Thursday, 5 August 2010

Information Edge Five - Speed & Selectivity

Last year I wrote four times* on hedge funds utilising an information edge. The edge has to be exploitable and and through repetition become significant to the return series of the fund. This applies whether it is an edge in understanding (say the ability to extrapolate from few data points or read across from related industries) or more pertinently an edge in receiving proprietary information not generally available. There can also be an edge in hedge funds in the speed of reciept of information.

This last (superficially unlikely) edge is exploited by high frequency trading algorithms and stat arb managers. The extreme case of this need for speed is evident in the clustering of server farms near the exchanges. There can be a milli-second advantage of signals being received and sent over short physical distances between the computers generating trades and the exchanges fulfilling the orders down the pipes, whether on dark pools or electronic books on the bourses.  These systems are blind with no emotional or subjective inputs, but rely on accurate, logical and continuous pricing. As such, automatically traded capital is subject to event risk - if you like, the news on companies and sectors can get in the way of treating shares as trading chips for playing with minute-by-minute. 

This topic came to mind in reading a press release from Selerity, which describes itself as a low latency, real-time fact aggregation and event data company that caters to sophisticated investment firms including hedge funds, banks and proprietary trading firms. The Selerity technology searches and extracts "event data" from real-time primary sources, and delivers to clients machine-readable, actionable input.

In a new development Selerity software can now work with earnings pre-announcement event data as part of its content offerings to allow traders to use or avoid unexpected, market-moving events in their trading strategies. Corporate preannouncements are real-time, breaking updates released by companies regarding changes in their outlook or guidance ahead of scheduled earnings announcements, and as they are unscheduled fundamenal inputs these announcements can be disruptive to cluster trading, pairs trading and other high frequency trading strategies.

One of the essential tasks managers of high-frequency trading strategies (and nearly all quant managers) is to define their universe to exclude the outliers, anomalous stocks - those subject to corporate activity, and those having scheduled news releases for example - and make an effort to leave in their programmes only suitably qualified stocks by their criteria. Promptly identifying stocks that are subject to unexpected changes in earnings guidance from company management would help algorithmic traders and traders who deal in whole portfolios from being left marooned with unintended idisyncratic risk in the holdings/portfolios. 

There are many hedge funds that engage in event-driven strategies. The larger and more sophisticated ones use software systems to trawl through court documents, SEC Filings and other lengthy documents to find the few nuggets of information they seek. It is the document equivalent of conference delegates that come away with two or three quality pieces of insight or new understanding from a whole day sitting listening to presentations. The more deep and diverse the information inputs a manager uses in their process the more applicable are these tools.

The Selerity product is a specific example of a delivering appropriately framed and focussed information on a timely basis to a hedge fund or  investor of risk capital. But nearly all hedge fund managers need information of that sort, whether it is delivered via human vectors or software crawlers and spiders. The information edge has to supported and developed, and have a thorough process behind it.


No comments:

Post a Comment