Custom Search

Application of TOPFINDER techniques

The foregoing examples underscore the circumstance that application of TOPFINDER techniques frequently involve ambiguities   either in the choice of launch and "fit" points, or in the possibility for more complex simultaneous structures. While a regrettable   frustration in our search for the magic wand that turns all to gold, there do occur "textbook" examples devoid of such   complications which have great profit potential.

So to end on this positive note, let's go a bit further with QCOM and look at the next bull leg – another double from 16 to 32. As   seen in the fourth figure, the straightforward T1 associated with the new S1 works   like a charm catching the top exactly. So   something is clearly going on here and in the next article we'll examine the TOPFINDER algorithm and try to understand why it works.

Strongly trending price moves are distinguished by their motion away from the theoretical S/R level launched at the start of the   move. In several examples we have seen how these accelerated moves find support (resistance) at a TOPFINDER   (BOTTOMFINDER) curve which frequently has the ability to predict thecumulative volume at which the move will end. In this   article we present the TOPFINDER algorithm and discuss its implications.

To understand TOPFINDER, it is worthwhile to rewrite slightly the algorithm used to generate the S/R levels. In the first figure   we collect all of the MIDAS algorithms, written in their most useful form. The S/R levels are given by the quantity called "P-Bar- Star", where P stands for price, Bar for average, and Star being an acronym for "subsequent to a reversal". By this device, the   symbol itself - when given its mathematical pronunciation - describes how it is to be computed (the average price subsequent to a reversal)!

From the equation for P-Bar-Star, it is seen that it is now expressed in terms of the cumulative volume difference (or   "displacement" as we shall call it) between the launch point and the point of computation. The S/R level is simply the cumulative   price*volume at the given instant minus the cumulative price*volume at a point d units of cumulative volume earlier, all divided   by d, where d is the displacement measured from the (fixed) launch point.

We now define a new quantity called "P-Hat", where "Hat" is now an acronym for "hitting a top". P-Hat is the TOPFINDER (or   BOTTOMFINDER) curve, and is computed by an equation which looks very much like that for P-Bar-Star. The difference lies in   the relationship between the effective displacement, e, used in the P-Hat equation and the displacement, d, used in the P-Bar-Star   equation. Specifically, e is related to d parabolically through the equation e=d*(1-d/D). D is now a new parameter which we shall   call the "duration" of the move since e goes to zero when d approaches D.

The TOPFINDER algorithm (i.e. the equation for P-Hat) simply says that instead of keeping the launch point fixed (as with P-Bar- Star) one allows it to move forward in time towards the present. In effect the launch point can be visualized as chasing the present,   finally catching up when the D units of cumulative volume have built up subsequent to the starting launch point. Putting it another   way, as the move uses up its allotted duration, the topfinder curve represents an average price taken over successively shorter   intervals, whereas the S/R level is an average taken over successively longer intervals.

This is illustrated in the second figure which shows the relationship between e and d. We have drawn the actual parabolic curve   for e, as well as three linear tangents and their corresponding cumulative volume "zones" I,II, and III. In zone I, the start of the   move, e and d are very close to each other so the topfinder launch point, while actually moving away from the starting launch   point, doesn't move very far. The TOPFINDER and S/R urves are thus close to one another in this initial zone. (Indeed, it is the   failure of the price to pull back fully to the S/R curve, that alerts us to the fact that topfinder is coming into play since a shorter   displacement than d is required to "fit" the actual price pullback).

In zone II, the topfinder displacement is roughly constant at a value close to one-quarter of the duration

D. If D were 100,000 shares, for example, then in zone II the price is finding support (or resistance) at a price

average taken over the past 25,000 shares. This is behaving, in effect, like a conventional moving average,

taken over a fixed number of shares rather than a fixed number of days.

In zone III, the effective displacement is rapidly diminishing. In fact, to a first approximation, for every new

share traded while in this zone, the averaging interval shortens by one share. It is this remarkable feature of the

climactic end of the move that gives us a clue as to what might be going on.