PARADIGM's highly theoretical and statistical approach to index construction creates, in our view, the most accurate set of hedge fund indexes and sub-indexes available. PARADIGM is the industry pioneer in the identification of four major hedge fund data biases. These biases must be corrected to obtain meaningful data about hedge fund performance.

An unfiltered hedge fund benchmark shows that the passive rate of return is an unrealistic 20-25% per year. However, taking into account the four biases (listed below) and correcting the data for upward skewing, we have determined that investors should expect the broad index of hedge funds to generate about 10.5% per year with a standard deviation of 6.6%. PARADIGM's mandate is to outperform this passive benchmark with lower volatility.

Survivorship Bias:

Hedge funds report their monthly rates of return to various databases voluntarily. Since hedge funds are generally exempt from registration, there are no publicly available sources where this data can be easily retrieved. Unless it is collected real-time, this data is almost impossible to obtain. This is particularly true for managers who have suspended or ceased trading. PARADIGM became aware of this problem and has been carefully collecting and preserving the data of defunct managers since 1992. PARADIGM has the world's largest database of defunct managers, which enables us to construct accurate historical hedge fund benchmarks that not only tell us what the real annualized returns and risks were but also, and more importantly, how these returns and risks moved over time.

Because performance databases often do not include the performance histories of hedge funds that have gone out of business (usually as a result of bad performance), the absence of this performance generates a database and subsequent benchmarks that are too good to be true. According to our research, this bias alone can account for more than 500 basis points of exaggerated annualized returns.

Self Selection Bias:

Why is the first year of a hedge fund manager's performance track record so often his best year? According to PARADIGM's research, and contrary to conventional wisdom, this phenomenon has nothing to do with the size of the assets under management. Many managers generate their initial returns through luck and self-select themselves as hedge fund managers and begin reporting to the databases. Since luck does not persist, their performance eventually will decline. Investors rarely participate in these luck-based returns because they only become involved with these apparently successful traders after the fact. Because these returns will not persist, they should not be included in any hedge fund benchmark. PARADIGM's statistical analysis estimates and eliminates this bias of about 200 to 300 basis points.

Liquidation Bias:

Global information shocks such as the bond market crash of 1994, the Asian crash of 1997 and the Russian default of 1998 often lead to hedge fund disasters like the Long Term Capital Management fiasco. Hedge funds that go out of business due to market-related shocks have no incentive to report their final monthly return. PARADIGM estimates this bias to be about 50 basis points on an annualized basis. This bias, however, is concentrated around certain global crisis events and can dramatically change the shape of the benchmark at these moments. Any statistical analysis using a benchmark that has not corrected for this bias would be fruitless.

Bull Market Bias:

Most hedge funds came into existence in the 90's and have not operated outside the recent bull market. Since many so-called hedge funds are actually overpriced closet mutual funds, their presence in the databases of hedge funds will necessarily skew the data upward. Evidence of this bias is the large correlation and beta that hedge fund indexes have with the S&P 500. According to PARADIGM's model and definition of hedge funds, a portfolio of hedge funds should have no correlation or beta with any long-only index. Consequently, PARADIGM's actual 14-year track record has 0.0 correlation with the S&P 500 and the Lehman Aggregate Bond Index.

Past performance is not necessarily indicative of future results.