How to Invest in Forex Now?
By Irene AldridgeMany investors, wary of the volatility in the stock markets, are looking for a safe asset class to place their hard-earned nest eggs. Foreign exchange, or forex, is one such class right now. However, for many investors, forex is an unknown and untested entity. This article offers a case study to illustrate an approach to invest in a foreign currency, US-to-Canadian dollar exchange rate, denoted USD/CAD.
Figure 1 shows the price levels of USD/CAD over the past two years. To an untrained eye, these price levels are random and can be intimidating. Academics have proposed various diverging strategies for predicting the fair price levels of currencies, including USD/CAD, yet, most of these theories have failed the test of time and data.
One particular economic theory still appears to have legs and may prove useful to investors interested in forex: the Purchasing Power Parity (PPP). PPP states that the forex rate is really a reflection of the two countries relative inflation rates. Both the U.S. and Canadian governments report inflation as a change in the monthly Consumer Price Index (CPI). While the PPP addresses respective inflation rates, this article shows that the U.S. inflation rate alone has a significant enough impact on USD/CAD to warrant a sound investment strategy.
Figure 2 illustrates the monthly changes in the U.S. CPI over the past two years. While it is difficult to estimate the relationship across the inflation rates and the USD/CAD just from the charts, some additional math uncovers the following dependency: following each announcement of rising U.S. CPI, USD/CAD falls with a very high probability, often approaching 99.99%.
Figure 3 illustrates the phenomenon: following each increase in the U.S. CPI, USD/CAD drops sharply for 5 trading days following the CPI increase announcement (conversely, USD/CAD rises for 5 trading days following an announcement of a fall in the U.S. CPI). On average, for every 1 point increase in the U.S. CPI, USD/CAD falls over 600 bps over the following 5 days, as measured from the closing price of USD/CAD on the day preceding each U.S. CPI announcement. In addition, as the right-hand axis of Figure 3 shows, the response of USD/CAD is extremely statistically significant: the likelihood of USD/CAD falling (rising) within the first few days after a rise (fall) in the U.S. CPI is well over 95%.
To actively invest in USD/CAD, investors should closely monitor the U.S. CPI announcements. The next U.S. CPI announcements are expected at 8:30 AM on November 17 and December 15, 2010; see The Quant Investor’s Almanac 2011 for all the dates and times of the U.S. CPI releases in 2011. As soon as possible after a U.S. CPI release, place a trade in USD/CAD: after a positive change in the U.S. CPI, sell USD/CAD (buy Canadian dollars with U.S. dollars); after a negative change in the U.S. CPI, buy USD/CAD (sell Canadian dollars and buy U.S. dollars instead). On the fifth trading day after the announcement, reverse your position, capturing the gain.
Irene Aldridge is a quantitative portfolio manager and managing partner at Able Alpha Trading, LTD., a U.S. Registered Investment Advisor in New York City. Aldridge is also a co-author of The Quant Investor’s Almanac 2011: A Roadmap to Investing (Wiley & Sons, October 2010).
October 30th, 2010 at 1:30 am
Hi Irene,
I appreciate this article very much. Enough that I attempted to verify the USD/CAD strategy above, but I’m not seeing a correlation between the 5-day price change and the CPI change. My data is below (apologies in advance if this doesn’t format well), and BPS (profit) is normalized to a CPI change of |.1|. What am I doing wrong here?
CPI Date CPI change Day 0 price Day 5 price Delta BPS
01/16/09 -0.7 1.23843 1.26607 0.02764 395
02/20/09 0.3 1.26368 1.24635 -0.01733 578
03/18/09 0.4 1.26146 1.23565 -0.02581 645
04/15/09 -0.1 1.21491 1.23284 0.01793 1793
05/15/09 0 1.17577 1.14858 -0.02719 0
06/17/09 0.1 1.14118 1.14526 0.00408 -408
07/15/09 0.7 1.12236 1.10445 -0.01791 256
08/14/09 0 1.08500 1.10948 0.02448 0
09/16/09 0.4 1.06987 1.07655 0.00668 -167
10/15/09 0.2 1.03337 1.03153 -0.00184 92
11/18/09 0.3 1.04680 1.05866 0.01186 -395
12/16/09 0.4 1.05989 1.05547 -0.00442 111
01/15/10 0.1 1.02630 1.04522 0.01892 -1892
02/19/10 0.2 1.05148 1.05603 0.00455 -228
03/18/10 0 1.01032 1.01908 0.00876 0
04/14/10 0.1 0.99677 1.01846 0.02169 -2169
05/19/10 -0.1 1.04621 1.05947 0.01326 1326
06/17/10 -0.2 1.02467 1.02243 -0.00224 -112
07/16/10 -0.1 1.04508 1.03644 -0.00864 -864
08/13/10 0.3 1.04141 1.02764 -0.01377 459
09/17/10 0.3 1.02680 1.02350 -0.00330 110
Average BPS: -22
Cumulative BPS: -471
Delta = Day 5 price - Day 0 price
BPS = Delta/CPI * -10000 [*-10000 normalizes to BPS profit when using the strategy above]
November 1st, 2010 at 10:41 am
Greg,
Thank you so much for your interest and attention to detail.
In quantitative finance, the impact of a particular event is most often assessed through a quantitative technique known as the regression, not correlation. With a regression, you are able to separate the effects of changes of, in this example, U.S. CPI, on the security prices from other factors that may also influence the security prices. The correlation procedure does not provide the same ability — correlation just throws all potential impacts into the same bag, leaving us unable to distinguish any event-specific impact.
Check out my book, “The Quant Investor’s Almanac 2011: The Roadmap to Investing” (Wiley, 2010) for more practical results of various quantitative studies.
Best Regards.
November 2nd, 2010 at 7:41 pm
Greg,
Having re-read your email, I’ve noticed a couple of potential issues:
1) it is not clear how you are defining a Day 0 price;
2) not sure what you are trying to get using correlation.
If you are looking to learn the underlying methodology of my analysis, the following paper by Kothari and Warner (2004) may be a good start: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=608601
Hope this helps.