**EBI - Risk Assessment**

**Mean annual return**

As a measure of risk I estimated the probability distribution of the mean annual return obtained with EBI rules by resampling observations from the group of countries ranked first (the "Top 1" group). More specifically, a new sample was created by randomly replacing a country from the "Top 1" group by other country from the same group and for each new sample a mean annual return was calculated. This operation was repeated 50 thousand times. The distributions were estimated for the retrospective (assuming the "unlucky" scenario) and prospective samples separately.

The mean annual return in the retrospective sample is 34.7% (95% Confidence Interval: 12% - 57%) . The mean annual return in the prospective sample is 32.5% (95% CI: 11% - 55%). Although the distribution of the prospective sample is relatively skewed to the left, there is not a fundamental difference between them.

**Annual return and probability of loss**

A slightly different question refers to the range of the return that could be obtained in a year. Bootstrapping is used again to select four quarterly returns from the retrospective and prospective periods separately, and these are used to generate annual returns.

The estimates show that with retrospective data the annual return is above -14% with a 95% probability. More importantly, the returns obtained in the prospective sample indicate that with a 95% probability the annual return will be above -5%. The probability of having a positive return is 92.2%. In other words, the probability of suffering a loss in any year is 7.8%.

The risk analysis will be updated as more data come out of the prospective phase.

The risk analysis will be updated as more data come out of the prospective phase.