How may we be able to estimate the credibility of one currency?

When we abolish the gold standard, the values of modern currencies are based on the issuing countries' credibilities. Different countries obviously have different credibilities, at least the markets believe they have different credibilities. Some may be more credible than the others due to their different sizes of economies, stabilities of regimes, credibilities of central banks, and many other factors. When the currencies are backed by the countries with credibilities, same information or factor change could cause different influences on different currencies. For example, when the possibility of a hard Brexit is currently increasing, the British currency GBP depreciates against other currencies. Because different countries have different credibilities, their currencies are influenced by the news and appreciates against the British sterling to different extents. When they appreciate at different rates, the exchange rates among them also change. Therefore, we could use these cases to test the countries' different degrees of credibilities. We can set up a benchmark for the foreign exchange market and test every single currency's price change against the benchmark over time. The benchmark could be gold or USD. I prefer using gold, as no country directly influences on the gold price.

If we can have a wide time range that we can a large sample of as many data as possible, we can estimate how different the responses of different currencies to information. However, there is a problem: among the appreciating currencies, the more credible currencies should have greater appreciation rates than the less credible ones; on the other hands, among the depreciating currencies, the more credible currencies should have smaller depreciation rates than the less credible currencies. Therefore, when we should contain a dummy variable to state whether the currency is facing an appreciation pressure or a depreciation pressure so we could see that the currency's responses to appreciating pressure as well as depreciating pressure. Moreover, of course, there is some information that directly links to one country, so the impact on the country is definitely greater than other country; however, if we can have a sample that is large enough, such cases will be more likely to equally distribute among all countries, then the errors caused to the modelling system might be mitigated in general.