Our research suggests we can forecast nearby magnitude ie the next 10-20 ticks (in our case this equates to 5 - 10 mins worth of data) with varying degrees of accuracy ranging from 65% - 80% + reliability depending on volatility.
The new sys generates a table of probabilities with the highest & lowest probabilities forming the basis of a trading strategy.
However the application still needs to be able to incorporate risk and firm money management principles to avoid getting things badly wrong.
Our experience suggests that extreme volatility preceeded by spikes are information based and cannot be forecast. However an analysis indicates spikes ,in certain circumstances, indicate a change or very strong enforcement of sentiment.
An example would be the London market expecting US base rates to rise so the Ftse trades down between -7 and -10 for most of the day. However US rates are moved higher or information accompanies the move that indicates more severe action in the future . The market spikes moving by 2 or three pts in the space of a minute. The difficulty is judging whether this is false or real ( ie prices will rebound quickly or continue down). Our new algorythms are able to assign probabilities to these scenarios that early testing indicates could be traded profitably.
