How do you run backtests on specific symbols for specific dates?

Hi Jared

We can download and save the spreadsheet as a variable first. Assuming your spreadsheet is a .csv file, with columns of "Date" and "Stock_to_be_traded", each row as a data point:

# sample CSV
Date,Stock_to_be_traded
2021-01-02,SPY.QQQ.AAPL
2021-01-03,JPM
2021-01-04,AAPL.MSFT

from io import StringIO

def Initialize(self):
    stock_list = self.Download("https://www.dropbox.com/xxxx/xxxx.csv")
    self.stock_list = pd.read_csv(StringIO(stock_list), index_col="Date")
    self.stock_list.index = pd.to_datetime(self.stock_list.index)

    self.SetUniverseSelection(
    	ScheduledUniverseSelectionModel(
    		self.DateRules.EveryDay(), 
    		self.TimeRules.At(8, 0), 
    		self.Selection
    	))

def Selection(self):
    today = self.Time.date()
    if today in self.stock_list.index:
        return [Symbol.Create(str(ticker), SecurityType.Equity, Market.USA) for ticker in self.stock_list.loc[today, "Stock_to_be_traded"].split(".")]
    return []

Note that in live trading, Initialize will not be called subsequently. So, in order to fetch any update on the spreadsheet, we will need to put the 3 self.Download lines at the beginning of the Selection method.

Best
Louis

Hi Louis,

Thanks for the response.  I am not worried about importing the spreadsheet.  My question is, how do I run back tests on specific dates and specific symbols at the same time.  I want to test if my code will work on a know set of dates and symbols.  Currently I can only do one at a time.  I don't know how to run for example:

2021-01-01 AMZN

2021-03-01 TSLA

2021-03-01 QQQ

Hi Jared

You may still use the above code snippet example, but using dataframe.groupby method to merge same date indexes:

# sample csv 
Date,Stock_to_be_traded
2021-01-03,JPM
2021-01-04,AAPL
2021-01-04,MSFT

def Initialize(self):
    stock_list = self.Download("https://www.dropbox.com/xxxx/xxxx.csv")
    self.stock_list = pd.read_csv(StringIO(stock_list), index_col="Date")
    self.stock_list.index = pd.to_datetime(self.stock_list.index)
    self.stock_list = self.stock_list.groupby(self.stock_list.index)["Stock_to_be_traded"].apply(lambda x: eval("['%s']" % "', '".join(x)))

Best
Louis

Thanks for the reply.  That isn't the part I have an issue with.

The tutorials use a structure like below.  You add equities and set a start date and end date.  I don't see any way to loop over any of this?  The only way I could see implementing it with a list of stocks and dates would be to add all symbols, add the max start date and end date, and return out of functions if the current date and symbol aren't in the list.  That sounds like a bad way of doing things.  The structure below doesn't look well suited for what I am trying to do.

class OpeningRangeBreakout(QCAlgorithm):
    
    bar1 = bar2 = bar3 = None 
    
    def Initialize(self):
        self.symbol = "OCGN"
        self.SetStartDate(2021, 10, 14)  
        self.SetEndDate(2021, 10, 14)  
        self.SetCash(10000)
        self.AddEquity(self.symbol, Resolution.Minute)
        self.Consolidate(self.symbol, timedelta(minutes=5), self.OnDataConsolidated)
        
    def OnData(self, data):
        
    def OnDataConsolidated(self, bar):
    def ClosePositions(self):

Hi Jared

The AddEquity method is no longer needed as we have a Universe Selection method. That will return us its selected securities as the ActiveSecurities keys. So you may directly refer to my code snippets in the previous replies.

Best
Louis