Stop talking about gold anymore. Bitcoin is not a safe-haven asset.

When Bitcoin Bottoms

The crypto market is not exactly hot right now. While stock and metal prices continue to hit new all-time highs, cryptocurrencies have been stuck on the “pain train” since October last year.

Recently, a common narrative has been circulating: “Funds are rotating from precious metals into crypto assets, and a shift is imminent.” Unfortunately, those pushing this idea are often known as the industry’s “talking heads,” whose only consistent profit comes from monthly engagement revenue on X platform.

I want to take some time to analyze whether this so-called “rotation from precious metals to crypto” has any real basis (spoiler: it doesn’t), and then share some key historical turning points in the crypto market, along with how you can identify these moments.

The Relationship Between Bitcoin and Gold

First, an obvious question: if we want to find a relationship between gold peaks and Bitcoin performance, the premise is that gold itself must frequently “top out.” But in reality, over the past decade, genuine peaks in gold have been few and far between.

It’s tempting to get excited, but when sharing opinions online, it’s better to have data to back it up—otherwise, you risk sounding like a complete fool. Over the past ten years, gold has only experienced three significant corrections: in 2018, 2020, and 2022. That’s just three data points. Just on that basis alone, I could stop here; but for the sake of completing this article, let’s take a closer look.

If you examine the chart above, you’ll notice that two of these gold peaks occurred before Bitcoin entered a downtrend—specifically in 2018 and 2022. The only time Bitcoin strengthened after gold’s correction was during the 2020 risk-on frenzy.

Over roughly the past 10 years, the overall correlation between Bitcoin and gold has been close to 0.8, which isn’t surprising—because both markets have been trending upward over the long term. But correlation alone doesn’t answer the question you really care about.

If you want to determine whether there’s a “counter-movement, rotation of strength, and eventual reversion” relationship between two assets, looking at correlation isn’t enough. You need to examine cointegration.

Cointegration

Correlation measures whether two assets tend to move together in daily fluctuations—whether they rise and fall in tandem. Cointegration, on the other hand, asks a different question: do these assets maintain a stable long-term relationship, such that deviations from this relationship are eventually corrected?

Think of it like two drunk friends walking home together: they might stagger and veer off course (non-stationary), but if they’re tied together with a rope, they can’t stray too far apart. That “rope” is the cointegration relationship.

If the narrative that “funds are rotating from gold into crypto” has any real substance, then at minimum, Bitcoin and gold should exhibit cointegration—meaning that when gold surges and Bitcoin underperforms, some underlying force should pull them back onto a shared long-term trajectory.

Looking at the data above, the actual message is that the Engle–Granger cointegration test found no evidence of cointegration.

The p-value for the full sample is 0.44, well above the typical significance threshold of 0.05. Further, in rolling two-year windows across 31 intervals, none showed cointegration at the 5% significance level. Additionally, the residuals from the spread are non-stationary.

A simpler view of the BTC/gold ratio appears slightly “more optimistic,” but not by much. An ADF test on this ratio yields a borderline stationary result (p = 0.034), suggesting it might have very weak mean-reversion properties. However, its half-life is about 216 days—nearly 7 months—which is extremely slow, almost drowned out by noise.

Currently, Bitcoin’s price is roughly equivalent to 16 ounces of gold, about 11% above its historical mean of 14.4. The corresponding z-score is -2.62, indicating that Bitcoin seems relatively “cheap” compared to gold from a historical perspective.

But here’s the key point: this reading is mainly driven by gold’s recent parabolic rally, not by any reliable mean-reversion relationship that would pull the two together.

In fact, no robust cointegration exists. They are fundamentally two very different assets: gold is a mature safe-haven asset; Bitcoin is a highly volatile risk asset, just happening to trend upward during the same period.

If you’re completely lost with all this, here’s a very simplified crash course in statistics:

Engle–Granger test is the standard method for detecting cointegration. It involves regressing one asset on the other, then testing whether the residuals (the spread) are stationary—meaning they fluctuate around a stable mean rather than drifting infinitely. If residuals are stationary, the assets are cointegrated.

The Augmented Dickey-Fuller (ADF) test checks whether a time series is stationary. Essentially, it tests for the presence of a “unit root”—whether the series tends to diverge or revert to a mean. A p-value below 0.05 allows you to reject the “unit root” hypothesis, indicating the series is stationary and exhibits mean reversion.

Half-life measures how quickly the spread reverts to the mean. If the half-life is 30 days, it takes about a month for the deviation to be halved. Shorter half-lives imply more tradable opportunities; longer ones suggest the spread is slow to revert and less useful for trading.

Ultimately, I’ve always thought trying to force Bitcoin into the mold of any traditional financial asset is inherently absurd. Most of the time, people use such comparisons to fit the narrative they prefer: today, Bitcoin is “digital gold”; tomorrow, it’s “leveraged Nasdaq.”

In contrast, its correlation with stocks is much more tangible. Over the past five years, Bitcoin’s peaks and troughs have been highly synchronized with the S&P 500—until now. The S&P 500 remains near all-time highs, while Bitcoin has already retraced about 40% from its peak.

Because of this, you should treat Bitcoin as an independent entity. It’s not a metal—no one considers an asset with over 50% annual volatility a safe haven (compared to gold’s roughly 15% annual volatility, which is already high for a store of value). It’s not a stock index—Bitcoin has no component stocks; it’s essentially just a piece of code.

Over the years, Bitcoin has been repeatedly wrapped in various narratives: payment tool, store of value, digital gold, global reserve asset, and so on.

While these ideas sound appealing, the reality is that this is still a relatively young market. It’s hard to say whether it has developed clear, stable practical uses beyond being a “speculative asset.” Ultimately, treating it as a speculative asset isn’t wrong—what matters is maintaining a clear, realistic perspective.

Bottom

Getting a reliable, steady bottom for Bitcoin is extremely difficult—any market is, but Bitcoin’s rapid evolution over the years makes historical patterns less and less relevant.

Ten years ago, the market structure of gold and the S&P 500 was not very different from today; but in 2015, one of the main reasons for holding Bitcoin was still to buy heroin online.

That has obviously changed dramatically. Today’s market participants are much more “serious,” especially after the surge in CME Bitcoin futures and options open interest in 2023, and the launch of Bitcoin ETFs in 2024—marking the entry of institutional capital on a large scale.

Bitcoin is a highly volatile market. If there’s one conclusion we can be relatively confident in, it’s that market bottoms are often marked by extreme overreactions and “liquidation cascades” in derivatives.

This signal appears both in native crypto indicators—such as extreme swings in open interest and funding rates—and in more institutional metrics like options skew and abnormal ETF fund flows.

I’ve personally developed an indicator that consolidates these signals into a composite regime (note: this indicator is not publicly available at the moment, sorry). As shown in the chart, red zones typically correspond to periods of extreme market despair: declining open interest, negative funding rates, traders paying high premiums for puts, and realized volatility exceeding implied volatility.

Meanwhile, the spot-volatility correlation in Bitcoin, although still somewhat chaotic overall, is increasingly exhibiting characteristics similar to stock indices.

Summary

If you’re looking for “entry points, stop-loss, take-profit” signals, I’m afraid I can’t help you (not that I’d want to).

The main purpose of this analysis is to clarify a seemingly obvious but often overlooked fact: Bitcoin is an independent market. It sometimes behaves like gold, sometimes like stocks, but fundamentally, there’s no inherent reason for them to move in sync over the long term.

If you’re staring at a falling price and trying to guess when the bottom will form, it’s better to focus on the real data that matter—like the structure of positions, which often tell the most honest and brutal stories.

And don’t forget: most genuine bottoms are formed when nearly everyone has given up.