antgoldbloom/xbrl_parser

Parsing Raw XBRL to JSON

xbrl_parser's Issues

Replicating AMZN income statement for 2020-03-31

From Presentation XML
document date -> Statement -> metric -> order [DONE]
document date -> Statement -> metric -> link from [DONE]
document date -> Statement -> metric -> link to [DONE]

From Instance XML
document date - > Statement -> metric -> date -> value [DONE]
[QUESTION]: Am I handling YTD sensibly?

From Label XML
document date -> Statement -> metric -> label

From XSD
document date -> statement -> statement name

Add 20-F and 40-F filings

These are for foreign companies listed on a US exchange.

40-F is for Canadian companies. 20-F is for everything else.

Replicating all AMZN statements and disclosures for 2020-03-31

This is what I'm aiming to replicate
https://www.sec.gov/cgi-bin/viewer?action=view&cik=1018724&accession_number=0001018724-20-000010&xbrl_type=v#

Splice history for companies that change their CIK code

We lose history for companies that changes their CIK code (GOOG from 2015 Q3, ZG from 2015 Q1).

Handle amendments

This security has an amendment
https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&owner=exclude&start=0&count=100&CIK=VER&type=10-Q&dateb=None&output=atom

Can see the amendment as 10-Q/A. Behavior we want is to replace version with the amendment before the amendment.

Fill missing values by inferring change in xbrl tag even when there's no overlap

A good example is in ZG
us-gaap_costofrevenue___z_homesmember is missing for 2019-12-31 and 2018-12-31

However there is a 2019-12-31 and 2018-12-31 with the tag us-gaap_costofrevenue___z_homessegmentmember. This is the missing value and it wouldn't be hard to identify that.

Could use cosine similarlity on tags as part of the solution to address this issue.

Handle 10-KT filings

E.g. this company has a 10-KT
https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000007084&type=10-K&dateb=&owner=include&count=40&search_text=

It's a transitional 10-K for when companies are changing their year end filing date:
https://www.investopedia.com/terms/s/sec-form-10-kt.asp

OPEN missing Cost of Revenue

For December, September and June.

Make statement names more comparable across companies

There's no standard naming for statements and disclosures.

Balance sheets are referred to as CONDENSED CONSOLIDATED BALANCE SHEETS (ZM) or Consolidated Balance Sheets (AMZN). Come up with a standard naming for all financial statements.

Make row labels more comparable across companies

Currently using GAAP tags as labels. They are not always comparable across companies. For example, some companies report revenue as us-gaap_revenuefromcontractwithcustomerexcludingassessedtax, while others report as us-gaap_revenues. It'd be helpful to have a label that's just "Revenue".

Handling filings where metrics are quarterly or annual

For example, AAP reports revenue for the last four months:
https://www.sec.gov/cgi-bin/viewer?action=view&cik=1158449&accession_number=0001158449-20-000149&xbrl_type=v#