Idea Streams #11 – Replicating a Goldman Sachs Index Using Quant Strategies

 

On January 20, 2021, the Financial Times reported on the spectacular performance of the Goldman Sachs Non-Profitable Technology Index. As we can see in the image below, this basket of companies remained relatively flat from 2014 to 2020. However, following the coronavirus crisis in March 2020, the index has spiked almost 400%.

We aspired to replicate the performance of this Goldman Sachs index, so in Idea Streams episode #11, we focus on creating a portfolio of the most unprofitable technology companies.

Our Process

To match the start date of the plot above, we first set the start date of our backtest to September 10, 2014. We then set up two universe selection methods to provide the algorithm with the securities we need. To ensure we only trade large companies and avoid penny stocks, we set our initial universe selection method to return securities that have a daily trading volume worth at least $10 million.

def CoarseFilter(self, coarse):
    return [c.Symbol for c in coarse if c.DollarVolume > 1e7]

Since we only want to invest in unprofitable technology companies, we narrow down our universe further to only include the top-50 most unprofitable technology companies. To classify companies into their respective sectors, we utilize the sector codes provided by MorningStar.

def FineFilter(self, fine):
    tech = [x for x in fine if x.AssetClassification.MorningstarSectorCode == MorningstarSectorCode.Technology]    
    unprofitable = [x for x in tech if x.FinancialStatements.IncomeStatement.NormalizedIncomeAsReported.ThreeMonths <= 0]    
    sorted_revenue = sorted(unprofitable, key=lambda f: f.FinancialStatements.IncomeStatement.TotalRevenue.OneMonth, reverse=True)    
    return [f.Symbol for f in sorted_revenue[:50]]

With our universe of securities selected, we constructed a portfolio that gives equal weight to all 50 securities. The portfolio is rebalanced at the beginning of each month. To ensure the trades are filled at the opening auction price, we set our data resolution to daily.

Results

Here is the value of the portfolio throughout our backtest:

Overall, it looks like we were able to replicate the Goldman Sachs Non-Profitable Technology Index fairly accurately. Not only do we see the 400% spike leading into 2021, but even the fluctuations in value leading up to that time closely match the performance of the index.

With all of that in mind, we hope you now have a better idea of how to approach creating a portfolio that is rebalancing monthly based on certain fundamental data points, like an asset’s sector classification and profitability. To view the source code of this strategy, clone the algorithm below.

Derek Melchin

By: Derek Melchin

01.06.2021