Market Geometry in Trading: Angles, Ratios and Shapes, and Which Ones Are Tested
A map of the geometric chart tools, from Gann angles to Fibonacci ratios and pitchforks, why the chart scale changes every angle, a retracement computed on real prices, and the part of market geometry our bench can actually test.
On this page
Market geometry is the family of chart methods that treat price and time as distances on a chart and look for regular shapes in them: angles drawn from a pivot, ratios between two swings, parallel lines, channels and repeated proportions. Gann angles and squares, Fibonacci retracements and extensions, the Andrews pitchfork, price channels, Elliott wave counts, harmonic patterns and classic chart patterns all belong to it. The common idea is that a market that has moved a certain distance in a certain time will tend to react at a related distance or time. Some of these shapes can be tested against chance; for the drawing tools, we found no reproducible test.
Gann fan angles
- Above 1x1
- steeper angles, price rising faster than time
- Below 1x1
- flatter angles, price rising slower than time
The main market geometry tools
| Tool | What it draws | What you must choose | Tested on our bench? |
|---|---|---|---|
| Gann fan and angles | Lines from one pivot at fixed ratios of price to time (1x1, 2x1, 1x2) | Pivot, price per bar | No |
| Gann Square of 144 | A 12 by 12 grid | Anchor, price per cell, time per cell | No |
| Fibonacci retracement | Levels at 23.6%, 38.2%, 50%, 61.8% and 78.6% of a swing | Swing low and swing high | No |
| Fibonacci spiral | A spiral of quarter circles growing toward the golden ratio | Center, first radius, direction | No |
| Andrews pitchfork | A median line and two parallels from three pivots | The three pivots | No |
| Harmonic patterns | Five-point swings whose legs follow set ratios | Detected by rule | Yes |
| Classic chart patterns | Triangles, channels, tops and bottoms | Detected by rule | Yes |
Why the chart scale matters
An angle on a chart has no meaning until you say how many units of price equal one unit of time. A 45 degree line on a daily chart becomes a steep or a flat line as soon as you stretch the price axis or switch to a weekly chart. Gann angles avoid the screen angle by naming ratios instead of degrees: a 1x1 line rises one unit of price per unit of time, a 2x1 line two units, a 1x2 line half a unit. The scale is still a choice, and every line moves with it.
The diagram shows what that does to a reading. The price path is the same in both cases; only the unit changes. At one scale the 1x1 line is broken early and reads as failed support; at the other it is never broken and reads as a trend line that held.
One pivot, one 1x1 angle, two chart scales
Ratios between swings do not have this problem. A retracement measures a pullback as a share of the move before it, which stays the same at any zoom. That is why ratio tools can be computed from real prices without hand drawing: the diagram below shows the construction, and the chart under it computes the levels on real prices.
How a Fibonacci retracement is drawn
Fibonacci retracement computed on the S&P 500 ETF
- Price, daily candles (hollow up, filled down)
- 61.8% retracement level (line)
- Other levels, measured from the end of the move back to its start (line)
- Swing highs and lows: a high above (a low below) the 5 bars on each side
| Swing low | 714.8, Jun 26, 2026 |
|---|---|
| Swing high | 781.6, Oct 6, 2026 |
| 0% level | 781.6 |
| 23.6% level | 765.9 |
| 38.2% level | 756.1 |
| 50% level | 748.2 |
| 61.8% level | 740.3 |
| 78.6% level | 729.1 |
| 100% level | 714.8 |
The swing points above come from a fixed rule (a high or low that stands above or below the five bars on each side), not from an eye on the chart. Change the rule and the anchors change, and so do all the levels. Ratios also overlap: with seven levels on a single swing, price will often be near one of them, which is why a level being touched proves little on its own.
Parallel lines and channels
The Andrews pitchfork, which Optuma’s history of median line analysis attributes to Dr. Alan Andrews, starts from three pivots. The median line runs from the first pivot through the midpoint of the other two, and two parallels pass through those two pivots. Price channels are a simpler structure: the Donchian channel, named after Richard Donchian, follows the highest high and the lowest low over a set number of bars, 20 by default on TradingView. Unlike the pitchfork, it needs no choice of pivots, only a period.
Andrews pitchfork
- M
- midpoint of B and C, the point the median line passes through
Waves, harmonics and chart patterns
Elliott wave analysis, set out by Ralph Nelson Elliott in The Wave Principle (1938), reads trends as five waves followed by a three-wave correction. Harmonic patterns such as the Gartley, in their modern form, require each leg to sit near a set Fibonacci ratio of the previous one. H. M. Gartley described the pattern on page 222 of his 1935 book, Profits in the Stock Market, without Fibonacci ratios; according to the Harmonic Trader site, the exact ratios were assigned to it later, with the book The Harmonic Trader (1998). Classic chart patterns, such as triangles, channels and head and shoulders, are geometric shapes too, and they are the part of market geometry that can be detected by rule and counted.
Detected by rule means tested: our bench applies a fixed rule to historical data, records what it detects and compares what followed with random bars of the same assets and measures the excess return over the rest of the market. The cards below show the current results for a few of them.
Does market geometry work?
For the shapes that can be detected by rule, there is an answer, and it is on each pattern’s page, market by market: some patterns did better than random entries on some markets, many did not. The cards above pool all markets. For the drawing tools (angles, squares, spirals, pitchforks), we found no reproducible test against random lines, and we do not run one. The free settings of these tools are the main obstacle: a test has to fix the anchor rule and the scale in advance, or it will measure the analyst’s hindsight rather than the tool. Our methodology explains how the pattern tests are built.
For the full set of guides, see technical analysis, and for Gann’s own tools, the profile of W. D. Gann.
Frequently asked questions
What is market geometry in trading?
It is the use of geometric shapes on price charts (angles, ratios, parallel lines, channels, waves and patterns) to find levels or times where the market might react. Gann tools, Fibonacci tools, the Andrews pitchfork and chart patterns are the best-known examples.
Is market geometry the same as Gann theory?
No. Gann’s angles and squares are one part of it. Fibonacci ratios, pitchforks, channels, Elliott waves, harmonic and classic chart patterns are others, and some of them have nothing to do with Gann. See the W. D. Gann profile for his own tools.
Does market geometry work?
It depends on the tool. Shapes that can be detected by rule, such as chart patterns, can be tested, and our bench shows mixed results. Drawing tools with free settings have no reproducible test that we know of. Read our methodology for how the tests are done.
Why do Gann angles look different on two charts of the same market?
Because an angle depends on how much price one unit of time is worth on the chart. Change the price scale or the time frame and the drawn line changes. Gann angles are written as ratios (1x1, 2x1) for that reason, but the ratio still needs a price unit chosen by the trader.
Which market geometry tool should a beginner learn first?
Horizontal support and resistance and the Fibonacci retracement: both are easy to place and easy to check after the fact. Learn how often a level is touched without any reaction before you rely on it.
Sources
Passionate about finance, quantitative trading and data analysis. Amaury built the PatternsWizard test bench, which detects each pattern on full price history and measures what happened next against random entries.
PatternsWizard tests chart and candlestick patterns on full market history, against random entries. How we test patterns.