Inspired by what I had mentioned about linking directly to a component of an SEC XBRL filing (as opposed to having to get the entire filing), a statement made to me by a CPA about how it would be nice if I connected two things together, and some information provided to me by the SEC in response to a question I posed to email@example.com; I created this prototype mashup.
The mashup is similar to several things I already had so it literally only took me two hours to create. The first puzzle piece was this list of the taxonomiesfor the top 100 SEC XBRL filers by total assets. That is useful, but it leaves a lot to be desired in terms of what I was trying to achieve which is allow a CPA to quickly browse through SEC XBRL an see how others expressed their financial information in XBRL.
The second thing I did was this set of what I call exemplars which allows you to look at information for the same 100 SEC filers as above, but this time, it breaks the pieces of the filings into components; thus letting you focus on one specific component and browse through that by filer. For example, you could browse through the document information or the balance sheet or the income statement.
The obvious thing which is missing from that exemplar viewer is the actual rendering of the financial report. If you had that, plus the taxonomy you can really look at SEC XBRL filings in a very helpful way. So when someone showed me that Rivet CrossView Preview, what I saw was this. That was good, it makes it easier to browse SEC filings faster than you can at the SEC web site; but my first question was could I go directly to a component of one filing. Someone kindly pointed out that I could do that, and then this was the view I was able to have, linking directly to the Exxon income statement.
So my next question was: Could I generate the URL from my software, allowing me to autogenerate the links to this information.
Well, I know everything in that URL (the CIK, the accession number, the company name) EXCEPT for that funky "R2.htm". I new what it was, if you go to this link you can see these pages. But how do you figure out those page numbers?
Well, I looked at the pages and noticed that the pages correlated to the networks in a filing. So, I tried that and sure enough I was pretty successful predicting which network was in which of those "R" whatever HTML pages.
And that is how I created that mashup of my taxonomy information for one network and the corresponding rendering, complements of the SEC! Pretty slick.
But, what I noticed is that (a) I cannot be sure this is reliable, in fact it is not reliable. I have some glitch in my algorithm. If you go to filing #9 (AMERIPRISE FINANCIAL INC) you can see that I have the wrong file for the network I wanted to grab. My metadata is fine, I point to the correct balance sheet network (i.e. my taxonomy relations are fine). So, for this reason this approach may not be reliable. Also, I noticed some filings which have like 60 or so networks and there are only R files which go up to R42.htm. Not sure what the deal is with that.
Also, if you use those files, you will notice that the handy little pop up with the concept name, references, and other detailed information does not work.
I do believe that you can see how useful this functionality is. This will help me create this Accounting Trends and Techniquestype functionally for SEC XBRL financial filings. Do you think something like that would be useful with both the taxonomy relations, the viewer rendering, and all this information usableinto an application which generates XBRL? Let me know.