AUD/JPY and EUR/CHF leading indicator history, 2002-2009

User Rating: / 0
PoorBest 
Written by Forex Automaton   
Wednesday, 15 April 2009 12:34

This is a follow-up report on the predictive feature discovered in the correlation analysis of AUD/JPY and EUR/CHF. The feature was found by studying the time-integrated correlation of logarithmic returns in the two time series. A detailed time-evolution study is necessary to tell whether this effect is merely a result of a single high-impact event or a recurrent feature. The effect in question was found not to exist in the Asia-Pacific trading, therefore in this report I cover the trading hours from 4am to 8pm New York time, which excludes Asia. I extend the period of observation up to April 2009 and analyze the time stability of the leading indicator provided by AUD/JPY for EUR/CHF.

Historical bar chart of AUD/JPY since 2002, hour. Historical bar chart of EUR/CHF since 2002, hour.

Fig.1:Historical bar charts of the data used in the analysis. Top: AUD/JPY, bottom: EUR/CHF. Time scale is hour. Time axis is labeled in MM-YY format.

By visual inspection of the AUD/JPY and EUR/CHF charts, Fig.1, it's impossible to tell the "instantaneous" correlation, devoid of predictive value as such, from predictive correlations (those associated with non-zero time lags). The charts look somewhat positively correlated or the two price quotes sometimes move in tandem. Quantitative analysis is needed to measure the magnitude of the correlation and the range of time lags affected by it.

Time history of the AUD/JPY and EUR/CHF Pearson correlation coefficient, 2002-2009 Time history of AUD/JPY and EUR/CHF intermarket correlation, including non-zero time lags, 2002-2009 Time history of AUD/JPY and EUR/CHF intermarket correlation, including non-zero time lags, 2002-2009, zoomed

Fig.2: Correlation of logarithmic returns in AUD/JPY and EUR/CHF, derived from the hour-by-hour time series for the liquid trading hours, 4am to 8pm New York time, October 2002 to April 2009. Top: time history of the magnitude of the zero time-lag Pearson correlation coefficient. Middle and bottom: time history of the correlation in the vicinity of zero time lag. The peak structure is represented by three correlation values: the one for the zero lag (essentially a volatility measure) downscaled by 10 for easier visual comparison, the one at one hour lag and the one at minus one hour lag. The bottom panel is a zoomed view of the middle, focusing on the lower volatility period.

Fig.2 presents the correlation between AUD/JPY and EUR/CHF as a Pearson correlation coefficient, taking volatility (variance) change with time out of the picture. That the volatility does change can be seen from the bottom panel by looking at the history of the 0-lag correlation magnitude, a fair measure of variance in this situation (of near-zero average time series element). While the 0-lag correlation value is the highest, correlations with non-zero time lags are of particular importance for the purpose of forex trading system development.

From inspecting the pre-crisis structure of the correlation (before August 2007, best seen in the bottom panel), a curious insight is that the predictive correlations existed just fine even before the zero-lag correlation became significant, as seen from the bias of the red- and to a lesser extent green-hatched quantities towards the positive values. The result orginally reported came from the time series fragment ending in February 2008, and the correlation peak had strong content in the -1 hour lag bin. Fig.2, the two bottom panels, present the time history of that bin, integrated over time in the original article.

During the volatile fourth quarter of 2008, the peak remained strongly asymmetric with the asymmetry as originally inferred, that is for the AUD/JPY to lead EUR/CHF. The +1 hour lag, the other side of the peak, shows much less constancy in behaviour, getting strongly negative in the fourth quarter of 2008.

Bookmark with:

Deli.cio.us    Digg    reddit    Facebook    StumbleUpon    Newsvine
Last Updated ( Monday, 04 January 2010 12:37 )