I've updated my Directional Similarity spreadsheet through Oct 3, 2014. The complete description of the spreadsheet can be found here (post 165 of "My Multi-Currency" thread):
http://www.forexfactory.com/showthre...=487576&page=9
In a nutshell:
I make trading decisions based off the weekly charts, and because I'm interested in setting up pseudo-hedges, I wanted to answer this simple question: How many times during the past 52-weeks did the week-to-week exchange rate of one currency pair move in the same direction as another currency pair? I call this "DIRECTIONAL SIMILARITY". The calculation of the week-to-week directional similarity of any two currency pairs is very simple: for both pairs, calculate their most recent weekly close minus their weekly close from the previous week, and then compare these two weekly pip differentials. If both weekly pip differentials are positive, or if both are negative, or if both are zero (this never has happened), then the comparison is scored as a "1". Otherwise ALL OTHER weekly pip differential comparisons are scored as "0", i.e. If one weekly pip differential is positive and the other is either zero or negative = score "0", or if one weekly pip differential is negative and the other is either zero or positive = score "0", or if one weekly pip differential is zero and the other is either positive or negative = score "0". These weekly binary scores are then added up for the previous 52 weeks and the Percent Directional Similarity is calculated (i.e. 52-week Sum of the Binary Score/52). Note that the "move in the same direction" can be as little as 1 pip. This is not “correlation” since the statistical parameters used for correlation are much more comprehensive than the calculations used in my simple algorithm. The Directional Similarity calculation described above can be computed for any time frame.
So for instance, as of October 3, 2014, the 52-week Directional Similarity of EURUSD vs. USDCHF = 6%. This means that the exchange rates of the EURUSD and USDCHF moved in the same direction only 3 weeks out of the previous 52 weeks (Oct 1, 2013 - Oct 3, 2014). Compare this to EURJPY vs. CHFJPY; their 52-week Directional Similarity = 86%, meaning that both exchange rates moved in the same direction 45 weeks out of the previous 52 weeks.
Also included in the spreadsheet for each pair comparison is the year-to-year "Average Directional Similarity" (calculated over the time period of January 2008-Oct 3, 2014, N=7 yearly time periods, although the current year is not yet complete, so the column marked "2014" is actually the rolling 52 week period of Oct 1, 2013 -Oct 3, 2014 ), and the year-to-year standard deviation which gives an indication of the volatility of the "sameness" of the directional movement over the seven year period. Obviously, as time passes and more data is collected, the Average & standard deviation will be a better reflection of the directional similarity over time...
For me, the information in this spreadsheet is used to determine which currency pairs should be hedged (positive or negative) based on their historical similarity of price direction. I hope that this information may be useful to you as well...
Dave
http://www.forexfactory.com/showthre...=487576&page=9
In a nutshell:
I make trading decisions based off the weekly charts, and because I'm interested in setting up pseudo-hedges, I wanted to answer this simple question: How many times during the past 52-weeks did the week-to-week exchange rate of one currency pair move in the same direction as another currency pair? I call this "DIRECTIONAL SIMILARITY". The calculation of the week-to-week directional similarity of any two currency pairs is very simple: for both pairs, calculate their most recent weekly close minus their weekly close from the previous week, and then compare these two weekly pip differentials. If both weekly pip differentials are positive, or if both are negative, or if both are zero (this never has happened), then the comparison is scored as a "1". Otherwise ALL OTHER weekly pip differential comparisons are scored as "0", i.e. If one weekly pip differential is positive and the other is either zero or negative = score "0", or if one weekly pip differential is negative and the other is either zero or positive = score "0", or if one weekly pip differential is zero and the other is either positive or negative = score "0". These weekly binary scores are then added up for the previous 52 weeks and the Percent Directional Similarity is calculated (i.e. 52-week Sum of the Binary Score/52). Note that the "move in the same direction" can be as little as 1 pip. This is not “correlation” since the statistical parameters used for correlation are much more comprehensive than the calculations used in my simple algorithm. The Directional Similarity calculation described above can be computed for any time frame.
So for instance, as of October 3, 2014, the 52-week Directional Similarity of EURUSD vs. USDCHF = 6%. This means that the exchange rates of the EURUSD and USDCHF moved in the same direction only 3 weeks out of the previous 52 weeks (Oct 1, 2013 - Oct 3, 2014). Compare this to EURJPY vs. CHFJPY; their 52-week Directional Similarity = 86%, meaning that both exchange rates moved in the same direction 45 weeks out of the previous 52 weeks.
Also included in the spreadsheet for each pair comparison is the year-to-year "Average Directional Similarity" (calculated over the time period of January 2008-Oct 3, 2014, N=7 yearly time periods, although the current year is not yet complete, so the column marked "2014" is actually the rolling 52 week period of Oct 1, 2013 -Oct 3, 2014 ), and the year-to-year standard deviation which gives an indication of the volatility of the "sameness" of the directional movement over the seven year period. Obviously, as time passes and more data is collected, the Average & standard deviation will be a better reflection of the directional similarity over time...
For me, the information in this spreadsheet is used to determine which currency pairs should be hedged (positive or negative) based on their historical similarity of price direction. I hope that this information may be useful to you as well...
Dave
Attached File(s)
