Financial markets are an example of a complex dynamical system with embedded feedback mechanism. Because of their complexity, the markets appear random most of the time. Usually a single price move on a weekly, daily or intra-day time-scale can be ignored as noise. There are moments, however, when price movement causes a strong reaction from market participants that exaggerates the move and turns it into a trend. Many momentum-based (trend-following) investment strategies exploit this quality of the markets. The challenge, of course, is to tell 'random' price movements from those that are likely to create a positive feedback for the price. Another challenge is to determine when the trend is about to end.
ChaosMonitor algorithm detects complex patterns in the time-series that normally occur before a major change in system behaviour. We call it system 'critical points'. The algorithm is based on the cutting edge research in the field of complex non-linear systems with applications in weather forecasting, studies of earthquakes and such fields as prediction of epileptic seizures.
A system in a critical state that we call a 'compression' point is likely to get out of its unstable equilibrium and to develop a positive feedback (think of a price breaking out of an established trading range, leading to aggressive position closing on part of some investors, and 'bandwagon' activity on part of others).
A system with a strong positive feedback (a strongly trending market such as an investment bubble or a market panic can be an example) can reach another critical state that we call an 'extension' point. In this state the system is likely to abruptly lose its momentum and get back to mean-reverting mode.
Above patterns in market system behaviour are not dependent on the underlying asset class. They are mathematical properties of the system itself, and can be detected on individual stocks, equity indices, fixed income instruments, commodities, currencies -- even on spreads between traded instruments. Time-scale also does not matter. We can detect critical points on intra-day time-series, and on weekly or even monthly data.
It is important to note that ChaosMonitor does not make forecast of future market price trajectory. We only detect moments of when a 'regime change' in the particular market is likely due to system instability. One consequence of this is that in case of a compression signal we do not know the future direction of the emerging trend. A wobbling pencil that stands vertically on a table can fall to the right or to the left. But once it starts falling, it is likely to continue to move in the same direction.
Another consequence is that ChaosMonitor can rather accurately predict the moment when a strong trend is becoming unsustainable and is likely to end (an extension signal). But we do not know if there will be a dramatic trend reversal or if the price will only settle in a sidewise trading range. Using the same analogy, we will warn our clients when the fall of the pencil is likely to end, but we don't know how hard the falling pencil will bounce.
Still, the information our clients receive from the analysis of market non-linear dynamics is extremely valuable, especially when there are signals on a weekly or monthly time-scale. One of the most dramatic illustrations of ChaosMonitor predictive power was the compression signal on VIX volatility index that appeared just before the market panic in the autumn of 2008.
Most people would find the idea that ChaosMonitor predicted the collapse of Lehman Brothers and the market turmoil that followed ridiculous. And we would agree with them. The truth is of course that Lehman Brothers was not the cause of the market sell-off in 2008. The chart above shows that equity markets have entered the state of extreme instability by the end of August 2008. The resulting spike in volatility and risk-aversion has affected Lehman Brothers' price of funding as well as value of their assets so much that they went bankrupt by mid-September. Their demise has further intensified the panic providing a nice illustration of system positive feedback in action. ChaosMonitor did not predict the collapse of Lehman Brothers but it correctly and timely identified the conditions leading to it.
ChaosMonitor algorithm detects complex patterns in the time-series that normally occur before a major change in system behaviour. We call it system 'critical points'. The algorithm is based on the cutting edge research in the field of complex non-linear systems with applications in weather forecasting, studies of earthquakes and such fields as prediction of epileptic seizures.
A system in a critical state that we call a 'compression' point is likely to get out of its unstable equilibrium and to develop a positive feedback (think of a price breaking out of an established trading range, leading to aggressive position closing on part of some investors, and 'bandwagon' activity on part of others).
A system with a strong positive feedback (a strongly trending market such as an investment bubble or a market panic can be an example) can reach another critical state that we call an 'extension' point. In this state the system is likely to abruptly lose its momentum and get back to mean-reverting mode.
Above patterns in market system behaviour are not dependent on the underlying asset class. They are mathematical properties of the system itself, and can be detected on individual stocks, equity indices, fixed income instruments, commodities, currencies -- even on spreads between traded instruments. Time-scale also does not matter. We can detect critical points on intra-day time-series, and on weekly or even monthly data.
It is important to note that ChaosMonitor does not make forecast of future market price trajectory. We only detect moments of when a 'regime change' in the particular market is likely due to system instability. One consequence of this is that in case of a compression signal we do not know the future direction of the emerging trend. A wobbling pencil that stands vertically on a table can fall to the right or to the left. But once it starts falling, it is likely to continue to move in the same direction.
Another consequence is that ChaosMonitor can rather accurately predict the moment when a strong trend is becoming unsustainable and is likely to end (an extension signal). But we do not know if there will be a dramatic trend reversal or if the price will only settle in a sidewise trading range. Using the same analogy, we will warn our clients when the fall of the pencil is likely to end, but we don't know how hard the falling pencil will bounce.
Still, the information our clients receive from the analysis of market non-linear dynamics is extremely valuable, especially when there are signals on a weekly or monthly time-scale. One of the most dramatic illustrations of ChaosMonitor predictive power was the compression signal on VIX volatility index that appeared just before the market panic in the autumn of 2008.
Most people would find the idea that ChaosMonitor predicted the collapse of Lehman Brothers and the market turmoil that followed ridiculous. And we would agree with them. The truth is of course that Lehman Brothers was not the cause of the market sell-off in 2008. The chart above shows that equity markets have entered the state of extreme instability by the end of August 2008. The resulting spike in volatility and risk-aversion has affected Lehman Brothers' price of funding as well as value of their assets so much that they went bankrupt by mid-September. Their demise has further intensified the panic providing a nice illustration of system positive feedback in action. ChaosMonitor did not predict the collapse of Lehman Brothers but it correctly and timely identified the conditions leading to it.
= large price move next
or 
= this price move ends