CHF/JPY intraday seasonality overview, 2003-2010

User Rating: / 2
Written by Forex Automaton   
Tuesday, 15 February 2011 12:47
Article Index
CHF/JPY intraday seasonality overview, 2003-2010
Variances and correlations
All Pages

This CHF/JPY report completes a set of systematic intraday pattern research reports involving JPY as the base currency. This will be followed by a synthesis report for JPY, following the format already tried for USD.

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

As I strive to make each of these reports self-contained, they are somewhat repetitive, especially when it comes to describing the analysis approach. The reader familiar with the approach can skip to Fig.1.

As always, the quantities we are going to look at are not the actual hourly 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 to the values they had during the previous hour-long time interval.

As in other posts dealing with intradey seasonality, I use Central Europen here. The convenience of using CET is that one gets 5 non-interrupted, full 24-hour long trading days per week, as explained in an article on CET which also provides a time zone conversion table.

Seasonality effects in first-order statistics (means) are straightforward to utilize in trading. A significant positive mean of log return for an hour means that on average, the quote tends to move up during that hour. Likewise, a significant negative mean indicates the opposite. An efficient market "theory" partisan therefore could argue that such effects are not likely to be found, or in any event would not exceed the costs of execution. Nevertheless, previous intrady seasonality studies involving USD indicated that such effects exist -- an example of how rational expectations regarding markets do not work. To look for the effect, we average logarithmic returns for each hour of the day separately using profile histograms. The resulting histograms are plotted in Fig.1.

Same analysis and plotting approach was tested on "placebo" random walk data to ensure that interpretations are not misguided by misunderstood features of analysis techniques.

CHF/JPY averaged log return vs CET hour of the day, in 2003, 2004, 2005, 2006 1.1 CHF/JPY averaged log return vs CET hour of the day, in 2007, 2008, 2009, 2010 1.2 CHF/JPY 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 CHF/JPY. 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. The usual caveat is that since the distributions of (even logarithmic) returns are generally non-Gaussian with heavier tails, confidence intervals to use with these errorbars are lower than standard.

Fig.1 presents hourly "seasonal" averages of the hourly logarithmic return in CHF/JPY. The hourly averages can deviate from zero being as large as 2 pips. The log return behavior that takes place during the "forex night" between 19:00 and 3:00 very much resembles what has been seen in AUD/JPY, EUR/JPY, and USD/JPY, The predominantly downward trend for the hour ending at 00:00 CET (marked 24 in the figures), seen also in those other studies involving JPY, is very strongly pronounced here. The different datasets sharing the same base currency (JPY) are not fully idependent, and JPY plays a big role in creating similarities in daily patterns.

Last Updated ( Thursday, 17 March 2011 17:13 )