Forecasts

Forecasts reflect assumptions about assets' return distributions, derived from other sources than your historical data. They may come from current data, i. e. yield information for bonds, or from more generally held principles of finance. The forecasts described on this page are generally single-asset forecasts. For forecasts relating multiple assets to each other, use forecast contrasts.  Forecasts must include return and risk projections, but correlations are optional.  Alternately, a previously saved risk-return set can be loaded to the Forecasts Worksheet.  

Forecasts are combined with the historical data analysis when the Run Bayes button in the Estimator ribbon is clicked. The result combines the two signals with relative strengths determined by the standard deviations and correlations. Larger standard deviations indicate less certainty, and thus less strong of an influence in the result. Ignoring correlations for the moment, a forecast with an equal standard deviation to the historical standard deviation will produce a result mean (expected return) halfway between the historical return and the forecast return for that asset.

• Forecast means are your estimates for how a particular asset class will perform in the next year.

• The standard deviation of forecast is a subjective estimate of the uncertainty of the user's return forecast, indicating the strength of the forecast.  If you are uncertain about a particular return estimate, enter a high percentage, maybe even as high as 50%.  The standard deviation dictates whether the return observations or your forecasts dominate the results.  

• • One rule of thumb is to think of the forecast return as x% +/- y, where x is the return and y is the standard deviation.  Therefore, a forecast return of 10% with a standard deviation of 5% implies a forecast range of 5-15%, while a standard deviation of 20% implies a forecast range of -10% to 30%.  A 5% standard deviation indicates a more precise forecast and has more influence on the final result.  

  • The historic mean and standard deviation appear for reference.  Another rule of thumb to consider is that if the standard deviation of the forecast is a smaller number than the historic standard deviation, the result will be closer to the forecast since its certainty is greater.

  • Standard deviations must always be positive.  

  • Entering zero, a negative value, or a non-numerical value for the standard deviation causes the cell to be replaced with the text “N/A".  Depending on what is selected in the Bayes Forecasting drop down menu, a "N/A" in the Std Dev of Forecast column causes an error (Forecasts), nothing (Historic), or for the specific asset to be unaffected by running Bayes estimation.  Note that turning the prior off for an asset is not the same as increasing the standard deviation to a very large number, because in the latter case the mean estimates will still be influenced by the correlations.  Turning the Bayesian functionality off for an asset has the additional property of setting all (forecast) correlations for that asset to zero.

  • The Result Mean Column only populates after running Bayes.  This column appears for your reference as you adjust forecasts.

• Select the source of the correlation matrix used during estimation from the Forecast Correlations drop down menu.  

• • You may have little reliable forecast information on the correlation of assets in the future.  In that case, select the Use Historic Option to use the historic correlations.  The correlations on the Forecasts Worksheet remain grayed out to indicate that the matrix on the Historical Worksheet will be used.  This is generally recommended.

  • Selecting the Use Forecast Option causes a blank correlation matrix to appear.  Complete the matrix with your correlation projections.  Alternately, copy the historic correlations to the Forecasts Worksheet using Excel copy and paste. Use caution in manually creating forecast matrices as it is easy to create matrices that are not positive definite, i. e. are not allowable correlation matrices.

• A more flexible and advanced tool for entering relative information comparing or combining multiple assets is forecast contrasts.  Choose whether or not to display contrast fields in the Display Menu.

• To combine forecasts with the historical analysis in the results, click the Run Bayes button.

The input forecasts are used according to your further selections on the Forecasts Worksheet.