Return

As we mentioned in addressing relative risk, PAST PERFORMANCE IS NOT AN INDICATION OF FUTURE RESULTS, so why would one consider it?

From our grading routine we have other elements that compensate for this reality and so we can include past return (and risk) as criteria to expose what less sophisticated grading/ranking systems do not.

In three studies we performed, testing our methodology covering various three-year time periods, we discovered that 53-55% of all funds materially underperformed their best fit benchmark. Interestingly, this is about the same advantage casinos have over you in roulette. It isn't surprising because expenses within funds obviously make the odds of beating the benchmark a difficult hurdle to pass. "Materially underperformed" as used in FUNDGRA+DES® is measured by having a return which is less than the benchmark return by more than 10% of the standard deviation of the benchmark. This is important when you are comparing returns to the asset class in a strategic asset allocation for which you are selecting a fund instead of just a relative and somewhat subjective "peer group" rank as used in other systems.

Returns are obviously going to move around from negative to positive depending on the time period being measured. The volatility and return results are both going to expand and contract over time. Our method of ranking returns needed to be objective, had to avoid peer rankings so that investors could examine a fund for how it fit into their strategic asset allocation, and also needed to consider how variance would simultaneously expand and contract, sometimes in the opposite direction of returns. Our approach dynamically takes all of these factors into consideration.

For example, say the benchmark return over the time period being measured was 20%, a very strong market, yet the volatility during this time period was only 10%, a relatively low volatility for that high a rate of return. A grade of C would be assigned to a fund that returned between 19%-21% for this class, during this period. This range is dynamically calculated by taking the grading rule of C equaling a benchmark return within 10% of the volatility of the benchmark (in this case, 10% of 10%, or 1%) and establishing this as a plus or minus to the actual benchmark return. A more volatile market that had a 20% standard deviation would provide a grade of C if the return were between 18% and 22% (10% +/- tolerance relative to the market for a C grade, times 20% standard deviation for the market, equals +/- 2% relative to a 20% market benchmark return).

This methodology, while a bit complex to calculate dynamically, constantly and consistently can be used across all asset classes and market environments without needing to cross the line to use potentially misleading "peer group" rankings. Less volatility makes for tighter grades. Lower return markets make the volatility tolerance more sensitive. All of the returns are always relative to the benchmark, and to get a grade of better than C, the fund must beat the benchmark. The amount one must beat it by to score the top grade of A+ (about 2% of the graded funds) varies based on the volatility of the various markets.

With a return grading system as dynamic as this, you would think we would want to weight it more heavily in the overall grading criteria, however, as we stated before, past performance is not an indication of future results.

We did a study of the top 5% return grades for the three years ending in the first quarter of 2003 and compared those top rated results (by return) to all funds as of the end of the first quarter of 2007. It made no statistical difference whether the fund was a top performer in the past or not, 54% of the prior top performers materially underperformed their benchmark four years later, just as 54% of all funds had. This is why you cannot depend on grading/ratings systems based on returns to show you anything predictive.