Forecasting, meteorological or financial, needs massive amounts of data. For decades some financial analysts have drawn charts of price movements in an attempt to find patterns. Analysts try to detect when prices are breaking out of a range that is drawn on the chart. This is called "technical analysis" or "chartism." Traditional chartism is slow and crude. It drastically over simplifies what is a highly complex, dynamic and fast-moving process and so was regarded with contempt by many analysts. Computerized pattern recognition, on the other hand, is immensely more detailed and sophisticated. Computers can detect patterns that humans cannot detect.
Chartists usually studied charts that showed the end-of-day closing prices. Computers drew lines on such charts and tried to indicate when a price trend was breaking out of an established pattern. This was crude and the results were not too impressive. However, computers can analyze all the data they can get. When applied to forecasting, they need to see every price movement, not just end-of-day prices. They need tick-by-tick details of every change in quoted price and the volume and price of every trade. Such data can be collected from the stock-market feeds of Reuters, Knight-Ridder and Telerate. Quoted prices of major currencies can change 20 times a minute or more. The US dollar/deutschmark rate, for example, can change 18,000 times in a single day.
Such a body of data made it possible to test hypotheses about price movement and experiment with proposed techniques for automated trading. Investment tools were programmed, run against historical data and modified until they produced the best results achievable.
This data sits in "data warehouses." Tick-by-tick details of every movement sound like a massive amount of data; but, in fact, it is much less than that in many commercial data warehouses (for example, Wal-Mart's warehouse). Warehouses for the computerization of investment have billions of bytes of data; Wal-Mart's warehouse for optimizing the delivery and positioning of goods on supermarket shelves has trillions of bytes.
A data warehouse for finding investment patterns precise enough to help predict future prices can fit in the desk drawer of a personal computer user today. A billion bytes fit on a pocket JAZ disk.
In 1986, Richard Olsen, a Swiss, whose father was a banker and whose mother was a professor, started a company in Zurich, Olsen and Associates, with the goal of predicting the future prices of financial instruments using only information from past prices. He recruits the brightest young Ph.D. physicists and statisticians he can find from CERN and other top research centers. He tells them that the science of financial markets is more intellectually fascinating than particle physics. He says that this new science is in the position today that quantum mechanics was in the 1920s before it wreaked havoc in physics. The understanding of foreign exchange markets is being condensed into reliable theory that will wreak havoc in the financial world. It will be made practical by computers. Olsen's team couldn't be less like a Swiss bank. They wear jeans, not ties, and indulge in intense intellectual arguments. Olsen gives them massive computer power, detailed data warehouses and infectious enthusiasm. His team is convinced that it is involved in an area of science that is about to explode. Olsen's favorite metaphor is that events affecting the market are like stones dropped into a lake. His company is not interested in trying to predict the stone's fall, but thinks that it can predict the pattern of waves that spreads from it.
After many years of research, Olsen and Associates claims to be predicting the general direction of currency markets 60 to 65 percent of the time for short term (three-month) predictions and 70 to 75 percent of the time for longer term predictions. These figures are significantly better than those of the highly paid floor traders.
A number of other firms have set out to find computerized techniques that might enable them to predict price movements. Even if they only nudged the probability of being correct by a minor amount, that could translate into big money. The realization that exchange rate forecasting might be possible has lured brilliant mathematicians from fields like nuclear physics--and weather forecasting. Can financial storms be predicted? More alarming, can a highly leveraged investor make storms happen so as to profit from them?
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access