The Rhythm Hiding Inside Gold's Weekly Chart

Where this started

I found this by watching a video I probably wasn't supposed to take that seriously

A few months ago I stumbled across a video explaining gold through cycle analysis. Harmonic families, 84 months, 42 months, 21 months, all layered on top of each other like a tuning fork, plus fibonacci confluence zones where two or three of these cycles agreed on a turning point. My first reaction was skepticism. I have spent enough time around Gann squares and cycle theory to know how easily this stuff turns into pattern matching after the fact.

My second reaction was to open Pine Script, because that is what I do with skepticism now. If the idea has any teeth, I can build it and test it against real weekly closes instead of arguing with a chart screenshot.

12A · 21A · 42A · 84A

Four cycles, nested like a tuning fork

The core of the indicator is a set of arcs, drawn straight onto the price chart rather than tucked away in a separate pane. Each arc represents a harmonic period, twelve months for the smallest fractal, twenty one for the next layer up, forty two for what I think of as the medium cycle, and eighty four for the major one that has, historically, marked most of gold's real inflection points since 2008.

The arcs are anchored to a fixed calendar date and scale to the instrument's own all time price range, so a major dome on a fifteen year chart looks proportionally the same as it would on a fifty year one. It sounds like a small detail until you actually try to overlay a cycle study on an asset that went from four hundred dollars to over five thousand. At that point most naive implementations either disappear into the bottom of the chart or blow the scale out completely. Mine does neither, on purpose, after a few frustrating iterations that did exactly that.

Fib A / Fib B

Two fibonacci grids that draw themselves

Underneath the cycle arcs sits a pair of fibonacci retracement and extension grids. The first, the major grid, locks onto the largest swing high and swing low the script can find using pivot detection, not a date I typed in six months ago and forgot to update. The second grid does the same thing on a tighter lookback, so it tracks the most recent meaningful swing while the major grid holds the bigger structure.

Both grids recalculate on their own as new pivots confirm. That single change, moving from hand entered swing dates to automatic pivot detection, is the difference between a tool you have to babysit and one you can actually forward test. It draws the same lines a trader looking at the chart for the first time in 2031 would see, not the lines I happened to draw in July 2026.

The confluence line

When a cycle turn and a fibonacci level land on the same week

The most useful signal the indicator produces is not any single element. It is agreement between elements. A vertical confluence marker appears every eighty four months from the anchor date, and separately the script tracks a shorter forty two month marker I use as an early warning. When one of these markers lines up with a fibonacci level or a genuine swing point already on the chart, that is the zone I actually pay attention to. Everything else on the chart is context.

Above the current all time high, the same logic runs in reverse. Once price actually forms a lower high after a peak, the script connects that peak to the new lower high and extends the resulting line forward as a declining resistance level, the kind of falling ceiling you would otherwise have to draw and redraw by hand every time the market prints a new local top.

Where price sits right now

Two grids, two very different stories

Plotting the actual dollar levels next to each other says more than the percentages alone. The major grid is built on the 2011 to 2015 swing, the minor grid on the 2022 to 2026 swing, and gold currently trades around $4,052.

Look at the major grid for a second. Price is sitting above every single level on it, including the 261.8% extension near $3,151. The swing that grid was built on simply stopped being the relevant measuring stick a while ago, which is exactly why the minor grid exists, it picked up the job once the old one ran out of road. On the minor grid, price is holding above the retracement zone and pushing toward the 127.2% extension at $5,023, the same level that reversed price seventy percent of the time historically.

Swing accuracy

How often the cycle markers actually caught a turn

The eighty four month confluence marker fired three times in this dataset, in November 2008, November 2015, and November 2022. All three landed within a few weeks of a real swing low or swing high on the weekly chart, which for a marker with fixed calendar spacing and zero curve fitting is a better hit rate than I expected going in.

The forty two month marker is noisier by design, it fires roughly twice as often. Out of six occurrences it caught a genuine swing four times, in 2010, 2014, 2018 and 2021, and missed narrowly in 2007 and again most recently in 2024. I treat it as a heads up rather than a signal on its own. It earns attention when it lines up with something else on the chart.

How I actually use it day to day

I do not trade off any single element here. What I watch for is stacking, a cycle marker landing within a couple of weeks of a fibonacci level, sitting near the edge of the support box, or resting under the declining resistance line. When two or three of those line up, I start paying real attention to price action on that timeframe. When only one shows up, I note it and move on.

The support box and resistance line are both automatic now too, drawn from the recent trading range and from the actual sequence of lower highs after a top, so I am not the one deciding where they go. That was the point of rebuilding this the way I did. I wanted the chart to tell me when something matters, instead of the other way around.