(5-engine Rating) Engine Ratings: ValuEngine's proprietary Engine Rating system provides an overall assessment of a stock's attractiveness. It combines valuation, risk-return tradeoff, momentum, market capitalization and forecasted future return data into an easy to understand and actionable format. Only two percent of our stock universe receives a 5-engine rating--which indicates a strong buy, while the worst stocks receive a 1-Engine--or strong sell. The tables above and below demonstrate how well the Engine Rating system and various ValuEngine portfolio strategies perform and indicate that the rating system is predictive and the portfolio strategies beat their benchmark indices.

Click Here for more Engine Rating back test results.

In addition to the back-test results you see here, you can also examine our live-tracked performance numbers as well as the current holdings for each ValuEngine Benchmark Portfolio. You can find that information in addition to a discussion of our benchmark portfolio strategies in our Strategy Library or click HERE to download VE Portfolio Historical Performance Study released on Sept 2, 2009 (pdf, 461K).

The ValuEngine Research Findings page explores the following research topics:

• Test Methodology
• Findings on Investment Strategies and the VE Stock Valuation Model
• Results for the VE Forecasting Models

ValuEngine currently conducts back tests and performance tracking on US stocks for the period January, 1990 to the present. To maintain the integrity of our research and development process, all of our back tests are conducted under rigorous guidelines designed to eliminate the most common problems that arise when attempting to vet a strategy via the use of historical market data. ValuEngine's back tests eliminate any and all survivorship, forward-looking, and data-snooping biases. This means that while the tests are conducted in the present, the buy/sell decisions are based solely on data that was contemporary to the time period as well as a historically accurate stock universe.

This "out-of-sample" decision rule applies to every investment strategy and/or model we test. For example, when we test our book/market ratio strategy for March, 1988, only the book/market ratio of each stock in March 1988 is used to determine which stocks should be bought or sold during that month. Similarly, strategies based on the return forecasting models for March 1990 used only the data available up to that month to estimate the forecasting parameters applied to predict the post-March 1990 returns.

In other words, no information that later became known is utilized by our models. This rigorous test methodology allows greater confidence in our strategies and helps to insure that they perform in a variety of market conditions.

<Top>

The following investment strategies were compared against each other via back testing studies:

• VE Standard 20 -- Long Strategy
• VE Forecast 22 -- Long Strategy
• VE Top Rated 20 -- Long Strategy

A description of the investment objectives in the above strategies can be found at our VE Benchmark Portfolios page.

Results for the VE Forecasting Models

To demonstrate the significance of our forecasting models, we ran a separate set of back-tests using the same stock universe employed to test the above investment strategies. Specifically, for each past month in the study, we first estimated the forecasting coefficients for every model using only data available up to that month. In step 2, we made forecasts for each stock's one-month-forward return. In step 3, we divided all the stocks of that month into 10 decile groups (i.e., 10 groups of equal number). With an equal dollar amount invested in every stock included in a given decile group, we obtained 10 equally-weighted decile portfolios for the month going forward. In step 4, we computed the actual returns for each of the decile portfolios over the immediate month. Then, we repeated these steps for all the months in the study.

The chart below shows the average annualized returns for the best and the worst forecasted-return portfolios. It clearly shows that our forecasting model, like our engine rating system, is predictive and performs remarkably well.

The following chart also demonstrates the effectiveness of our forecast model. As you can see, the two different portfolios perform as predicted with the top-ranked portfolio providing major increases in performance when compared to the bottom-ranked portfolio and the overall stock universe.

<Top>