-A python library for pulling financial data from Morningstar and Yahoo.
-Includes over 100 functions for income statement, balance sheet, and cashflow items, as well as various valuation metrics.
-Data is fetched from Morningstar and Yahoo Finance.
-FREQUENCY refers to Annual (A), or Quarterly (Q) for financial metrics, TIME is how many columns (Years) the script scrapes from the downloaded excel file (1 - 5) For a pure DCF Implementation the default TIME should be 5 representing the last 5 years of financial data
- Credit to VincentRaia for the skeleton code of pulling the financial data from Morningstar and Yahoo
-urllib2
-pandas
-Quandl
===============
THIS IS ENTIRELY DEPENDANT ON MORNGINGSTAR; IF THEY DECIDE TO CHANGE THEIR URL ADDRESSES OR SHUTDOWN THIS PROGRAM BECOMES USELESS.
FREQUENCY refers to Annual (A), or Quarterly (Q) for financial metrics, TIME is how many columns (Years) the script scrapes from the downloaded excel file (1 - 5) For a pure DCF Implementation the default TIME should be 5 representing the last 5 years of financial data
************************************** TO DO ******************************
Still need to fix all the functions on the function page, follow format of functions.revenue, or functions.cost_of_goods, functions.income_before_taxed, etc.
Still need various QA implementations, and more efficient and cleaner code to grab all the data. Maybe download the income statement, balance sheet, and cashflow sheet ONCE and put them into three separate data frames. From there you can parse through in order to find the relevant metrics for each functions.
Implement different functions for overall DCF process and segment the parts of DCF process into smaller functions such as the implementation of the Income statement, Balance sheet, and Cashflow, add functionality for other Metrics as well.
Also need to figure out a way to deal with modularity and flexibility. By these I mean what happens when the rows don't exactly match the ones in your array. What happens when the numbers don't add up or a stock doesn't label their data the exact same as you have it. This will take a lot of trial and error, try and except blocks. You also thought about using either the row #, or the attribute name as it appears in the excel file as keys for the dictionary. If you pair the attribute name and then continue with the low level approach of working down with the excel file, I think that it would absolutely ensure accuracy when placing the data, even if the row and column NUMBERS change.
Maybe implement machine learning functions in order to make educated guesses on assumptions
You for sure can pull data for the initial global assumptions though
Modularity is HUGE
Safe guards to check and see if the stock's net income, free cash flow, etc. match up with the ones found in the csv files and the company's reported numbers.
Make sure to check and see if the data is reported in millions or thousands; if thousands then just truncate the nums