How to Consistently Extract Revenue Data from Income Statements

[mvinoba]

Hi,

I'm trying to extract Revenue data from the income statements of various companies. However, I've noticed that different companies label Revenue differently in their 10-K filings. For example:

• Nvidia and Microsoft use the term "Revenue."
• Apple uses "Net Sales."

Is there a standardized GAAP code or a consistent approach I can use to extract Revenue data across companies? Ideally, I'm looking for a method that handles these naming variations reliably.

[Lanikeha]

Hi. I am not an expert on edgartools but have faced the same problem. Currently I use list consisting of possible options regards GAAP "concept" column. For revenues that is:

gaap_revenue_list = ['us-gaap_RevenueFromContractWithCustomerExcludingAssessedTax', 
                     'us-gaap_RevenueFromContractWithCustomerIncludingAssessedTax', 
                     'us-gaap_Revenues']

These are options based on 20-30 companies, maybe there are more. I update them as I discover new ones.

[mvinoba]

Thanks! That could work I guess. It's a shame it's not built-in into the .get_income_statement() method. Do you mind sharing the mapping lists? Or are they stored anywhere publicly anywhere? I think it would be very useful if we could also add such info to the edgartools documentation.

[Lanikeha]

I have just started so treat the lists as very incomplete. So far:

gaap_general_administrative_expenses_list = ['us-gaap_GeneralAndAdministrativeExpense',
                                             'us-gaap_SellingGeneralAndAdministrativeExpense']

gaap_net_income_list = ['us-gaap_ProfitLoss', 
                        'us-gaap_NetIncomeLoss']

[mvinoba]

Thanks! I'm marking your reply as the answer! That said, if @dgunning has a more complete or standardized approach, he's welcome to share it as well.

[dgunning]

Standardized statements and having the ability to compare companies seems to be a popular topic.

The current implementation of financials is geared around getting financials 'as-reported' but standardized statements are also supported. For instance you can call financials.get_income_statement(standard=True) and get standardized statements.

The current limitation is that it currently uses the most likely fact that matches a category e.g. Revenus is us-gaap_RevenueFromContractWithCustomerExcludingAssessedTax.

The code for standardization is here https://github.com/dgunning/edgartools/blob/main/edgar/xbrl/statements.py#L67. I think it can be changed to have multiple fallback facts from a list, but that list needs to be learnt from actual financials of many companies. If someone already has a list then that would be welcome.

I might spend some time of the weekend rerunning a script I had for approx 85 public companies and build the list.

[windy-day-ca]

I noticed that for Apple, Inc. (AAPL), the revenue returned by that call ( financials.get_income_statement(standard=True)) seems to work inconsistently. For example, for 2021 it correctly returns the correct total revenue ($365.817M) but for 2022 it reports only $316.199M (which was only the company's product revenue, excluding service revenues) and for 2020 it gives only $53.768M (which was the company's service revenues. leaving out product revenues).

I am very much a python beginner, and your tool looks very promising!