Portfolio Allocation size in a Dynamic Universe

Hi Patryski,

You are very close to the answer already so you're learning fast. In your code:

self.portfolioTargets = [PortfolioTarget(symbol, 1/(len(self.activeStocks))
                        for symbol in self.activeStocks]

you allocate a % of the portfolio to each symbol. We do not have your complete code but I am assuming that if you change that code for this:

self.portfolioTargets = [PortfolioTarget(symbol, 0.9/(len(self.activeStocks))
                        for symbol in self.activeStocks]

you will have 10% of your portfolio always liquid for future buys. You can use 0.8, 0.7, as you wish. You will most likely have to add some code so you can use that spare cash if you spot an opportunity. At a high level, you would need to add those new assets to self.activeStocks which you probably do when you want them to become active and your SetHoldings() call buys them.

Fred

Hi Fred Painchaud,

thank you for your input! Yes, I have done that before, but the issue is exactly that I dont know how to use the then remaining buying power or what kind of code is needed to use the remaining buying power during the next rebalancing.

Patryski

Hi Patryski,

I'd say that if you can share your complete code, it would be easier to help you out with that.

Fred

Fred Painchaud 

Okay, I hope this helps. I attached a backtest with the a variation of the code.

Ok. Looking at that right now.

Re-reading your original post, what you want to do is if your activeStocks (your Universe basically) contains say only 1 asset, you do not want to put 100% of your portfolio to that asset, you would like to enter say only 30% of your portfolio value. So that when new assets are added to your Universe, you already have room to buy them, and SetHoldings will not have to liquidate part of your previous position before buying new ones. Is that it?

If so, I am not currenly able to help you as I still don't know those details of the platform. What you need to do for sure is to track how many % of your portfolio is currently invested. That may or may not already be accessible somewhere, that's one thing I don't know right now. Say it is 30%. Then, if you have enough new assets in the Universe (enough is for you to decide, what is enough, 2? 3? 4? etc), and you do not want to touch the 30% already invested, you need to take 70% and assign it to the new assets, so say there are 3 new assets, that would be .7/3 for each of them. Really sorry I don't know what you don't know here…

A quick hack which might work is:

def OnSecuritiesChanged(self, changes):
        for x in changes.RemovedSecurities:
            self.Liquidate(x.Symbol)
            if x.Symbol in self.activeStocks:
                self.activeStocks.remove(x.Symbol)
                
        for x in changes.AddedSecurities:
            self.activeStocks.add(x.Symbol)
      
		# 1 here is your threshold for "too few assets"
		# 0.5 is your max % threshold to invest when there are too few assets (here 0.5 works because it makes sense with 1)
		# but if your first threshold is say 2, 0.5 twice would still be 100%, so you would need to put something like 0.25...
		# so then you could use code like this:
		# percent = (1/len(self.activeStocks)) if len(self.activeStocks) > 3 else (0.5/len(self.activeStocks))
		
		percent = (1/len(self.activeStocks)) if len(self.activeStocks) > 1 else 0.5
      


        self.portfolioTargets = [PortfolioTarget(symbol, percent)
                        for symbol in self.activeStocks]

So yeah, that could do it. But that being said, I am wondering why you say that “it is obviously not optimal to have 100% allocated to for example 1 positions.”. If that asset is the only one that meets all your criteria, it should be optimal. But maybe it is because then, you think it exposes you too much / too much risk?

Anyway, it's none of my business. Hope the code above and comments help to some extent.

Fred

Thank you Fred, the code you provided seems logical. 

Yes, it is a matter of risk.