To read Advisor Perspectives’ analysis of FundGrades, see the accompanying article in this issue.

We had the opportunity to speak with David Loeper, Financeware’s CEO on November 28, 2007 about FundGrades.

Background

Loeper has been using the methodology behind FundGrades to support Financeware and his advisory services business, since he founded the company in 1999.  He decided to release FundGrades to serve as a tool for advisors and sophisticated investors, with an additional goal of helping advisors that serve in fiduciary roles and are subject to ERISA guidelines.  ERISA requires diversification which, Loeper argues, contradicts how others grade funds.  While FundGrades can reward funds for making big winning bets through concentrated positions in three of the five grading criteria (return, risk, or risk of material underperformance), such bets will likely be exposed in a low diversification grade.  ERISA also requires reasonable expenses, based on the product offered, and FundGrades provides objective grades of expenses based on related offerings.  Lastly, ERISA requires that “fiduciaries should avoid any investment risk unless it is prudent to do otherwise.”  Loeper expands on this by positing that traditional measurements of risk (i.e., standard deviation and volatility) are useful metrics, but do not tell the complete story behind the risk levels of mutual funds.  To meet ERISA standards, Loeper has created a proprietary risk metric, which he calls the Risk of Material Underperformance, which shows the extent to which a fund’s performance is the result of a few good months or a steady, consistent history of results.

“The problem with peer groups is that there is widespread mis-classification of funds, partly because relative fund performance can be improved by putting funds into lower performing peer groups, survivor bias and many other well known problems” says Loeper, adding that the CFA Institute has published statements cautioning CFAs about the problems in using peer groups.  FundGrades provides an alternative framework for categorizing funds.  Using three years of historical data, each fund is compared to a series of over 30 asset class benchmarks, and a correlation coefficient is computed for each benchmark.  Funds are ranked by their correlation coefficients, from the “best fit” to the “worst fit” benchmark, making it very easy for advisors to see where a fund belongs, and FundGrades also uses macro holdings data to suggest a “best fit.”

Loeper likes to use the example of the Fidelity Contra Fund (FCNTX), which Morningstar categorizes as a large cap growth fund.  But FundGrades shows there are actually 13 other asset classes FCNTX resembles more closely.  “You can grade it against large cap growth, but you can find a better asset class,” says Loeper.   FundGrades’ categorizations show the best fit for a fund, and do not suggest whether a fund is better than other funds.

Five Criteria for Grading a Fund

The Fund Grades web site provides documentation on their methodology, which we summarize here.  FundGrades displays a list of asset classes for a given fund, ranked in decreasing order based on the correlation coefficient to the asset class benchmark.  If a fund has a correlation coefficient of less than .82 against all benchmarks resulting in a diversification grade of F, it is not assigned an overall grade  – since it does not behave like any asset class tracked by FundGrades.  Advisors select an asset class for analysis, and the results can vary significantly depending on the asset class chosen.  FundGrades looks at five criteria for a fund, and assigns a grade to each, ranging from A+ to F.  Each of the five components of the FundGrades grades has an honor roll – those funds that meet the highest standards.

• Diversification – Greater diversification earns higher grades, as measured by the correlation coefficient against the asset class benchmark.  As one would expect, Index funds dominate the honor roll for the diversification criterion because they have near perfect correlation to appropriately selected benchmarks.

• Expense – Lower expenses earn higher expense grades.  FundGrades assigns each fund to an asset class based on best fit and holdings, and determines the average expense ratio for the lowest 10% of funds based on expense ratio.  Grades are assigned relative to this average expense ratio.  An additional adjustment is made to lower the grade if the fund has front-end or deferred sales charges.  As with diversification, lower cost index funds generally will earn the highest expense grades and dominate the honor roll of expense grades. 

• Relative Risk – Funds with lower volatility relative to a benchmark earn higher grades.  Specifically, the volatility of each benchmark is measured.  If a fund’s volatility tracks within +/- 1% of the benchmark’s volatility, it earns an average (C) grade.  Lower volatility earns higher grades and vice versa.  Unlike diversification and expense grades, index funds do not appear on the honor roll for relative risk; they earn average grades, as would be expected.  In fact, many actively managed funds earn high risk grades because they hold cash, which reduces their volatility.

• Return – Funds that exhibit higher returns, relative to the return and volatility of the benchmark, earn higher grades.  For example, if the benchmark for an asset class earned 20% with 10% volatility, a grade of C would be assigned to any fund earning between 19% and 21%.  A grade of C implies a return range of +/- 10% of the volatility of the benchmark (10% of 10% is 1%).  Generally, about 10% of the funds make the return honor roll.  As the volatility of the benchmark increases, the hurdle necessary for higher return grades increases.  Index funds earn average grades and essentially have no chance of making the return honor roll.

• Risk of Material Underperformance – Funds exhibiting a more consistent history of returns earn higher grades.  This is FundGrades’ unique and proprietary metric, and is designed to convey additional information about the history of a fund’s returns.  Funds that exhibit high returns because of superior performance over relatively short periods of time are penalized relative to funds that exhibit a consistent pattern of returns, as measured by the percentage of periods over the measurement interval that the fund outperformed the benchmark, after adjusting for expenses.

These five grading components are averaged (with equal weightings) to provide an overall grade for the fund.  Again, as with the individual criteria, the overall grade will depend on the asset class chosen; a fund may earn a high grade in one asset class but a low grade in another. 

Interpreting the Data

Our first question for Loeper was whether FundGrades offers any predictive ability vis-à-vis future risk-adjusted returns.  Loeper argues this is not the right question to ask when evaluating funds in a retirement account.  “What matters is wealth, the actual dollars accumulated or spent, and that is not necessarily correlated to return or risk,” says Loeper, adding that “if two funds deliver the same compound returns but with different risks, it may not matter when it comes to wealth accumulation.”  Loeper believes that advisors should be focused on managing life relative wealth instead of focusing on market relative returns. This is the basic difference between dollar weighted returns versus time-weighted returns (that may or may not be tied to wealth), he says. “ He continued, “For someone dependent on their portfolio for their retirement lifestyle, looking at risk adjusted time-weighted returns is like trying to argue that you won a football game based on the number of third down conversions. In the end, what matters is the points on the board or, in investing, the dollars you can spend.”

Loeper is very skeptical when it comes to using historical data to predict future performance, noting that “uncertainty is, well…certain.”  He offered some results from studies his team conducted with FundGrades.  Of the top 5% of funds at the end of March 2003 (based on return grades), there was a 54% chance that these funds would have been below average (C- or worse) based on return for the three years ending March 2007.  In fact, this same 54% probability held true across all funds, leading Loeper to conclude that historical returns offer no statistically significant ability to predict future returns (at least over these two three year periods and several others they have tested as well).

Loeper looked at the 5% of funds that would have made the overall honor roll during the same time period, and found that 66% had a return grade of C or higher, about 50% better than for all funds.  Loeper cautions this does not imply that the overall honor roll has predictive ability for returns, only that it has more statistical significance than looking at pure returns, or risk adjusted returns.

Lastly, Loeper found that if a fund was in the overall honor roll its chance of earning a terrible (F) grade for returns was cut in half (relative to the chance of any fund earning an F grade).

Future Plans for FundGrades

Loeper discussed a number of plans for enhancing FundGrades.  First, retrospective grades will be provided, so that advisors can see what a fund’s grade would have been at a given time in the past.  This will allow advisors to gain confidence in the methodology, by seeing how a fund would have been rated prior to a significant event.  For example, Bill Miller’s Legg Mason Value Trust would have earned an overall grade of D- in 2003, just prior to when its 17-year record of beating the S&P 500 ended.  The fund’s diversification and risk of material underperformance grades were alarmingly low, relative to the S&P, due to concentrated positions.  Loeper notes that Miller argued that the S&P was not the relevant benchmark for the fund, but FundGrades’ versatility allows the fund to be compared against other benchmarks.  “The fund had a lot of things going against it – such as expenses and risk,” notes Loeper, adding that “this would have showed up whether it was compared to large cap growth, large cap core, or anything else that was reasonable.”

The second plan for FundGrades is to allow advisors to view the grades of a portfolio of funds.  This is partly in response to advisors that argue they use particular funds, even though they might have poor grades along certain dimensions, because they work well in an overall portfolio.  Loeper predicts that advisors will be surprised to see the results, and looks forward to providing this capability.

Loeper’s believes the ultimate value of FundGrades is in building portfolios based on asset allocation models, and may provide features specifically to address this need.  He cites the example of an advisor that has a portfolio consisting of large cap core equities, taxable bonds, and cash.  If the advisor is looking to replace the taxable bond funds with muni bond funds, he or she will want to know whether the diversification or risk is being compromised.  Such analysis is easily performed with FundGrades, and diversification and risk can be examined independently.   Loeper notes that “everyone else is trying to explain what a given fund is; we are starting with the asset allocation pie and trying to find the best fund for each slice.”

Although the FundGrades methodology rewards index funds in certain criteria (e.g., diversification and expense), this is not the intent.  The intent is to respect fiduciary standards.  “A hole exists, where nobody is warning advisors when funds are placing big bets,” says Loeper.  Loeper also adds that index fund investors will see that different index funds offer different fits against a benchmark, and FundGrades allows advisors to easily see which fund offers closest correlation.

Loeper does not treat FundGrades as a profit center.  He will not allow fund companies to advertise (although many have asked), and will not charge usage or subscription fees.  He is not positioning FundGrades to replace or compete with any other system, and sees it as a complementary offering in the market.  An open architecture allows other web sites to easily link to FundGrades pages, and FundGrades offers hyperlinks to other vendors’ fund ratings and information.  “Thousands of advisors have started using it in one form or another” notes Loeper, “and our success will be measured in the improved wealth accumulation in today’s retirement accounts.”

Display article as PDF for printing.

Would you like to send this article to a friend?