The appeal, and the trap inside it
Looking at gold across several decades, the eye finds structure immediately. Long periods of nothing, then a sustained advance, then a peak that holds for years. It is natural to conclude that this is a cycle with a length, and from there it is a short step to drawing the next one. The step is short and it is unsupported.
The arithmetic problem is simple. Depending on how you define a cycle, the modern gold market offers only a handful of complete examples. No pattern can be established from that many observations, no matter how convincing it looks on a log chart. What can be studied is the mechanism, because a mechanism can be reasoned about and checked against conditions today, while a cycle length can only be fitted to the past. The distinction between a convincing picture and a measured edge is the subject of why a score is not a probability, and it applies here with full force.
The engine that appears every time
Gold produces no cash flow. It has no coupon, no earnings and no tenant. That single fact means its main competitor is whatever safe cash pays after inflation, which is the real return on short term government debt and deposits. When that number is comfortably positive, holding metal costs you something measurable every year. When it is negative, the cost disappears and the metal competes on its own terms.
Every major gold advance maps onto a period in which real returns on safe assets were poor, falling, or expected to fall. The 1970s advance followed the end of dollar convertibility in 1971 and ran through an era when nominal rates chased an inflation rate that kept getting away from them. The 2000s advance ran alongside a long decline in real yields. The most recent one has the same signature. This is the closest thing to a durable law in gold pricing, and it is the first number to check before accepting any other explanation.
The second engine is confidence
The real return story is incomplete on its own, because it does not explain why gold sometimes rises while real returns are unremarkable. The missing piece is confidence in the institutions that set the rules: the credibility of the currency, the predictability of policy, the question of whether official debt will be repaid in money that holds its value. Gold is the asset you own when you doubt the answer, which is why it responds to events that have nothing to do with growth.
You can see the same mechanism at different scales. A banking stress episode such as the one in 2023 is a short, sharp version of it, where the doubt is about specific institutions rather than the system. A multi year advance is the slow version, where the doubt is about the arithmetic of debt and the willingness to live with higher inflation. Neither is measurable. Both show up in the price before they show up in the commentary, which is the part that frustrates people who want a fundamental trigger to point at.
The shape is usually the same shape
The structural pattern repeats more reliably than the timing does. A long base, where the asset is ignored and liquidity is thin. A slow markup that is dismissed as noise for a surprisingly long time. A broadening out, where new types of buyer arrive because access improved or because the story reached them. Then a late acceleration, usually with the loudest conviction and the worst risk reward, followed by a peak that takes years to be recognised as one.
What makes the shape repeat is not mysticism, it is participation. Each stage brings in a slower group of buyers, and the last group to arrive is the group with the least tolerance for a drawdown. That is why tops are violent and bases are boring. Knowing which stage you are probably in is more useful than knowing a date, and it is read from the weekly chart rather than from an overlay, as set out in what the weekly timeframe is for.
What brings one to an end
Advances end in one of two ways and both are about the alternative rather than about gold. Either the price of money is raised hard enough that safe cash offers a real return nobody can ignore, which is how the advance into the 1980 peak was broken, or confidence is restored and the reason for owning the asset quietly evaporates, which is closer to what followed the 2011 peak. In the second case the top is less obvious, because nothing visibly breaks. Demand simply stops arriving.
This matters because the two endings look different on a chart. A policy driven top tends to be sharp and dated, with an identifiable event. A confidence driven top is a long rolling affair of lower highs that takes a year or more to confirm. Expecting the first kind while living through the second is a common and expensive error, since every failed rally gets interpreted as the start of the next leg rather than as part of the distribution.
What does not repeat at all
Here is the list cycle enthusiasts skip. The monetary regime differs every time: a fixed convertible system, a free floating one, and one in which the official sector itself is a persistent buyer are three different markets wearing the same ticker. The identity of the marginal buyer differs: coins and bars, then futures, then listed funds, then reserve managers. Access differs, which changes how fast demand can arrive and leave. Information speed differs by orders of magnitude.
These are not details. The marginal buyer sets the price, so if the marginal buyer has changed, the response function has changed with it. A reserve manager buying on a multi year mandate behaves nothing like a leveraged fund buying a macro theme, and neither behaves like a household buying coins. Any overlay that matches two cycles by shape is implicitly assuming the buyers are interchangeable, and they are not.
The timing trap
The most popular misuse of cycle thinking is the fitted calendar: overlaying a previous advance on the current chart, aligning the starting points, and reading dates off the result. It feels rigorous because it uses real data. It is not, because the alignment point is chosen by the analyst, the scaling is chosen by the analyst, and the comparison is abandoned quietly once it stops working.
The same objection applies to fixed length cycles in general, and it is worth reading alongside the case for and against time cycles in gold. Seasonality is the slightly more defensible cousin, because it has a plausible mechanism in physical demand and fiscal calendars, but even there the effect is noisy and unstable across decades, as covered in how seasonal tendencies behave in practice. The test to apply to any cycle claim is this: was the prediction written down with a date and an invalidation before the period began, or was the match noticed afterwards. Almost always, afterwards.
Using cycle context without predicting
There is a usable residue. You can ask what the real return on safe cash is doing and whether that is getting better or worse for a non yielding asset. You can ask whether the market appears to be in a quiet base, a steady markup, or a late acceleration, based on the structure of swings on a weekly chart rather than on a date. You can ask who is likely buying now and how reversible that demand is. All three are observable and none require a forecast.
On the live XAUUSD chart this translates into ordinary work: higher timeframe structure first, then the regime label to tell you whether conditions are trending or consolidating, then execution on a timeframe you can actually manage. The honest caveat for the whole subject is that cycle context never improves an entry. It only changes how much patience you extend to a position and how surprised you are by a shock. If a cycle argument is doing more work than that in your plan, it has been promoted beyond its evidence.
FAQ
Do gold bull markets have a fixed length?
There is no good evidence for one. The modern market offers only a handful of complete advances, which is far too few observations to establish a period. The forces behind each advance can be studied and compared, but a cycle length fitted to three or four examples is description rather than prediction.
What do all gold advances have in common?
A poor or falling real return on safe cash, so holding a non yielding asset costs little, combined with some questioning of policy or currency credibility. Structurally they also share a long quiet base, a widening buyer base as the move matures, and the fastest movement arriving late.
What normally ends a gold bull market?
Either the price of money is raised far enough that safe cash offers a real return that is hard to refuse, or confidence is restored and new demand simply stops arriving. The first produces a sharp dateable top, the second a long rolling one that takes a year or more to confirm.
Why are overlays of past cycles unreliable?
Because the analyst chooses the alignment point and the scaling, and quietly drops the comparison when it stops matching. More fundamentally, the marginal buyer changes between cycles, and since the marginal buyer sets the price, two advances with similar shapes can have entirely different response functions.
Can cycle analysis help a trader at all?
In a limited way. Judging whether the market looks like a quiet base or a late acceleration can inform how much patience you give a position and how much volatility you expect. It does not improve entries and it should not be used to pick dates or set targets.
ⓘ See these ideas on real price: open the free XAUUSD live chart.