Now that asset managers have access to financial information resources like FactSet the hurdle to entry into the mainstream use of quant techniques has been lowered.
Managers who want to gain or maintain a competitive advantage must focus on areas with a higher entry hurdle, where the information is less easily available and less easily processed.
One area where this is possible is ‘market microstructure’. This is an hour-by-hour or sometimes minute-by-minute examination of the trading process to discover how stock prices adjust to reflect new information.
Examples include whether transactions take place at bid or ask prices, and the volumes that are traded at different prices and times during the days.
State Street Global Advisors (SSGA), through its Advanced Research Centre, is one asset manager that has identified this microscopic view of market movement as a way of adding value to the investment process.
Work in progress includes the use of portfolio flow data to help forecast industry returns. The aim here is to gauge the movement of portfolios in and out of industry sectors in advance so that SSgA’s portfolio managers can tilt their portfolios towards the right sector.
State Street bank’s custody operation, which handles some 15% of the world’s tradeable assets, provides the basis for this data. State Street’s proprietary Equity Flow Indicator, aggregated from about $9.5trn (e7.9trn) assets, measures total equity flows - net buys to net sells - as an indicator of investor demand.
Mark Hooker, director of SSgA’s Advanced Research Centre (ARC), explains: “Liquidity considerations or transaction costs may lead investors to gradually shift their portfolios in one direction or another, and picking up the movements in a timely way may be useful.
“For example, if the banks’ clients are moving as a whole from telecommunications to utilities, and they are going to be doing this over the next couple of weeks, if we can a signal three or four days into that portfolio reallocation that this shift is taking place and lean a little bit towards utilities in our portfolio, then that may be advantageous.
“People will point out that for every buyer there’s a seller, so why should portfolio balancing by our clients be any more different or informative than anyone else’s? We think it is because State Street’s clients are some of the larger and more sophisticated investors. Since we are measuring smart money relative to less smart money, these movements might be informative.
“We’re expecting the majority of excess return to come from choosing companies within each sector. But we are willing to take small sector bets. We use a combination of information – some top-down, some bottom-up with portfolio flow information coming in from the side.”
This microstructure driven predictor has been successful in every year except 1996 and 2000. More impressively, the power of this predictor has been persistent enough to predict returns not only on a short-term horizon but out as far as three months.
The hypothesis here, says Hooker, is that when the investors in the custody database are making a significant move from one industry to another they are basing that on information that is to some degree private and it will become more public over time. “Investors that are more insightful are guessing better than their competitors.”
Another area of market microstructure research is the use of high frequency data to predict market response to ‘gap’ events – sudden and unexpected movements up or down in a stock’s price. Gap events are usually triggered by information such as a company’s profits warning being revealed to the market.
The ARC has two objectives: to predict when a gap event is likely to take place; and, more important, to predict which direction the gap movement will go – up or down.
“We have looked a historical dataset of gap events and then measured the information that that was available leading up to these events and tried to infer the patterns.
“Since these gap events are measured in minutes, it makes sense to focus on a high frequency analysis. For the information leading up we look at days - up to 21days before the event - and then after the gap occurs we trace out the next five hours or so in periods of 15 minutes and see what typically happens.”
This minute-by minute analysis enables asset managers to see the precise point at which there is money to be made buying stock, he says. Typically the period between a negative gap event and a bounce back in the price is no more than a few minutes.
“The place where you have an opportunity to trade is within minutes of the event. If there was a 10% fall in the price, and if you could predict the gap response you would earn a 10% return in a matter of minutes. You could close your books and go home for the day.”