Anchored backtesting

In QuantConnect, you can use the History method to access historical data. This method returns a pandas DataFrame indexed by time and symbol. You can use this data to train your model. Here is an example of how you can use the History method:

# Request historical data
history = self.History(self.symbols, 30, Resolution.Daily)

# Loop over the data
for symbol, data in history.groupby(level=0):
    # Use data for model training
    pass

In this example, self.symbols is a list of symbols you are interested in, 30 is the number of days of historical data you want to retrieve, and Resolution.Daily specifies that you want daily data.

Remember that in QuantConnect, the algorithm can only access information up to the current time in the backtest, similar to live trading. This is to prevent look-ahead bias, which is when you use information from the future to make decisions in the past.

For more information on how to use historical data in QuantConnect, you can refer to the Understanding Time page in the documentation.