Four Market Properties:

The Cornerstones of Market Predictability

The Scaling Law

Scientists have demonstrated how natural phenomena - from the leaves of a fern to the structure of blood vessels in the lungs - display fractal properties. Through our study of high-frequency pricing data, Olsen has uncovered similar fractal patterns in the fx market.

The pattern of volatility measured at 10-minute intervals is similar to that measured at one-hour intervals. We have demonstrated that this fractal "scaling law" holds for periods up to two months and beyond. It also holds for the frequency with which prices change direction: the pattern for the direction of price change at one-hour intervals is similar to that for one-day intervals.

 

Seasonality in the Markets

In addition to displaying fractal regularities, the fx market exhibits daily and weekly "seasonality". Volatility and other market variables follow patterns, or seasons, according to the time of day and the day of the week.

FX markets operate 24 hours a day around the world during the business week. In the Far East, trading begins on Monday morning and builds up steam as Tokyo and Sydney are joined by Singapore and Hong Kong. At 4:00 a.m. (Greenwich Mean Time) there is an abrupt decline as these markets break for lunch; an hour later trading picks up again and remains strong throughout the morning as the closing Asian centers are replaced by markets on the European continent and in London.

The overlap of European and U.S. markets then produce the peak of daily activity, which declines steadily as trading centers across the U.S. close.

Seasonality is not a quaint anecdote. It is visible in the volatility of prices, the relative spread, tick frequency, the tendency of volatility to follow trends, and the frequency with which prices change direction. And seasonality can be unique to specific financial instruments: for example, U.S. traders are more interested in the Yen than the Swiss Franc, and this predilection affects activity in these currencies through the course of the day.

 

The Business Times Scale

To accommodate this seasonality in its intra-day analysis, Olsen has developed an adaptable model of time: when fewer traders are present in the market, data is compressed; for more active periods, data is expanded (given greater emphasis). This "business time scale" allows for more appropriate weighting of information: the behavior of the market at any particular moment is interpreted in the context of the number of participants present.

In addition to daily and weekly seasonality, Olsen applies business time in a more general way: during periods of high activity, we measure it at a higher resolution in order to capture the full detail of market dynamics. Conversely, when activity is low, we take a lower-resolution view.

The presence of scaling law and seasonality enables forecasting of price evolution. The fractal nature of the market means that short-term volatility (for example, over one hour) can be predicted by longer-term volatility (over one day).

 

Market Heterogeneity

Conventional wisdom says that all market participants are much the same. The same information is available to everyone, and investors must inevitably respond to news in the same way.

On the contrary, we believe market participants fall into quite distinct groups. Depending on their characteristic trading patterns and appetite for risk. FX dealers and market makers, for example, typically trade for profit at relatively high risk and relatively often - perhaps many times in one day. At the other extreme, central banks trade infrequently, usually for political or national economic reasons and not for profit. Similarly, companies and pension-fund managers tend to trade infrequently and avoid risk wherever possible.

Because this range of groups respond to events (internal and external) from a diversity of viewpoints and trading habits, patterns emerge.

Market heterogeneity means that the impact of an event does not play out immediately in the market, but slowly dissipates at different rates over time. The initial reaction to an event is followed by series of secondary reactions (aftershocks)

Why do these insights matter?