Hi .ekz.,

Please attach a backtest that throws the insufficient buying power error so we can assist further. The margin for spreads should be correct.

Best,
Derek Melchin

Thanks for the response, but this algo isnt one i can share publicly, 

That said,  this is a known issue. See this post that describes the issue accurately, with a backtest that reproduces the issue: 

Are you suggesting that I do not need to use the custom buying power model : OptionStrategyPositionGroupBuyingPowerModel, and that it should ‘just work’ ?

Also sharing this, as I believe it is related. 

Hi .ekz.,

We calculated the changes in margin and cash using this page as a guide and got

Buy to open 1 SPY 321 Call for 3.86
   3.86 * 1 * 100 = 386
Sell to open 1 SPY 320.5 Call for 4.19
   4.19 * 1 * 100 = 419
Margin requirement = (321 - 320.5) * 1 * 100 = 50
Net Credit = 419 - 386 = 33
BP effect  = -50 + 33 = -17

The logs of the attached backtest show

Quit(): Contract_1 strike: 320.5; Contract_2 strike: 321.0 Margin used: 50.0 Margin Remaining: 9944.0; Cash: 10031.0; TPV: 9994.0

that the margin used matches the calculation and the total portfolio value is

Value of 1 SPY 321 Call for 3.84
   3.84 * 1 * 100 = 384
Value 1 SPY 320.5 Call for 4.21
   4.21 * 1 * 100 = 421
Value = 384 - 421 = 37

TPV = Cash + Credit - Value - Fee:

9994.0 = 10000 + 50 - 37 - 2 = 9994.0

and the margin remaining is TPV - margin used:

9944.0 = 9994.0 - 50

The insufficient buying power error in this thread is likely caused by how the quantity is calculated.

max_risk = self.df_calls.at[uppercall_ind,'strike'] - self.df_calls.at[lowercall_ind,'strike']
max_risk_contract = max_risk * 100
max_investment = math.trunc(self.investment_limit / max_risk_contract)

Best,
Derek Melchin

Aha. Got it. I appreciate the thorough investigation and detailed response. 

I'll go take a look at my code again, and revisit how i calculate the # of spreads I can afford. 

Thanks a lot.

Trying this out, with a bull put spread on SPY. 

I'm calculating the affordable quantity, based on buying power effect. The math looks right, but apparently I'm trying to buy too much because i am still getting the error. Any thoughts?

Here's my debug log below, and the backtest attached

spreadWidth 	  = 321.0 - 320.5 = 0.5
marginReqPerSpead = 0.5 * 1 * 100

spreadCost 		  = Short BidPrice - Long AskPrice
spreadCost 		  = 1.82 - 1.71 = 0.11
creditReceived 	  = 0.11 * 100 = 11.0

requiredOrderfees = 2
buyingPowerEffect = 11.0 - (50.0 - 2) = -37.0

marginRemaining   = 10000.0
qtyWeCanPurchase  = math.floor(abs(10000.0 / -37.0)) = 270

Order Error: id: 1, Insufficient buying power to complete order (Value:-49545), Reason: Id: 1, Initial Margin: -49612.5, Free Margin: 10000
Order Error: id: 2, Insufficient buying power to complete order (Value:45765), Reason: Id: 2, Initial Margin: 45832.5, Free Margin: 10000

See attached backtest. 

Hi .ekz.,

The option margin model is not currently equipped to determine the maximum position size for orders with OptionStrategies, but we have an open GitHub Issue to have the problem resolved. Subscribe to our progress here:Combo buying power model  #2709

For the time being, we can use the following snippet to determine the quantity of each contract we can afford without receiving the margin error:

def GetAffordableQty(self, theShortContract, theLongContract):
   quantities = []
   
   marginRemaining   = self.Portfolio.MarginRemaining
   for contract in [theShortContract, theLongContract]:
       security = self.Securities[contract.Symbol]
       marginParameters = InitialMarginParameters(security, 1) 
       marginRequirement = security.BuyingPowerModel.GetInitialMarginRequirement(marginParameters).Value
       affordable_quantity = int(marginRemaining / abs(marginRequirement))
       quantities.append(affordable_quantity)
           
   return min(quantities)

See the attached backtest for reference.

Best,Derek Melchin

Thanks Derek Melchin . I admit it's good to know that I was not mistaken and my concerns are valid.

With that said, I have two important questions.

1. Will this workaround calculation give us the true affordable Qty of spreads we can open? Or is it just a work around to avoid the error? Another way to ask this: when the GitHub issue is resolved and we can do the math 'correctly', will the new calculation give me the same Qty?
2. What is the timeline to fix this issue? I see it has been open for a few years. With more folks trading options, I'd imagine this might warrant some escalation / re-prioritization

Hi .ekz.,

1. The workaround calculation is just to maximize the position size without hitting the margin error that's currently thrown. Once the issue is resolved and we can fill the true maximum position size, we should be able to use the formula above that results in a quantity of 270 contracts. We will test it once the patch is merged.
2. We've opened up a newer GitHub Issue to give it a higher priority  (Can't Maximize Position Size of OptionStrategies  #5693) and closed the old one. The engineering team will address it as soon as possible.

Best,
Derek Melchin

Ah, thats what I thought, thanks for confirmation Derek Melchin.   

Follow up qq: if i were to live trade, would the engine use the broker's margin model / calculation?

Hi .ekz.,
QuantConnect's margin model is applied in live mode to avoid placing bad orders. To use the brokerage margin model, we have to "disable" the margin model.
In this algorithm just this part to turn off the HasSufficientBuyingPowerForOrder check:
 

class CustomBuyingPowerModel(BuyingPowerModel):
   def HasSufficientBuyingPowerForOrder(self, parameters):
     return HasSufficientBuyingPowerForOrderResult(True)

Best,
Varad Kabade

Got it. Thanks Varad kabade