Incorporating seasonality into Heidi. A concept of a better forecasting component for an intraday trading system. - Intraday Season 1 Optimization: 8-10pm ET, 2-4 CET

User Rating: / 3
PoorBest 
Written by Forex Automaton   
Monday, 07 March 2011 17:21
Article Index
Incorporating seasonality into Heidi. A concept of a better forecasting component for an intraday trading system.
Intraday Season 0 Optimization: 5-7pm ET, 23-1 CET
Intraday Season 1 Optimization: 8-10pm ET, 2-4 CET
Intraday Season 2 Optimization: 11pm-1am ET, 5-7 CET
Intraday Season 3 Optimization: 2-4am ET, 8-10 CET
Intraday Season 4 Optimization: 5-7am ET, 11-13 CET
Intraday Season 5 Optimization: 8-10am ET, 14-16 CET
Intraday Season 6 Optimization: 11am-1pm ET, 17-19 CET
Intraday Season 7 Optimization: 2-4pm ET, 20-22 CET
All Pages

Season 1 presents somewhat of a dilemma when it comes to optimizing the prediction for hourly close. If the system were to only trade EUR/USD and USD/CHF, I would say that one would need Fred below 10. However, for the rest of the time series, such a choice would yield negative correlations, which is to say that playing against the forecast would be indicated. The compromise solution is to take Fred between 30 and 50 and flip sign.

For the hourly high and low, and the corresponding cumulant, the situation is staightforward and the compromise solution outlined above could also be the best solution from the point of view of these figures of merit, if Fred=33 is chosen again.

Dependence of Pearson correlation coefficient between predicted and actual logarithmic differences in hourly HIGH on the optimization parameter Fred. Season 1. 1.1 Dependence of Pearson correlation coefficient between predicted and actual logarithmic differences in hourly LOW on the optimization parameter Fred. Season 1. 1.2 Dependence of Pearson correlation coefficient between predicted and actual logarithmic differences in hourly LOW on the optimization parameter Fred. Season 1. 1.3

Fig. 1.1-1.3 Dependence of Pearson correlation coefficients between predicted and actual logarithmic differences (returns) in the three components of the hour candle (high, low and close, in that order) the optimization parameter nicknamed Fred. 1.1: hourly high, 1.2: hourly low, 1.3: hourly close. Data are for the Intraday Season 1.

Dependence of normalized 4-point cumulant among predicted and actual logarithmic differences in hourly HIGH and hourly LOW on the optimization parameter Fred. Season 1. 1.4

Fig. 1.4 Dependence of the normalized 4th order cumulant among predicted and actual logarithmic differences (returns) in hourly high and low on the optimization parameter nicknamed Fred. Data are for the Intraday Season 1.



Last Updated ( Saturday, 07 July 2012 10:44 )