Please Note: Blog posts are not selected, edited or screened by Seeking Alpha editors.

# F-Shift Forecaster probable market forecasts on IBM, RIMM and BA

In today's post, I am going to put the F-Shift Forecaster© to the test and post the probable ranges for 3 stocks that are widely traded. Call it an accountability check which in 12 trading days we will come back and review just how accurate or inaccurate, my proprietary approach to modeling future prices actually works. Before I do that, let's back test the F- Shift Forecaster's© accuracy with and without the weighting component. To understand this concept a bit clearer and the process that the F-Shift Forecaster applies prior to making a forecast you will need to revisit last week's blog post which can be found by clicking here http://fulcrumshifttra...

The forecaster applies 5 probable tests and they are as follows and are calculated on the forward looking 12th period. I say "period" because the platform allows for one to populate the platform with daily, weekly or monthly data:

• 1) The probability that price will be greater than (>) or less that (<) the current starting price
• 2) The probability that price will be greater than (>) 10% from the current starting price
• 3) The probability that price will be between 0 and 10% from the current starting price
• 4) The probability that price will be between 0 and -10% from the current starting price
• 5) The probability that price will be greater than (>) -10% from the current starting price

The significant difference between the F-Shift Forecaster's© approach and the typical price forecasting methodologies is in the way data is utilized. The F-Shift Forecaster© resample's ACTUAL historical changes in price data using a methodology called "bootstrapping" which effectively makes the assumption that the types of changes in closing prices over a defined look back period will eventually be similar in the future. Using the RANDbetween function within Excel, we can now run an unlimited number of scenarios as to the probable outcome "X" periods in advance. In the attached example that I am presenting here today "X" is represented in days. Twelve (12) days to be exact.  Allow me to explain why I chose 12 as the looking forward metric. Originally when this platform was designed, the data (which is downloaded from Yahoo finance) was monthly. Sixty (60) months or data points equates to 5 years of historical data. Once the platform has been populated with those data points, the next logistical step is to forecast, over the next year, (hence the twelve months/periods) where the stock has the highest probability of ending up. Now the platform can also model 12 weeks or 12 days into the future by simply changing the time frame of the data being input.

Below are the system back tested results as well as the future calls. Let's see how well the forecaster does and we will revisit these posted results 12 trading days from today, on May 20, 2009. In the meantime, if you wish to see any of my previous web tutorials or pics, you can view all of them via my blog at   http://www.fulcrumshif...  and then click any of the "screencast" links that can be found toward the end of each blog post.

Once a link has been clicked you will notice that all have been archived and if you look toward the top of the "screencast" screen, there are options to enlarge the video or pic and beside that are arrows allowing you to scroll though the entire library - both the videos and pictures. If you have any questions or suggestions, I am always eager to hear feedback, both positive and negative (I'm a big boy I can handle any and all constructive criticism!) and I can be reached by sending me an email via my website under the" contact" link, my direct email address at atlus1432@cogeco.ca or by leaving a comment directly on my blog or at the bottom of the videos or pics.

One final note - I have improved the F-Shift Forecaster both aesthetically and in its functionality.  In the looks department, I have re-shuffled the graphs around so that they are not cut off the screen. Prior to this change the user had to scroll down in order to view the complete picture. When running multiple scenarios the top 3 weightings are now colour coordinated so the user can see the impact of probable outcomes when adjustments to specific weightings are performed. This is bit confusing to grasp in print hence the reason for watching the latest 5 minute web tutorial which will visually illustrate what I mean and the improvement this feature has on the overall impact of the tool. I have also added some great macros which allow a user to "one click" for data population. The four (4) buttons now included are H.V. - Historical Volatility (for back testing purposes), F.V. - Future Volatility, H.P. Historical pricing (back testing on prices not volatility) and F.P. - Future pricing which forecasts probable outcomes over the next 12 periods. When any of the 4 options are selected the platform reflects that visually along the top so that you the end user know exactly which of the 4 outcomes you are looking at. Last but not least, I found a couple calculation errors that have now been corrected increasing the accuracy of the generated results and I won't bore you with the details on those.

As always, click the links below to open the visual of both the new and improved F-Shift Forecaster AND the 2nd link is the Forecaster results AND the future forecasts - enjoy

http://www.screencast....

http://www.screencast....