Threshold explorer

01 / How to read the chart

The thin line is the price path. Each thick line joins two confirmed extrema: a starting extremum F and the following extremum S. Together they form one event, F→S, in the terminology of the method. In standard terminology this is a directional-change event.

The latest extremum is drawn as an open ring. It is not yet an event endpoint. The shaded band shows how far price must still travel against it. Only when price crosses the dotted confirmation line does the ring become a confirmed point and a new event begin.

02 / Two numbers make the threshold

The author builds the threshold from a small unit rather than choosing it directly:

The module is the smallest price change worth using. A move smaller than the cost of a trade cannot be turned into a result even when its direction is called correctly. n says how many modules form one threshold. Price is the price of the current extremum, so δ is a share of price, not a fixed number of points. It is recomputed whenever the extremum moves.

The starting values, a module of 0.4 % and n = 5 (δ = 2 %), are the defaults of the author’s 2016 measurement script, TrendMetr.

The lower chart counts events on the whole path for every threshold from 0.4 % to 8 %. Halving δ produces many more events. No threshold is the true one: the threshold is a choice of scale, and the markup is a description at that scale.

03 / Why the last event is always open

Press Replay. While price keeps rising, the highest price so far is only a candidate. It becomes an extremum when price has fallen from it by more than δ. The square marker shows that moment for the selected event; the dotted corner joins it to the extremum it confirmed.

Event markup therefore lags price by one threshold. Every confirmed point is known only after the fact, and the most recent event is always unfinished. Any honest use of event markup has to show that open end rather than hide it.

04 / Contra and strength Φ

Contra is the largest pullback inside an event that does not exceed the threshold. It is measured from the highest price reached so far (or the lowest, in a falling event) to the deepest point of the pullback. It is bounded below by the module: when no pullback reaches one module, the module itself is used and the read-out says so.

Strength relates the event’s size to its contra:

A rising event has positive Φ, a falling one negative. Because a pullback inside an event cannot exceed δ without ending the event, |Φ| under this definition is at least about 1. Φ is the author’s own measure; no direct counterpart has been found in the literature. Its limiting value, Babylon, is an open research question.

05 / A caution: noise has events too

Every path on this page is random. It still breaks into clean-looking events, and some of them still have large Φ. Event markup will draw structure on any series. Whether market events differ from those of noise is an empirical question. It has to be answered by comparison, for example with shuffled versions of the same data. It cannot be answered by looking at the zigzag.

06 / Implementation notes

• Markup follows the method page: the extremum is tracked while price continues in its direction; it is confirmed when the reverse move exceeds δ computed at that extremum. The first direction is set by whichever comes first: a rise of δ from the lowest price or a fall of δ from the highest.
• Contra is the largest pullback from the running extremum between F and S, with the module of the price at F as its floor. This follows the definition the author set in 2026.
• The path is a random walk with slowly changing volatility, 500 steps, generated from the seed shown. The same seed always gives the same path.
• The calculation runs in this page. A tool for marking up your own quote files is in preparation.

Read the method →