A remarkable and unique technical indicator is built into FDC, and requires a bit of explanation. Everyone is familiar with trend channels, which can be used with any data series. What is not widely recognized, however, is that if an oscillator of a data series (say the difference of two moving averages or the daily difference of a single moving average with itself) is in a trend channel, then the original data series must be in a so-called parabolic channel – a channel whose top and bottom curves are parabolas (the mid cross sectional curves of satellite dishes). If a price series, say of a stock or commodity, has an oscillator that is in a linear (ordinary) channel, then that price series must remain in a parabolic channel as long as the oscillator remains in its linear channel. Such parabolic channels abound in the charts, but remain hidden from the view of most observers. They can be used to show, for example, that a top or bottom must occur at a certain price, unless the oscillator makes an exaggerated move up or down. If such a move has low probability for the oscillator, the predicted top or bottom of the price series is very likely.
The
parabolic prediction tool occurs on every command window, and is shown below,
in highlighted form.
That
tool opens a setup window that looks as follows:
In the above window you enter the name
of a Dataset to predict, or even an FDC expression that produces such a Dataset. The Dataset must have exactly one column, or
it must have three columns (representing High, Low and Closing price data)
or four columns (representing Open, High, Low and Closing price data). There are several choices to be made with regard
to your prediction. We have entered
default choices, but you are encouraged to experiment.
- Do you want the prediction to be based on closing data or midrange data (in the case of a 3- or 4-column Dataset)?
- Will your charted prediction be a single line or a channel showing volatility?
- You have the choice of calculating the prediction automatically or manually (called average approximation). These choices will be explained later.
- How many periods ahead do you want to forecast?
- Will
your oscillator be the difference of two moving averages, the difference of
a single moving average (any length you choose), or a weighted oscillator
of your choosing?
Click on the Begin Prediction button and a plot window with two panes is produced, showing the data to be predicted in the top pane, and an oscillator at the bottom. An example, showing the Yen/$ price in late 2001 is shown below:
You will note that a portion of the bottom pane is highlighted. This is accomplished by clicking on the chart with the mouse and moving it to select a time period on which to base your prediction, then clicking again. For this example we have chosen the “automatic mode”. The user then presses the Parabola tool:
In automatic mode, FDC computes a
least squares regression line to best fit the oscillator over the chosen time
period. That best-fit line on the bottom pane then
determines the corresponding best-fit in the upper pane for the same time
range. Both best-fit lines (linear
in the bottom pane, and parabolic in the top pane) are then extended into
the future (by using their formulae) to become your forecast.
As long as the oscillator can be
described or defined by its regression line, the price series must be described
by its own fitted line.
We now show you the actual market behavior following the prediction. As we had chosen to forecast a channel, you will see two dotted lines essentially defining both the forecast and the volatility.
Just to show you that forecasting is not as difficult as some might think, we show you the same market a little later in time, using the same type of analysis. Our point is that the Parabolic Predictor provides all market watchers with the tool to make valid predictions.
Another feature that remains to be explained is the second method of prediction, The “Average Approximation” method. The only place where that method differs from the procedure explained above is when the window is first opened. Instead of choosing a set of consecutive dates of oscillator prices, you can draw your own single line on the oscillator (bottom pane) specifying your best guess for a straight line approximation of the oscillator data over a given time period. Alternatively, you can draw several lines and let FDC average them. The resulting line is then used by FDC to construct the parabolic prediction channel. In the above case, with this method, the first thing we might do is draw the line shown below.
This line is then used directly by FDC, as the best approximating straight line to that portion of the oscillator data. You might do this if you thought that part of the chart was being overestimated by regression, perhaps because an obvious pattern had not yet formed. In any event, this lets you substitute your guess for the least squares line. Alternatively, you can draw several lines, say an upper and lower envelope, and let FDC take the average. This might be better than a freehand guess. You could also draw upper and lower envelope lines and a middle line giving a little more weight to the upper or lower part of the channel. For example, you could draw the lines shown below.
You can draw as many line segments as you like. FDC will look only at the dates common to all lines, and will average the lines on those dates. The rest of the process is just as described above.
There is one final tool, the CLEAR tool shown selected below.
This can be used at two different times, with two distinct effects. If you use this tool after drawing lines on the oscillator chart, but before clicking the parabola tool, then your lines are cleared, and you can draw again. If you use the clear tool after generating the parabolic prediction, then both the top and bottom panes are cleared of that data generated by the prediction, and are restored to their original state. You can then begin the process all over again in the bottom pane.
When you leave the prediction window, after having made a prediction, you can discard the data or save it as a real FDC Dataset. This allows you to see the predicted parabolic and straight lines as actual data, which can then be treated as any other data would be treated. The proper way to exit is to use the Exit menu, as described below.
Note: Should you have any question as to which method of prediction is best for you (“Automatic” or “Average Approximate”), it is suggested that you begin with “Automatic”.
Menus For the Parabolic Prediction Plot Window
The Exit menu expands as follows:
This is the proper way to exit a blank data screen, either retaining data, which then becomes an ordinary FDC Dataset, or discarding the data. If you simply close the window, you will be prompted as to whether to retain or discard the data.
The
expanded Set Type menu is shown below.
The choices here are self explanatory.
Finally, when the Help! menu is selected, the present help screen is presented.
A word of caution: Be objective. Do not fall into the trap of creating Parabolic Predictions that only justify your opinion of the market's future.
The
following is the plot template for a 1-column parabolic prediction.
