© Andrew Coles       Terms & Conditions | Privacy

MIDAS MARKET ANALYSIS
Modified VWAP Methodologies
Share on Facebook
Share on Twitter
Share via e-mail
Share on Stumble Upon
Share on Reddit
Share on Newsvine
Share on LiveJournal
Share on Google Bookmarks
Share on Digg
Share on Delicious

Essay one continued

Working with TB-F curves

So far as the interaction of TB-F curves with price is concerned, the first thing to say is that the implication of price breaking through them is much the same as the implication of price breaking through a standard MIDAS S/R curve, namely that the trend is likely to have ended.


However, there is a more subtle feature of TB-F curves that make them very different from standard M-S/R curves. This again connects to the fact that a TB-F curve requires a smaller amount of cumulative volume than a standard M-S/R curve to keep moving with the trend. This more subtle feature, as we’ll see below, relates to the fact that this smaller amount of cumulative volume is also finite, resulting in a quite dramatic form of chart behaviour in TB-F curves.  


Prescient readers will ask how we ensure that we do obtain a smaller amount of cumulative volume in the TB-F curve than in a standard MIDAS S/R curve. The answer is that the user of a TB-F curve himself supplies this smaller amount of cumulative volume manually. Alternatively there is a function within some software running the TB-F ensuring that this smaller amount of cumulative volume is automatically supplied to it according to user-based chart adjustments and inputs.


The next question prescient readers will ask is if the cumulative volume must be supplied to the curve manually (or by an additional software function), how do we know precisely how much cumulative volume to supply to the TB-F curve.


The answer to this question can be provided with the help of Figure 4.









Figure 4 consists of a hypothetical rising trend with the red (lowest) curve representing a standard MIDAS S/R curve which is no longer supporting the trend because the latter is moving too quickly away. This is precisely the scenario in which we normally launch a TB-F curve.


In Figure 4 the green curve represents a TB-F curve. Notice that it is launched from the same place as the standard MIDAS S/R curve and that it is fitted to the pullback (A) which the standard M-S/R curve fails to reach. Thus, the answer to the question of how much cumulative volume is required to keep a TB-F curve moving apace with a sharply moving price trend is that it is determined by fitting the curve to the first pullback (A). This fitting process is iterative, meaning that the amount of cumulative volume required is a trial-and-error procedure of inputting various amounts until there is a fit. This process sounds onerous whereas in fact it usually takes a couple of seconds. In automated systems, such as in the StockShare software for which David Hawkins acted as consultant, the amount of cumulative volume is determined automatically as the user drags the curve to the relevant pullback. In other current systems, such as the Metastock external DLL currently being used by Andrew Coles, the procedure is manual.


If we now go back to the diagram, the two blue lines represent hypothetical cases where too little or too much cumulative volume has been inputted. In the case of the latter, a TB-F curve will underfit the pullback, meaning it will be too far away from it (in effect, it is now behaving more like a standard S/R curve). In the case of too little cumulative volume inputted, a TB-F curve will overfit the curve, meaning that the displacement is now too small and that the curve is accelerating even faster than the price trend.


The finite cumulative volume to be added to a TB-F curve is known as D, which stands for the duration of the move. As noted above, it is this concept of finite duration that creates another dramatic difference from a standard M-S/R curve. For what a finite amount of cumulative volume implies is that a TB-F curve will literally come to a stop on the chart when it runs out of cumulative volume; in other words, when D, the duration, expires.


Think of D in terms of a finite source of energy stretching an elastic band. Because this energy is finite, the elastic band will be stretched to a point proportional to that energy before it snaps back to its original configuration. The pulling out of the band represents the accumulation phase of a market item being purchased or sold short, while the snapping back represents the distribution phase of the same item being sold or bought back. D reflects the energy equivalent to this accumulation phase very accurately. As a result, because the distribution phase mirrors the accumulation phase in D, the termination of a TB-F curve on a chart tends to be an accurate indicator of that accelerated portion of the trend reacting in some way.


It is important to use the term ‘reacting in some way’ instead of ‘ending’ because the latter would be a presumption and not a solid piece of information supplied by the TB-F curve. Think of it this way: if a finite energy source pulls out an elastic band and then expires as the elastic band snaps back again, it does not follow that a new finite energy source will immediately repeat the same effect straightaway. There may be a resting period in between. Thus, when a TB-F curve is indicating that the cumulative volume (D, the duration) is running down and that the curve will soon come to an end, all this means is that there is likely to be some response, not necessarily that there will be a change in trend. It may be, for example, that the market will now go into a resting phase, which is where again Andrew Coles’ MIDAS Displacement Channel would have a role to play.





previous page next page Figure 4