Creating our own Index Fund!

Sorry for some reason this isn't cloning properly. Please copy the code here:

class IndexInvesting(QCAlgorithm):

    def Initialize(self):
        self.SetStartDate(2012, 1, 1)  # Set Start Date
        self.SetCash(100000000)  # Set Strategy Cash
        
        self.SetBenchmark("SPY")
        
        self.UniverseSettings.Resolution = Resolution.Hour
        
        self.spy = self.AddEquity("SPY", Resolution.Hour).Symbol
        self.AddUniverse(self.Universe.ETF(self.spy, self.UniverseSettings, self.ETFConstituentsFilter))
        
        self.weights = {}
        
        self.Schedule.On(
            self.DateRules.WeekStart(self.spy),
            self.TimeRules.AfterMarketOpen(self.spy, 1),
            self.Rebalance)
    
    def ETFConstituentsFilter(self, constituents):
        self.weights = {c.Symbol: c.Weight for c in constituents}
        return list(self.weights.keys())
    
    def Rebalance(self):
        for symbol, weight in self.weights.items():
            if symbol in self.ActiveSecurities:
                self.SetHoldings(symbol, weight)  # Market cap weighted
                # self.SetHoldings(symbol, 1 / len(self.weights))  # Equally weighted
    
    def OnSecuritiesChanged(self, changes):
        for security in changes.RemovedSecurities:
            if security.Invested:
                self.Liquidate(security.Symbol, 'No longer in universe')
                if security.Symbol in self.weights.keys(): del self.weights[security.Symbol]

Brilliant! Thanks for sharing this. 

I recently had to do something similar, building a factor weighted portfolio using the historical constituents of the Nasdaq 100. It was a bit cumbersome as I had to buy the data, and load it manually via a CSV file on dropbox.  Using self.Universe.ETF will make life considerably easier. 

Will give it a try soon with NDX.  

Here is Ollie Hooper's original code with some lines I added to compare "own index fund" 
to the underlying ETF.

Cool stuff good work!

Hi QC Community

For those who are interested in this topic, we recommend further reading on Sparse Portfolios for High-Dimensional Financial Index Tracking (Benidis, Feng & Palomer, 2018). This paper demonstrates how to properly trimmed components to track index for similar results with a minimum number of positions, such that we can

• lower the capital/fee for creating our own index fund with similar performance
• explore arbitrage opportunity between the actual index fund
• to track only upside but not the downside

Best
Louis

Played with this a bit. Here’s a version that trades the top 10 securities in the QQQ ETF, based on Momentum. 

I’ve also externalized a few settings in case folks want to tinker with it, including: 

1. how many positions to hold
2. the lookback in days, 
3. the rebalancing period (monthly or weekly), and 
4. whether or not to honor the ETF’s weighting (or use equal weighting) for the stocks that are selected for holding.

I’m going to try some other performance metrics as well (e.g.: sharpe ratio, drawdown, etc), hence the extra ENUM code that's currently unused.