My most recent effort was to try and produce a very long term forecast. I wanted to see if I could generate a 10-year view on currencies from the various scenarios that have occurred, and led to the present FX valuations.
The concept I used was that it is reasonable to assume that varying time intervals represent different scenarios. For example, the last 4-5 years equate to what many call the “new norm” following the Lehman’s collapse; the last 8-10 years gives the boom and bust in many carry trade currencies; by going further back in time, we bring in the effects of pre-dotcom equity bubble, East Europe break up, oil crises etc. By taking such individual scenarios in turn, I carried out a large number of forward simulations – by large I mean; 100 trials per day x 2500 days ahead (a trading year roughly consists of 250 trading days) and repeat the process a further 100 times. This gives a grand total of 25, 000, 000 simulations per currency per scenario. I used 5 input scenarios, so the value of each currency pair will be ‘guessed’, using different assumptions, a total of 125 million times over a period of 10 years.
Clearly, processing such large volume of data needs careful considerations and the outputs must remain simple enough to understand – the first output is the probability of higher versus lower values; the second is the mean value; and the third (and perhaps the most important) output is the probabilities of the central forecast range, i.e. the relative likelihood of a currency moving by, say +10% and -10% about its present value.
As an example, I am attaching a 60-day forecast plot of Gbp/Usd to demonstrate how forecasts propagate over a short time of 3 months.
Figure 1: shows how the Pound's value versus the US Dollar might propagate over the next 60 days (3 months) from 08 February 2012. This plot shows that the uncertainty in Pound's value grows as we go further away in time from its present value.
The main output from my simulation software is available in the format presented below:
FX pair = | GBP/Usd |
Forecast date = | 08 / Feb / 2012 |
Forecast validity = | 60 days (3 months) |
Current value = | 1.5918 |
Future (average) value range = | 1.580 to 1.600 |
+/- 5% swing in value covers: | 75% of the forecast range. |
Probability of FX going up (over the full forecast) range# = | 46% to 50% |
Probability of FX going up (over +/- 5% of forecast) range# = | -7.11% to +0.23% |
# values below 50% or negative percentages indicate a greater chance of the FX pair falling. | |
Unfortunately such long term forecasts are expensive and not suitable for individuals. These forecasts are primarily aimed at company (or a corporation’s) Treasury Risk Management units.
Contact the author for pricing information.
Source | The Currency Forecasting Blog (CFB) | http://currency-forecasting.blogspot.com/2012/02/long-term-forecasts-ten-years-ahead.html
No comments:
Post a Comment