Vortex Indicator: Where It Came From and What It Measures

Five explanations of this indicator were read in full while preparing this page. They disagree about when it was published, they disagree about what period to run it on, and every claim any of them makes about whether it works comes from the same place. That last fact is the one worth knowing before loading it.

Where It Came From, and the Date One Platform Gets Wrong

The indicator is credited to Etienne Botes and Douglas Siepman, and all five sources agree on the names. They do not agree on the year.

One of the largest charting platforms states in its own support documentation that the indicator first appeared in the January 2020 edition of Technical Analysis of Stocks and Commodities. Three of the other four put it in January 2010, and the encyclopedia entry supplies the detail that settles it: an article by both authors in volume 28, issue 1, at page 21.

A decade is a long error, and it is not harmless. A reader who takes the later date away with them will assume the tool is recent, assume it has been tested against the last fifteen years of market data because it was designed in them, and treat the absence of independent study as a sign of newness rather than of neglect.

It is also the kind of error that spreads, because the platform stating it is where many readers go to load the indicator in the first place. The correction costs one sentence and nobody else on the first page of results makes it.

The date matters for one further reason. An indicator published in 2010 arrived after the directional movement system it is usually compared against, and was written as an answer to it, so the comparison is not incidental to the tool but the reason it exists. Reading it as a 2020 arrival loses that, and with it the question of whether the answer improved on what ADX measures.

Everything Known About Its Performance Comes From One Magazine Issue

Set the date aside and a larger problem appears. Every empirical claim about this indicator that any of the five repeats comes from the January 2010 issue, or from the software column published inside it.

The encyclopedia entry is unusually honest about this, though the honesty is easy to miss because it sits in a maintenance notice rather than in the text. That notice says the article leans too heavily on primary sources and asks for independent ones, and it has carried that request since 2010. Its reference list is the original article, the companion column, and the tests discussed below.

None of the other four mentions the situation at all. Two of them describe the indicator as popular among traders and one calls it lagging, and none of the three attaches evidence to the description. Popularity is not performance, and neither is repetition.

What this means in practice is narrow and worth stating plainly: there is no independent, peer-reviewed evaluation of this tool to consult, so anyone deciding whether to use it is deciding without one. That is a different position from the one a reader is in with what momentum indicators measure generally, where the underlying idea long predates any single publication.

The Two Tests in That Issue Reach Opposite Conclusions

The same issue carried two comparisons against the directional movement system, both by the same analyst, and putting them beside each other is something none of the five does.

The first ran across 38 futures contracts spanning index, currency, metal, energy and agricultural markets, over a window beginning in January 1978 and ending in November 2009, with both tools set to 14 periods. Over that window the newer indicator came out ahead.

The second used 101 NASDAQ stocks over a shorter window, from January 1992 to August 2009. There the older directional system came out ahead instead.

Two results pointing opposite ways is not a scandal, and it is not a reason to discard either. It is a reason to be precise about what was shown: an edge that appears on decades of futures data and reverses on seventeen years of individual shares is an edge tied to what was tested.

Anyone porting it to a currency pair is outside both samples. Both tests ran against the directional movement system, and the sample that favoured this indicator is the one built from futures rather than shares.

What the Two Lines Are Actually Built From

Four of the five give the formula and none explains what the parts mean, which leaves the reader copying arithmetic rather than understanding a measurement.

Both lines start from a distance. The upward line measures how far the current high sits above the previous low; the downward line measures how far the current low sits below the previous high. Each is a span between one bar and the one before it, so both are always positive and neither carries a direction on its own.

Those distances are then summed over the chosen period and divided by the summed true range over the same period. True range is the quantity what true range measures covers in full, and it is what makes the output comparable across instruments: dividing by it turns a distance in price into a ratio.

So the two lines answer one question each. The upward line asks how much of the period’s total range was spent reaching above the prior bar, the downward line asks how much was spent reaching below it, and because both share a denominator they can be read against each other. Nothing in either is a forecast, and nothing in either knows what a trend is.

The Settings Nobody Sources

The period length is where the five diverge most, and none of them says where its number came from.

One platform document offers a range from 14 to 30. A charting school calls 14 typical and then works its examples at 26 and at 23. A third source names 14 as the default and titles a scanner at 20. The encyclopedia entry works its illustration at 21. Threshold levels appear in one source at 0.90 and 1.10, again unattributed.

None of that is wrong, and a reader could reasonably use any of them. The problem is that the numbers are presented as settled when they are choices, and a choice presented as a standard cannot be evaluated. This is the same pattern our page on an indicator whose default period nobody sources found on a different tool, which suggests it belongs to how indicator documentation is written rather than to either indicator.

The practical consequence is that two readers following two of these pages will see different crossovers on the same chart on the same day, and both will believe they are using the indicator correctly. The True Strength Index shows the same problem in a starker form, because there the platform reference names three period inputs and publishes a value for none of them.

Reading a Crossover Without Treating It as a Signal

A crossover happens when the two ratios swap places, which means the period just measured spent more of its range reaching one way than the other. That is a description of what has already occurred.

Treating it as an instruction adds a claim the arithmetic does not contain, and the evidence available to support that claim is the single issue described above. A reader who wants the crossover to mean something has to supply the meaning, in advance, and say what would count as it failing.

There is a second reading that survives the problems on this page. Rather than acting on the swap itself, note how long each line has held the upper position, since a ratio that changes places every few bars is describing a range and one that holds for many bars is describing a stretch of one-way movement. That reading needs no claim about accuracy, because it only reports what the denominator already contains.

When This Indicator Has Nothing to Say

Because both lines divide by true range, a period of narrow bars makes the denominator small and the ratios jumpy. Readings taken in a quiet session move for arithmetic reasons rather than market ones.

It is also silent about level. It cannot tell you where price is, only how the last stretch of range was distributed, which is why it sits badly on its own and why why two brokers draw different levels is the more useful question when the decision is about a price rather than a direction.

Before relying on it, three checks are worth running. Confirm which period your platform loaded by default rather than assuming 14. Note whether the instrument you are on resembles either of the two samples that were ever tested, since a currency pair resembles neither. And decide in advance what a failed reading looks like, because an indicator with no independent evaluation behind it cannot supply that answer for you.