Exchange rate forecasts - We are not getting any smarter, use them at your own peril

Analysts have learned a lot about currency markets over the last few decades, yet exchange rate forecasters, on average, still cannot do better than the random walk model. Recent research which uses perhaps the most extensive database on forecast across currencies and over time has not found any improvement versus older survey and forecasting analysis. See "Analysts Forecasts and Currency Markets" by Florian Mair 

The Meese and Rogoff forecast puzzle continues. Macro fundamentals do not seem to help forecast exchange rates. The same hold for the Fama puzzle concerning the poor forecasting power of forward rates. We can talk about rational expectations, but the numbers still suggest biases. 

There are a number of possible explanations for these results and some research is in conflict, but the overall conclusion is that exchange rate forecasting is a game with few winners. Forecasts are negatively related to excess returns; however, the author finds that forecast dispersion is positively related to currency returns when added to a carry trading strategy. These forecasts also underperform for long/short portfolios relative to factor forecasts associated with carry, value and momentum. You can use these classic factor models as better alternatives to following the experts although the level of significance is not strong for differentiating between strategies.

The numbers on skill have not changed. You are not getting any value following the forecasts of other; however, if analysts are confused as measured by dispersion you are actually obtaining useful information.