Summary of the trading system optimization results. Step One. - Forecasting

User Rating: / 2
Written by Forex Automaton   
Tuesday, 21 July 2009 14:28
Article Index
Summary of the trading system optimization results. Step One.
How often to place new trades?
The multivariate optimum
Winners. Future plans.
All Pages

4. The forecasting parameter

The subject of the forecast is the difference between the closing price of the next (future) time step and the close of the present time step. It is expressed as a fraction of the present price. Essentially, this is the output of the forecasting procedure. It is this quantity that is filtered through the entry threshold discussed above. Currently, a single parameter controlls the way the forecast is obtained. For brevity's sake, this parameter is called Fred to avoid disclosing its nature.

Annualized return vs forcasting parameter

Fig.4.1. A profile histogram showing dependence of the annualized return (measured directly for each algo trader on the forecasting parameter by comparing equity at the beginning and end of trading history) on the stop-loss placement. Different symbols represent different trade entry parameter ranges. The unit of return is 100% (100%=1, 10%=0.1 etc).

Annualized return vs forecasting parameter, AUD/USD Annualized return vs forecasting parameter, EUR/USD
Annualized return vs forecasting parameter, GBP/USD Annualized return vs forecasting parameter, USD/CAD
Annualized return vs forecasting parameter, USD/CHF Annualized return vs forecasting parameter, USD/JPY

Fig.4.2. Collection of profile histogram fragments following the same format as Fig.4.1, but for the individual markets. Compare the trends in the distributions of the points. Click on any panel to get to the report with the full information.

As the figures 4.1 and 4.2 fail to hide from you, Fred is by far the least trivial parameter to optimize. It is not clear why the optima are where they are, and whether there are credible reasons for the differences in the Fig.4.2 shapes among the different markets. In all Fig.4.2 panels, the major maximum is located in the right half of the panel. The theme of a second hump (local maximum) at lower Fred values, those in the left half of the respective panel, seems to have some prevalence nonetheless: such a local maximum can be recognized without doubts in GBP/USD and USD/CAD, and it is not without a certain degree of credibility in AUD/USD, EUR/USD and USD/CHF. In GBP/USD and USD/CHF it is this left maximum that provides not just a local, but a global maximum.

If one allows oneself to be tempted to classify things a bit, GBP/USD and USD/CAD seem to form a "group", and then AUD/USD, EUR/USD and USD/JPY fall together in the opposite "group" on the basis of similarity of Fig.4.2 patterns. USD/CHF is then left by itself as an "intermediate" shape between the two groups.

The peak profitability at low Fred values, according to Fig.4.1, where the data from Fig.4.2 are mixed (averaged) together, is the property of the not-too-conservative trade-entry regimes (red symbols in the figure).

Last Updated ( Monday, 04 January 2010 12:30 )