Constructing Keltner Channels for Universe Selection within class SelectionData()

Hi Vncne,

Try replacing

with

If that doesn't fix it, please attach a backtest which demonstrates the issue so I can further assist.

Best,
Derek Melchin

Hi Vncne,

I encountered the same issue with my algorithm. I found that the issue came from modularisation for me. When I incorporated the universe selection into my main.py file, the algorithm ran without an issue. My algorithm is similar to yours, only it uses bollinger bands instead of EMA's. My code is attached, I hope this helps, it stumped me for days!

Hi Nicholas,

1. I'm curious why in your why in your def update, you only have the bollingerbands and not the keltner channels?

2. Also I don't see your OnData section.  I see you append your selected stocks to your selected list. I imagine you handle order something like this:

    def OnSecuritiesChanged(self, changes):
        for security in changes.AddedSecurities:
            self.selected.extend([security.Symbol])
            
    def OnData(self, data):
        for symbol in self.selected:
            if data.ContainsKey(symbol) and data[symbol] is not None:
                self.SetHoldings(symbol, 0.1)
        self.selected = [symbol for symbol in self.selected if not self.Portfolio[symbol].Invested]

What I'm trying to figure now is after we do all this is, do I need to also remove the securities after sending the order and when they are removed from the universe should I clear their data?  Basically I have an indicator condition I'm looking for and when that happens I want to send an order.  Once the condition is met to sell that stock I want to buy it again if the indicator condition presents itself again.

3.  Do you have more example code on how you handle the orders with this algorithim?

Hi @John J, 

You could maintain a list of your universe in OnSecuritiesChanged like this and send an order for each symbol in OnData (I created self.universe and self.removed in def Initialize)

def OnSecuritiesChanged(self, changes):
	
	for security in changes.RemovedSecurities:
		if security in self.universe:
			self.universe.remove(security)
			
		if security not in self.removed:
			self.removed.append(security)
			
	for security in changes.AddedSecurities:
		if security not in self.universe:
			self.universe.append(security)
		
		if security in self.removed:
			self.removed.remove(security)
		
		
def OnData(self, data):

	for security in self.universe:
		if not security.Invested:
			self.SetHoldings(security.Symbol, 1/len(self.universe))
		
	for security in self.removed:
		if security.Invested:
			self.Liquidate(security.Symbol)
	

Not the most elegant, but I guess it gets the job done, unless you were thinking of something else? 

Also, is your indicator condition an indicator that has to be attached to each symbol in the universe?