Bootstrap simulation is a simple technique that estimates the reliability or accuracy of forecast statistics or other sample raw data. It can be used to answer a lot of confi-dence and precision-based questions in simulation. For instance, if an identical model (with identical assumptions and forecasts but without any random seeds) is run by 100 different people, the results will clearly be slightly different. The question is, if we collected all the statistics from these 100 people, how would the mean be distributed, or the median, or the skewness, or excess kurtosis? Suppose one person has a mean value of say, 1.50 and another 1.52. Are these two values statistically significantly different from one another, or are they statistically similar and the slight difference is due entirely to random chance? What about 1.53? So, how far is far enough to say that the values are statistically different? In addition, if a model s resulting skewness is ?0.19, is this forecast distribution negatively skewed, or is it statistically close enough to zero to state that this distribution is symmetrical and not skewed? Thus, if we bootstrapped this forecast 100 times, that is, ran a 1,000-trial simulation for 100 times and collected the 100 skewness coefficients, the skewness distribution would indicate how far zero is away from ? 0.19. If the 90% confidence on the bootstrapped skewness distribution contains the value zero, then we can state that on a 90% confidence level, this distribution is symmetrical and...