The Combination of Macroeconomic Indicators as a Signal in the Forex Market
29 June 2022In earlier studies, we have considered the filtering of "Bollinger Bands Reversals" and “Head and shoulders” technical analysis signals using the macroeconomic background (MB). The effectiveness of such a filter has been revealed: the average rate of return with the MB filter has turned out to be higher than without it.
Today we will look at the work of the macroeconomic background filter in its clean form, without reference to technical analysis signals.
The market entry strategy based on the macroeconomic background filter is profitable.
8 economies of the world:
- The USA
- Canada
- The Eurozone
- The UK
- Switzerland
- Japan
- Australia
- New Zealand
Instruments: 25 currency pairs
Timeframe: D1
Period: January 2008 – February 2022
There are 5,052 signals in total
25 currency pairs participate in testing:
AUD | NZD | JPY | EUR | GBP | CHF | CAD | USD | |
AUD | ||||||||
NZD | 1 | |||||||
JPY | 2 | 8 | ||||||
EUR | 3 | 9 | 13 | |||||
GBP | 4 | 14 | 17 | |||||
CHF | 5 | 10 | 18 | 21 | ||||
CAD | 6 | 11 | 15 | 19 | 22 | |||
USD | 7 | 12 | 16 | 20 | 23 | 24 | 25 |
As an MB filter, we will take seven indicators for each of the 8 economies. A detailed list of indicators is presented in the appendix.
Recall that the MB filter "turns on" in the case when 8 evaluations out of 14 of macroeconomic reporting are unidirectional.
In the situation of 7 evaluations out of 14 - we refrain from transactions.
Example
The indicators of the macroeconomic background at the moment for the Eurozone are as follows:
Interest rate, GDP, Trade Balance, Business confidence, Inflation and Unemployment rate have risen while Retail sales have declined.
The indicators of the macroeconomic background at the moment for the USA are as follows:
Interest rate, GDP, Unemployment rate and Trade Balance have declined, and Business confidence, Inflation and Retail sales have risen.
MB Indicator | Base currency EUR | Quoted currency USD |
Interest rate | buying | buying |
GDP | buying | buying |
Unemployment rate | selling | selling |
Retail Sales | selling | selling |
Trade Balance | buying | buying |
Inflation | buying | selling |
Business confidence | buying | selling |
In this case, 8 out of 14 MB filter evaluations give a signal to buy the EURUSD pair. So, we buy a pair of EURUSD.
The "sentiment" of the filter will be determined by comparing:
- new actual values of MB indicators with the previous ones (fact/previous);
- new actual values of MB indicators with the forecast ones (fact/forecast).
Let's denote the "work" of the MB filter:
- F8 – 8 out of 14 evaluations to open a position;
- F9 – 9 out of 14 evaluations to open a position;
- F10 – 10 out of 14 evaluations to open a position;
- F11 – 11 out of 14 evaluations to open a position;
- F12 – 12 out of 14 evaluations to open a position;
- F13 – 13 out of 14 evaluations to open a position;
- F14 – 14 out of 14 evaluations to open a position.
Position opening:
- On the Open the next day after the appearance of the MB filter signal (entry when the signal appears);
- On the Open every day during the existing MB filter signal, starting from the next day after the appearance of the MB filter signal (daily entries on signal).
Closing a position:
- On the Close of the 1st day after entering the market;
- On the Close of the 5th day after entering the market;
- On the Close of the 10th day after entering the market;
- On Close on the 15th day after entering the market;
- On the Close of the day, the “sentiment” of the MB filter changes. For example, when instead of 8 evaluations of the MB filter to open a position, there are 7 of them, or instead of 9 evaluations, it becomes 8, 7 or 6.
The results are presented in diagrams:
Filter work when comparing of new actual values of MB indicators with previous ones
Filter work when comparing of new actual values of MB indicators with forecast ones
So, the macroeconomic background filter in its clean form, without reference to technical signals, is ineffective, since the average rate of return does not reach a significant value of 0.3%.
Next, let's take a closer look at currency pairs with the US dollar. And let's see how the overall results of the MB filter will change.
The US economic statistics, unlike many other countries, is updated on a variety of indicators of different categories.
Therefore, in addition to the basic indicators for each of the US categories discussed above, namely:
Category | Indicator |
Interest rates | |
GDP and industrial production | |
Business confidence index | |
Employment and unemployment (labor market) | |
Inflation | |
Trade balance | |
Retail sales |
we will consider the variant of the MB filter with several indicators in categories, each of which is assigned a certain weight:
Category | Indicator | Weight |
Interest rates | 1 | |
GDP and industrial production | 1 | |
Employment and unemployment (labor market) | 0.4 | |
0.3 | ||
0.3 | ||
Retail sales | 0.25 | |
0.25 | ||
0.15 | ||
0.15 | ||
0.2 | ||
Trade balance | 1 | |
Inflation | 0.4 | |
0.3 | ||
0.3 | ||
Business confidence index | 1 |
The results are presented in diagrams:
Filter work when comparing of new actual values of MB indicators with previous ones
Filter work when comparing of new actual values of MB indicators with forecast ones
Once again, the clean macroeconomic background filter is ineffective, since the average rate of return does not reach the significant threshold of 0.3%.
The macroeconomic background filter in its clean form, without reference to technical signals, is ineffective.
The effectiveness of the macroeconomic background filter in its clean form in market forecasting has not been revealed.
Detailed results are shown in the Appendix.
See also:
Comments
4