A process for selecting investments from a population of book-valued collective funds uses a determination of the past average investment performance of funds within an asset class. Past performance is measured in a "means-variance" analysis. The selection process determines anomalies from a theoretical population distribution that is uniformly random with a normal distribution around the measured past performance averages for the asset class. Investment return is calculated as the average of past periodic returns. Investment risk can be the variance of those past returns around their average, or the covariance of those past returns with those of a benchmark that can be the population average of past periodic returns or past periodic returns of an associated market index. Variances are identified preferably by dividing or scaling the performance distribution into four to twenty-five areas of what should be equal-sized fund populations assuming a normal distribution of risk and return about the center of each for the asset class population. The actual population in each equal area is measured, and these measurements are ranked. Investment selection is for those funds found in areas having a high ranking. The selected group exhibits a performance significantly stronger than the "asset class" average performance for at least thirty-six months.