OT: what looks good

All,
short answer: Verizon (VZ), Intel (INTC), Biogen (BIIB).

Much longer answer (possibly only of interest to software engineers):
6 years ago, I wrote a script that downloaded (and parsed) 10-Ks in XBRL, from the SEC website. The script started failing a few months back, because the SEC made their bot detection considerably tighter. Never mind that my ancient laptop in no way could do a DOS attack on SEC.gov any more than an ant could crush an elephant under its foot. So I had an email exchange with them. They were very nice, and pointed to a newer and much better way to get the same data. Turns out, they now publish the data in a much easier-to-parse JSON format. So I spent a week to rewrite my script to parse JSON.

The script is a stock screener and (in my imagination anyway), follows a Buffett-like approach. That is, it gives considerable weightage to predictability of earnings, tries to be conservative, avoids using a discount rate and (partly or wholly because I am a good but not a great programmer) ignores the “story” aspects of the stock such as its popularity. Unlike Buffett, it is completely backwards looking and assumes the future will look like the recorded past.

In the past it did pick out some of the same stocks that Buffett did, namely, IBM and Apple; but also some that he ignored, such as Viacom. It is interesting that the script is picking Verizon now, and Buffett invested in it comparatively recently. He is probably avoiding Intel and Biogen for good reasons that a script that can only analyze financial statements will not guess.

14 Likes

I hope Intel does well. This article talks about why the stock might not be the best choice for a while longer. Begin quote…

Intel CEO Is Tired of Wall Street Doubting His Comeback Plan

Intel Corp. Chief Executive Officer Pat Gelsinger, who took the job in February 2021, gives himself an A- grade for his first year running the chipmaker. Investors are proving to be tougher graders.

https://www.bloomberg.com/news/articles/2022-02-16/intel-s-c…

2 Likes

6 years ago, I wrote a script that downloaded (and parsed) 10-Ks in XBRL, from the SEC website. The script started failing a few months back, because the SEC made their bot detection considerably tighter. Never mind that my ancient laptop in no way could do a DOS attack on SEC.gov any more than an ant could crush an elephant under its foot. So I had an email exchange with them. They were very nice, and pointed to a newer and much better way to get the same data. Turns out, they now publish the data in a much easier-to-parse JSON format. So I spent a week to rewrite my script to parse JSON.

As a software engineer, I’m curious. My questions:

  1. Where is the data provided in JSON format?
  2. How well do companies comply? I’ve heard that poor compliance has always been an issue since XBRL first came out.
  3. Is your source code on GitHub/GitLab/BitBucket?
  1. Where is the data provided in JSON format?
    https://www.sec.gov/edgar/sec-api-documentation > https://www.sec.gov/Archives/edgar/daily-index/xbrl/companyf…
    There are other ways such as SOAP/REST. Also, there is another file to associate tickers with CIKs/names.

  2. How well do companies comply? I’ve heard that poor compliance has always been an issue since XBRL first came out.
    Good question. Compliance is very good. The problem is that accountants don’t understand software engineering very well and vice versa. So XBRL permits too many ways to state the same financial figure such as sales. Companies have too much latitude to use different tags for essentially the same line item.

  3. Is your source code on GitHub/GitLab/BitBucket?
    No, and never will be. I am hoping to monetize the script.
    The last five years have seen growth/story stocks triumph conclusively over value stocks. The script can only identify value stocks. Now the tide seems to be turning and hopefully the script’s record (my record!) will improve. We will see.

6 Likes