USD/CHF intraday seasonality overview, 2003-2010

User Rating: / 2
PoorBest 
Written by Forex Automaton   
Thursday, 20 January 2011 11:15
Article Index
USD/CHF intraday seasonality overview, 2003-2010
Variances and correlations
All Pages

USD/CHF tends to show bearish dynamics for several hours following the start of American trading. The European night has a bias towards mean reversion.

The report uses hourly data from January 1st, 2003 through January 1st, 2011.

As always, the quantities we are going to look at are not the actual low, high and close. Since prices are always positive, they are trivially correlated; this feature is absent in the correlations of the so-called logarithmic returns (or logarithmic increments) which are the ratios of price levels (low, high, close) to the values they had during the previous hour-long time interval.

Central European time is chosen  because it allows one to split the forex week into 5 non-interrupted trading days.

Seasonality effects in first-order statistics (means) are straightforward to utilize in trading, therefore one could argue that such effects are not likely to be found. Nevertheless, sometimes such effects are visible and significant. To look for them, we average logarithmic returns for each hour of the day separately using profile histograms. The resulting histograms are plotted in Fig.1.

USD/CHF averaged log return vs CET hour of the day, in 2003, 2004, 2005, 2006 1.1 USD/CHF averaged log return vs CET hour of the day, in 2007, 2008, 2009, 2010 1.2 USD/CHF averaged log return vs CET hour of the day, sum over 2003-2010 1.3

Fig.1. Averaged logarithmic return vs hour of the day in CE time for USD/CHF. 1.1: Years 2003, 2004, 2005, 2006. 1.2: Years 2007, 2008, 2009, 2010. 1.3: All years added. Vertical bars indicate RMS precision of the mean.

Fig.1 presents hourly "seasonal" averages of the hourly logarithmic return in USD/CHF. One curious feature is the time window from 1:00 to 7:00. For many other currency pairs, stable patterns are found in this time window, and for USD/CHF, the pattern looks very stable at least during 2003-2006. It remains visible in 2007-2010 and certainly survives the averaging over all years in panel 1.3.

CHF also seems to be taking a "lunch break" during the hour finishing at 13:00, sharply falling (USD/CHF goes up). Similar effects have been seen in EUR/USD (EUR moves down) and AUD/USD (AUD moves down).


In the following figures, the shape of the autocorrelation peak and its daily evolution is studied by looking at two autocorrelation values, for the zero hour lag (approximated well by the variance, Fig.2) and for one hour lag (Fig.3).

USD/CHF variance of log returns vs CET hour of the day, in 2003, 2004, 2005, 2006 2.1 USD/CHF variance of log returns vs CET hour of the day, in 2007, 2008, 2009, 2010 2.2

Fig.2. Hour-scale variance of logarithmic returns vs hour of the day in CE time for USD/CHF. 2.1: Years 2003, 2004, 2005, 2006. 2.2: Years 2007, 2008, 2009, 2010. Hatched bands in the plots are twice as wide as the measurement's precision of the mean. The mean is the middle of the band.

The two peaks correspond to European and American peak activity hours. European trading in USD/CHF has been gaining volume progressively during these eight years, as the comparison of the relative magnitude of the European and US volatility peaks indicates.

The transitions between mean-reversion and trend-following regimes could be seen in the magnitude of autocorrelations at non-zero lags, in particular, one-hour lag. This magnitude as a function of hour is presented in Fig.3.

 

USD/CHF 1-hour lag autocorrelation of log returns vs CET hour of the day, in 2003, 2004, 2005, 2006 3.1 USD/CHF 1-hour lag autocorrelation of log returns vs CET hour of the day, in 2007, 2008, 2009, 2010 3.2 USD/CHF 1-hour lag autocorrelation of log returns vs CET hour of the day, sum of years 2003-2010 3.3

Fig.3. 1-hour lag autocorrelation of logarithmic returns vs hour of the day in CE time for USD/CHF. Every hour's candle is correlated with two adjacent ones. 3.1: Years 2003, 2004, 2005, 2006. 3.2: Years 2007, 2008, 2009, 2010. Hatched bands in the plots are twice as wide as the measurement's precision of the mean. The mean is the middle of the band.

In the autocorrelation of returns, trend manifests itself in positive autocorrelation magnitudes at non-zero lags, while mean-reversion -- in negative ones. The two effects can coexist, if the range of non-zero lags with non-zero correlation signals is broad enough for that. Fig.3 looks only at one-hour lag.

Comparison of different years in Fig.3 indicates that the correlation regime evolves with time: what looked like a trend-following during the American trading hours in 2003-2006 becomes questionable and no longer dominates the picture in 2007-2010. Only the negative autocorrelation in the first two time bins of the plot looks relatively stable.

A hypothetic actionable strategy for USD/CHF, based on these data, would be to go short every day at 16:00 CET and close the position at 21:00. Fig. 1.3 and 3.3 are consistent in that while 1.1 records a negative trend, 3.3 supports the idea by pointing towards a bias of forming a trend (broad positive autocorrelation) rather than zig-zagging during these hours.

Bookmark with:

Deli.cio.us    Digg    reddit    Facebook    StumbleUpon    Newsvine
Last Updated ( Monday, 07 February 2011 13:47 )