tjbresearch.com
Thursday, December 3rd, 2009
behind the scenes

Any investment endeavor — from that of the biggest, most sophisticated asset manager to the day trader in his bathrobe in his basement — relies on accurate and timely information.  In a firm, doing the plumbing on those information pipes is part of “operations” or the “back office.”  Often, those areas are taken for granted.  (That is, until something goes wrong.)

The firms that settled under the Global Research Analyst Settlementthe research puzzle | This is the seventh in a series of postings about “the GRAS.”  This link will take you to an index of the postings on the topic that will include future postings as they are written. needed to work with outside vendors and consultants like me to ensure that the independent research they were obligated to deliver made its way to the intended recipients as it should have.  Along the way, the appropriate statistics had to be gathered for reporting to the regulatory authorities.

The specifics of the implementation differed by firm, and I won’t deal here with the intricacies of the application.  Instead, I’ll focus on four examples of situations with broad implications in the investment world.

Dan Brown has made a fortune off of tales of intrigue centered around the fictional Harvard “symbologist” Robert Langdon.   I dream of making a fortune as a symbologist on Wall Street.  I write a fair bit about the belief structures that drive investors and the symbols and shortcuts that they use, but here I’m talking about something different.  Stock tickers (or symbols), those short and unique representations of a security, pose all sorts of operational problems.  U.S. equities are as straightforward as you can get, but even they cause issues throughout the system of data terminals and repositories.  Static and functional most of the time, changes in them can have important repercussions on decision tools.

Most of those types of changes were more of a nuisance than anything else under the settlement, which dealt primarily with domestic equities.  Broader applications involving foreign stocks and other asset classes are more challenging, and algorithmic trading applications offer another level of complexity.

I have a hard time believing that this crazy quilt of security identifiers will be with us in twenty years.  New options symbols are on the horizon, Bloomberg is starting to establish unique identifiers for all securities, and there are other developments of note.  Time to work on my symbology chops.

Another operations issue is really a research reporting issue that ties back to my last post on performance.  Remarkably, we still have research reports that are created after market-moving news about a company has been released, but which bear the closing price of the security prior to the release of the news.  I grant you that it’s sometimes tough to find a better price if there’s no or limited off-hours trading, but stamping a price on a report that misrepresents the state of the world is unacceptable.  Our customs (write a report, use last price) get in the way of the accurate capturing of when and how a point of view has changed.

An educated reader of a report should be able to tell that the price is not representative, but a less sophisticated one will not.  In any case, in these unfortunate situations the metadata on the report can show a new rating based upon the new information, but an old price.  Some performance measurement services work directly from that metadata and therefore institutionalize the distortion.  Garbage in, garbage out:  Another reason to take performance rankings with a Minnesota road truck’s load of salt.

The consultants involved in the settlement were required to fill out annual reports about our work, and to include relevant statistics that had been compiled.  One set of particular interest was the usage numbers.  The debate about whether the settlement spawned enough usage of independent research can wait for another day; right now, I’m interested in the calculation and interpretation of such aggregations.  Too often, I see people assume a level of precision that’s not there.  There are a variety of issues with web analytics, and you ought to understand that before reporting or using information based upon clicks.Wikipedia | Speaking of sources of information on which there is some debate regarding accuracy, Wikipedia is useful for summaries of issues like this one. The metrics are broadly indicative and helpful, but any internal or external use of the information ought to be within the context of its shortcomings.

In every investment application, data quality is critical, yet many organizations don’t put forth enough resources (or structure them in an optimal fashion) to lower the rate or lessen the impact of data errors.  A year ago, I wrote about “the art of triangulation,” including the way in which analyzing data from different directions helps to improve its quality.the research puzzle | I also focused on triangulation in products, investment decisions, regulation, and due diligence. I saw much benefit during the settlement from using the separate databases of the investment bank and the technology aggregator in conjunction with my own; the comparisons helped to ensure that the data flow to the end user was as good as it could be.

We’re almost done looking at the settlement.  On now to thoughts about regulation in general and this specific case.