WhitmanTrading

What Is the Gann Square?

Gann square is a method of relating price levels to dates by arranging numbers in a spiral or grid and reading off positions that share an alignment. It requires converting price into time units, a conversion nothing in the market supplies, so the output depends on the scaling chosen.

What Is the Gann Square? — illustrated on a chart Watch: Ultimate Tutorial On Gann Box and Gann Square Trading Strategy

The Gann square is the most ambitious idea in this family: that price and time are related, and that the relationship can be read off a grid. It is worth understanding precisely, including where it depends on an assumption.

How it works

A price series with levels mapped against dates.
The Gann square maps price onto time. Illustrative chart - not real market data.

Numbers are arranged in a spiral. The square of nine starts at 1 in the centre and winds outward, so each full rotation lands on a square number — 9, 25, 49, and so on.

A steady series where price and time are assumed proportional.
It assumes the two move in proportion. Illustrative chart - not real market data.

Positions on that spiral are then read as related. Values sitting at the same angle from the centre — 90, 180 or 270 degrees apart — are treated as connected.

A rising series where an alignment is identified.
That assumption is the whole method. Illustrative chart - not real market data.

Prices and dates are both placed on it. A price level and a date that land on the same angle are taken to indicate a moment where the market may turn.

A falling series where the alignment produces nothing.
And nothing in the market enforces it. Illustrative chart - not real market data.

The assumption underneath

A choppy series where the conversion rate is arbitrary.
So the scaling is an input, not a discovery. Illustrative chart - not real market data.

Price is measured in currency. Time is measured in days. To place both on one spiral you must decide how many currency units equal one day, and that number comes from you.

A slow series where the mapping drifts over a long horizon.
And different again over a long horizon. Illustrative chart - not real market data.

Change the conversion and every alignment changes. The dates the method identifies are a function of the rate you chose, not of anything the market did.

A calm series where no alignment is visible.
A quiet stretch hides what it measures. Illustrative chart - not real market data.

And the spiral produces a great many candidates. Four angles per rotation, across many rotations, across any starting point — the number of alignments available is large enough that some will land near something notable.

A worked example

Take this site’s shared series. It runs 576 bars, the median bar range is 0.493 and the largest is 2.338.

Suppose you set one price unit equal to one bar. Alignments then fall at prices roughly 0.5 to 2.3 apart in the regions the series occupies — spacing similar to ordinary bar movement.

A falling series with a stop level marked.
A stop fills where the market is. Illustrative chart - not real market data.

Set one price unit to ten bars instead and the alignments spread ten times wider, producing an entirely different set of dates from identical data.

Neither scaling is more correct. There is no measurement that decides it, which means the method’s output is determined by a parameter with no external reference.

The honest way to hold it

Time-based analysis is not absurd in itself. Markets have genuine periodicity — sessions, expiries, earnings dates, month ends — and those recur for reasons that can be stated.

Anniversary effects are a real, documented phenomenon. Anticipating that a market may be active around a known date is different in kind from deriving a date from a spiral.

And marking dates in advance imposes useful discipline, the same way marking price levels does. Deciding when you will re-examine a position is better than deciding it reactively.

What does not survive is the mechanism. A number’s position on a spiral is a property of arithmetic, and no process connects it to what buyers and sellers will do. Anybody presenting it as one has moved from technique to belief, and the distinction is worth keeping visible.

The original data

On this site’s shared series: 576 bars, median bar range 0.493, ninetieth percentile 1.101, largest bar 2.338. Direction runs average 2.01 bars with a longest of 11. A round trip costs 0.0098, about 2% of the median bar range.

Those run figures are the practical test. A method identifying specific turning dates is claiming precision about a series whose direction persists, on average, for two bars.

A candlestick chart annotated with the cost of a round trip.
A round trip costs a share of a bar. Illustrative chart - not real market data.

And acting on each identified date costs the spread. A method generating many candidate dates generates many opportunities to pay 0.0098, which accumulates well before any of them is right.

A price series with volume shown beneath.
Volume and price measure different things. Illustrative chart - not real market data.

Why it persists

The output is impressive. A spiral of numbers with angular relationships looks like mathematics, and mathematics carries authority that a hand-drawn line does not.

Hindsight is extremely accommodating. With enough candidate alignments, some will sit near a real turn, and those are the examples that get shown.

And the originator’s reputation does heavy lifting. Claims about Gann’s personal trading results are repeated constantly and trace to promotional material rather than to audited records — this site does not publish unverified performance figures, and that includes historical ones.

Which leaves the method to be judged on whether it works, tested forward, by you, with a conversion rate fixed before the test rather than after. That is a reasonable thing to do and it is the only honest way to find out.

When it fails

The characteristic failure is fitting the scaling to the past. The square is applied, the dates it produces do not correspond to anything, so the conversion between price and time is adjusted until they do. The new scaling explains recent turns beautifully — it was selected for that — and is then carried forward as though it had been derived rather than fitted. Every parameter chosen to explain history will explain history; whether it explains anything else is a separate question the fitting process cannot answer.

A candlestick series with a gap through a level.
A gap skips the level entirely. Illustrative chart - not real market data.

A second failure is counting only the hits. With many candidate alignments, some will coincide with turns by arithmetic alone.

A third is treating the spiral as a mechanism rather than as an arrangement of numbers.

A fourth is repeating claims about Gann’s returns, which are not verifiable.

A declining series cut short at a decision point.
The date aligned. Did it have to? Illustrative chart - not real market data.

And a fifth is skipping the forward test. If the method works it will work on dates chosen before the outcome, and that test costs nothing but patience. Record the dates and the conversion rate in advance, in writing, then check them afterwards without adjusting either - that single discipline separates a method from a story more reliably than any amount of argument about the underlying theory.

Gann box covers the proportional tool from the same family. Gann fan covers the angular one and its own scaling problem. And Fibonacci retracement covers proportion-based levels without the time component.

What I actually do

This is the part of the Gann material I am most sceptical of, and I would rather say so than describe it neutrally. Price and time have no common unit. Any method that equates them has chosen a conversion rate, and the choice is doing the work that the geometry appears to be doing.

— Michael Whitman, from this video

This page is educational, not financial advice. Test every idea on your own charts before risking money.