Annual Protfolio Selection Notebook - stocks_price shift 90 days

Installations & Python Packages
1

Hi, 



I am trying to understand the Annual Portfolio Selection.ipynb notebook (see a link). 



From the script below you can see that the stocks_prices have 90 days shift and are grouped by year

# Create yearly groups of the dataset on shifted data to avoid forward bias
year_groups = stocks_price.shift(90).groupby([stocks_price.index.year])

Then it is passed to the annual_unverse_selector function, where stocks_prices has the 90 shift, but stocks_volume does not.  

 

def annual_universe_selector(stocks_price, stocks_volume):
   
    # Choosing from the volumes dataframe the index of the group
    stocks_volume = stocks_volume.loc[stocks_price.index]

    
    # Filter the top 100 stocks
    filtered_stocks = (stocks_price[:90] * stocks_volume[:90]
                       ).mean().sort_values(ascending=False).index[:100]


    # Return the filtered stocks for as a Series
    return pd.Series(filtered_stocks)

Shouldn't stocks_volume need to be shifted 90 days also? 



 

2

Hi Boris,



Thanks for pointing this out! Since the stocks_price is shifted by 90 days to avoid forward bias, the stocks_volume also needs to be shifted in order to calculate Average daily turnover. We will make the required changes. 



Thanks,

Akshay

3

Hi Boris,



Thanks for pointing this out! We have made the enhancement and the notebook is now updated.



Thanks,

Akshay