HAUGEN SYSTEMS PORTFOLIO BACKTESTS

In these backtests, we started with a universe consisting of the top 1,000 U.S. stocks (by market capitalization).

Starting with data from January 1996, we sorted based on the indicated methodology and took the top 50

stocks, which ranked highest under the methodology being tested. For the following month and for each month thereafter,

we rebalanced the portfolio by selling the four stocks with the lowest ranking. The four exiting position's value was

then calculated. We distributed the proceeds evenly among the four new stock picks, again which ranked

highest under the methodology being tested.

Once the backtest was complete, we linked the returns to get annualized results. Then we ran 11 more tests,

with each test beginning in a different month. For example, to see how the strategy would do if the test began in

February, we formed a portfolio of 50 stocks beginning in February, 1996; rebalanced four stocks each month, and

linked the returns.

The example worksheet shows how portfolios were compiled in the backtests.

The following two worksheets present the results from all 12 backtests for each methodology. In all the tests, except

for the ranking procedure, the exact same parameters were employed: the population consisted of the top 1,000 stocks,

50 stocks were held each month, and four stocks were rebalanced each month.

The cumulative worksheet graphically shows how four selected methodology's portfolios did over time.

Note: We have not made an attempt to replicate Greenblatt's strategy; for example our return on assets factor

does not include earnings before interest and taxes, but is rather the ratio of operating income to assets. In addition,

our earnings yield does not use enterprise value, but is rather the ratio of most recently reported 12 months earnings per

share to the current market price.

Differences in stock populations, numbers of stocks included in portfolios, ranking and rebalancing procedurecs betweeen

ours and those published in the Greenblatt book are due to differences in stock populations, ranking procedures, and time-

periods.