Coarse function takes top 10 holdings from an ETF

Hi Axist,
QQQ, like other ETFs, has a proprietary universe definition which might be difficult to reproduce using Coarse and Fine Fundamental data. So, we recommend using Manual Universe Selection of the top 10 holdings for securities in QQQ. Refer to the following code snippet.
 

def Initialize(self):    # Set requested data resolution
    self.UniverseSettings.Resolution = Resolution.Minute
    tickers = ["AAPL","MSFT","AMZN","TSLA","FB","GOOG","GOOGL","PYPL","NVDA","INTC"]
    symbols = [ Symbol.Create(x, SecurityType.Equity, Market.USA) for x in tickers ]
    self.AddUniverseSelection( ManualUniverseSelectionModel(symbols) )

Alternatively, we can fetch the latest top10 QQQ data from website using download method
Best,
Varad Kabade

Thanks for the response Varad.  The only issue I am seeing is that the download method wont allow for historical holdings of an ETF.  For example, with QQQ, their site doesn't unfortunately have old fact sheets or a history of what the top 10 holdings of the QQQ was in Q1 of 2018 for example.  

That being said, I take it my backtests will just take whatever the current top 10 holdings are, and run them against the time frame specified in back algorithm?  Which seems unfortunate. This may be a question for Google, but you wouldn't happen to know of any data sources that might provide this kind of information?

Hi Axist

We would not recommend copying an ETF from an online source as its constituents are ever-changing. Your backtest or live trade is not able to renew in time. However, if you insist, Yahoo Finance shall do the magic. You can search an ETF symbol and check on the "Holdings" tab (e.g. SPY) for the top 10 holdings. Please bear in mind this is not guaranteed to be up-to-date. Then you could create a .csv file to store ETF's constituents and maintain/update yourself in the form of

time,symbol,weight,rank
2021-7-24,AAPL,0.0590,1
2021-7-24,MFST,0.0560,2
2021-7-24,AMZN,0.0405,3
...

for example. Now, you can create a custom universe based on the .csv's data:

class CustomDataUniverseAlgorithm(QCAlgorithm):
    def initialize(self):
        # add a custom universe data source (defaults to usa-equity)
        self.AddUniverse(SPYComponents, "universe-SPYComponents", Resolution.Daily, self.spyComponents)    
        
    def spyComponents(self, data):
        return [ x.Symbol for x in data if x["Rank"] <= 10 ]
        
        
class SPYComponents(PythonData):
    def GetSource(self, config, date, isLiveMode):
        path = "your/PATH/of/xxx.csv"    # local date
        return SubscriptionDataSource(path, SubscriptionTransportMedium.RemoteFile)
        ### path = "url.of.csv/"         # online data
        ### return SubscriptionDataSource(path, SubscriptionTransportMedium.LocalFile)   
         
    def Reader(self, config, line, date, isLiveMode):
        # base on csv file example above
        csv = line.split(',')
        spy_components = SPYComponents()
        # the time and endtime should be 1 day apart as daily data
        spy_components.Time = datetime.strptime(csv[0], "%Y-%m-%d") - timedelta(days=1)
        spy_components.EndTime = spy_components.Time
        spy_components.Symbol = Symbol.Create(csv[1], SecurityType.Equity, Market.USA)
        spy_components["Weight"] = float(csv[2])
        spy_components["Rank"] = int(csv[3])
        return spy_components

You can study the details in this example.

Best
Louis Szeto