Inside buyer, a web service

I have recently started devoting disciplined energy towards becoming financially literate.  You may wonder, "What does financial literacy have to do with the O'Reilly audience?"  Well, after spending some time playing around with the different financial web sites and reading a really interesting book, I started to contemplate a new web site (destination) and web service (APIs). The book I just read is called, �The Vital Few vs. The Trivial Many,� by George Muzea, and what is interesting about it is that it introduces a few meaningful concepts that I hadn't previously contemplated. 

One is a market fundamental that sixty percent of all stock price movements are related to the overall trend of the market.  This suggests, somewhat intuitively, that your ability to make money in the stock market is in direct proportion to your ability to invest only when the percentages are with you.  It just so happens that there is a narrow window of time that the market percentages are cyclically unfavorable, allowing you to buy in at favorable prices, and a narrow window of time that the market percentages are cyclically favorable, allowing you an ideal sell point (the proverbial buy low, sell high).  Since search patterns are well formed for ferreting out such opportunities, but the requisite publicly available data sources (that must be parsed) are dispersed over the Internet, this is an ideal application for �information management� automation.

In a nutshell, here�s how it works.  You buy beaten down stocks in mid-October and sell them all in mid-February the following year, making this a four-month strategy (I wish I was writing this column in October, but that�s another story).  The logic is that the market is inherently down-pressured by the institutional tax loss selling that must be completed by October 31 as mandated by the IRS (as opposed to tax loss selling for individuals, which can be completed until December 31), and up-pressured by the new money that flows into IRAs and 401Ks at the beginning of a new year. 

So how do you automate this?  Step one: start with the NYSE "New Low List" (in Barron�s weekly and Wall Street Journal daily), which essentially is a list of companies hitting 52-week lows. Step two: subtract liabilities and long-term debt from assets on an identified company�s balance sheet, removing entries that are a negative number (this data can be accessed at CBS Marketwatch).  Step three: from the remaining list, search the ten-year chart of each remaining stock (also available at CBS Marketwatch). If the current stock price of a given stock in the list is not in the lower third of its ten-year price history, remove it from the list. Step four: from the remaining list, filter on institutional ownership, where the institutional ownership value equals 30% or greater (data that is available at MSN Money). Step five: filter out remaining entries that have not at least shown a penny of profit in the last quarter�s earnings (also available at CBS Marketwatch).  The interesting thing, and part of the reason automation is compelling for this exercise is that from a list of 100 or so "New Lows," maybe 10 will remain after completing this process, and these stocks will have both strong balance sheets and current earnings, be in the bottom range of their 10 year trading history, and down pressured by the institutional tax loss selling that must be completed by October 31st.  Once the cyclical down pressure is replaced by the cyclical up-pressure on January 1, these stocks should rise.  The author of the book suggests that you determine your total investment and buy equal dollars (not shares) of each of the quality stocks and then sell them all by mid-February. The author has used this strategy successfully in 24 out of 25 years. 

Similarly, the author separates the market into what he calls The Trivial Many, the mass investor market and so-called experts, who you effectively want to bet against when investing, and The Vital Few, corporate insiders, such as officers, board members and major shareholders, whose buying and selling actions you should track like a hawk.  Generally speaking, insiders sell into price strength and buy into price weakness (again, buy low, sell high). You want to look for new stock purchases where insiders are buying as the stock goes up or when insiders are buying a depressed stock, and you want to see them increasing their ownership percentages by at least 30 percent of their holdings in the company. Also, since it is normal for insiders to buy as their stock goes down and sell as it goes up, you particularly want to look for divergence from this normal behavior. For example, your eyes should be wide when you see an insider, especially the chief financial officer who normally sells stock only when price rises suddenly break this pattern by selling into price weakness. It usually means that the company�s business conditions have deteriorated and that bad news is coming. On the other hand, you should be really impressed when you see insiders buy at higher prices than their earlier purchases. This usually means that business conditions are at least as strong as when they first bought, and in many cases, getting stronger. Better than expected news will more than likely surface a few months later. 

So how do you take advantage of this one technologically?  First off, insider actions are required to be registered with the SEC within two days of their initiation (e.g., an insider sells a block of their stock holdings), and this data is accessible within a specific database, known as EDGAR.  While there are premium services that show correlations between a given insider�s action and movement of the stock, the basic data is out there gratis.  The specific type of filing to search for is known as Form 4, or �Statement of Changes of Beneficial Ownership.� 

As there is a fairly finite amount of insider activity, and insider buys are generally more predictive than inside sales, database size should be manageable even over a period of years.  As to why insider buying is generally more predictive than insider selling, consider that insiders may sell for any number of non-business related reasons (such as to buy that vacation home) but will generally only buy if they believe that the market has under-valued their company�s forward looking business momentum. 

One way of fine tuning such a search query string is to track all insider buying actions, in terms of adding them to the database, but set a special flag to alert you when there are follow-on stock purchases by the same insider in the face of an increasing stock price (i.e., an insider buys 10,000 shares when the stock is at $10, and then purchases a similarly meaningful amount of shares when the stock hits $13 a few months later.  The logic here is that follow-on purchases in the face is rising share prices are heavily predictive.

Again, you can tweak the model in terms of what constitutes a meaningful amount of shares (in terms of number of shares or absolute dollars relative to the insider�s holdings in the company).  Similarly, you can track and set your own flags for correlations between frequency of buys, multiple buyers within the company, timing of purchases and the stock�s price.

The key point in all of this is that you can build a web site that expresses your best stab at providing online answers to different �what-if� financial questions built around objective models tied to predictive data that is publicly available.  Further the underlying elements of this web site can be expressed as a web service, enabling like-minded peers to tweak the inputs and outputs to their hearts desire.  So, for example, if a consumer of your web service loves a columnist, like Ken Fisher or Herb Greenberg, or an online pub like Motley Fool, they can build and maintain a portfolio that tracks their favorite columnists� recommendations and cross-validates them around the insider-buying model.� Pretty cool, I think.