Bitcoin is around 37% below its October 2025 peak of $126,198 while the S&P 500 sits 1.5% below its own high.

That gap is significant.

Over the past cycles, the Bitcoin debate has never settled, not for the want of trying.

Believers have a story. Sceptics have a counterstory. But the only evidence was a price series short enough to fit both.

Money does three jobs: a medium of exchange, a unit of account, and a store of value.

As a medium of exchange, Bitcoin is poor. The base layer clears roughly seven transactions a second. Fees spike when demand does. Settlement takes minutes to an hour for confidence.

As a unit of account, Bitcoin has yet to succeed. Nothing meaningful is priced in Bitcoin. Even El Salvador, which made it legal tender, keeps its accounts in dollars.

As a store of value, the answer depends entirely on one’s horizon. Over its whole history, Bitcoin has compounded at a rate that dwarfs every other asset class. Over any given day, month or year, it is one of the most volatile instruments available to a retail investor.

The digital gold analogy has one strong leg and several weak ones.

The strong leg is supply: gold's annual mine supply runs near 1.5% to 2.0% of above-ground stock and responds to price, while Bitcoin's issuance is fixed by code on a published schedule regardless of price. On pure supply inelasticity, Bitcoin genuinely is harder than gold.

Secondly, Bitcoin is digitally transferable, a significant advantage over gold, especially if one needs to carry it across borders.

On the flip side, gold has better credibility, thanks to five thousand years of monetary use, industrial and jewellery demand that puts a floor under it, and price-insensitive central bank buying. In contrast, Bitcoin has only seventeen years of history, no industrial use, and no G20 central bank has bought it on the open market. The only central bank purchase on record is the Czech National Bank’s $1 million test portfolio from November 2025.

Bitcoin is neither money nor gold. It has a “price” denominated in fiat currencies, which does not make it a currency any more than calling it a “coin” makes it one.

That is fine. Plenty of what we trade is neither money nor metal. Bitcoin simply needs its own category rather than being blended into someone else’s.

Executive Order 14233, signed 6 March 2025, created two things.

The Strategic Bitcoin Reserve holds Bitcoin obtained through criminal and civil forfeiture, designated not to be sold. The separate Digital Asset Stockpile holds every other forfeited digital asset, and Treasury may sell from it. The order permits acquisition only where it is budget-neutral.

Since then, the federal government has not bought any Bitcoin. Treasury Secretary Scott Bessent said so plainly in August 2025, and repeated in February 2026 that he has no such authority. The reserve grows only when the government seizes coins.

The reserve created a great deal of noise for the crypto industry and gave the US Government a reason to keep what it had already seized. Two years on, nothing tangible has followed.

The ownership map has changed more in three years than in the previous ten. Roughly 3.2 million coins — about 15% of all Bitcoin that will ever exist — now sit in exchange-traded funds, listed company treasuries and government hands.

The exchange-traded fund is now the marginal buyer. But it is fickle.

ETF flows turned net negative for 2026 around 26 June after a record thirteen consecutive days of redemptions worth roughly $4.3 billion. They recovered on a three-week inflow streak of about $3.8 billion in August, leaving the year close to flat.

The institutional base is real but small and rotating. First-quarter 2026 filings showed advisers adding, hedge funds cutting by 39% quarter-on-quarter as the basis trade unwound, and banks rising sharply from a tiny base.

The stablecoin proxy has stopped working too. Total stablecoin market capitalisation is roughly $302.7 billion, essentially flat for 2026 after growing about 50% in 2025. As a demand indicator it now reads neutral, not bullish.

Morgan Stanley now calls the listed Bitcoin miners "Powered Shell Providers". These companies spent 2018 to 2023 acquiring something more valuable than the machines they put on it: energised sites with signed grid interconnections.

The arithmetic is not close. A mining site earns volatile commodity revenue that resets every two weeks with network difficulty. An AI hosting contract earns fixed rent from an investment-grade tenant for ten to twenty years.

Mining revenue is also gross, before power, and falls whenever difficulty rises or price falls. Colocation rent is contracted, with electricity passed through. Against that, the capital requirement is roughly ten times higher: $0.7 million to $1.0 million per megawatt for mining infrastructure against $8 million to $15 million for AI.

Cumulative contracted value across the listed miners passed roughly $70 billion by the first quarter of 2026. Miners with signed high-performance computing contracts trade near 12.3x enterprise value to forward sales against 5.9x for pure-play miners.

Bitcoin's hashrate posted its first quarterly decline in six years in the first quarter of 2026, while difficulty is down about 13% year-to-date.

The bear case is that a falling hashrate makes the network cheaper to attack. The bull case is that it survived losing most of its hashrate overnight when China banned mining in 2021, and difficulty adjustment is built for exactly that.

Quantum gets the headlines. Demonstrated hardware is still more than an order of magnitude short of what a break would need, so I will cover it when that changes. The risk that actually cost people money this year was mundane.

On 29 July 2026 attackers began sweeping Bitcoin from wallets generated on Coldcard hardware devices.

The cause was a firmware configuration error from a release dated 17 March 2021. Affected devices silently fell back to a weak software random number generator, and older models had effectively no secure hardware randomness at all. Attackers regenerated the reduced key space offline and swept the matches.

Independent on-chain analyses put the loss between roughly 1,596 and 1,816 coins, well over $100 million, across more than five thousand addresses.

The users who lost money had followed every piece of advice the security community gives, the "not your keys, not your coins" rule. Correct behaviour was not sufficient in this case, and that is what makes this attack more damaging to the self-custody case than any of the year's larger thefts.

The practical conclusion is not that self-custody is worse than a fund, or better. It is that they carry different risk types. A fund exposes you to custodian and regulatory risk that is visible, insured and audited. Self-custody exposes you to software and hardware risk that is invisible until it is exploited.

Last year, I made a claim that Bitcoin follows global money supply with a lag of ten to thirteen weeks. My thinking has evolved since.

The popular version cites global M2, which has steadily climbed throughout 2025-2026. On that measure liquidity is expanding and Bitcoin should be rising.

But that relationship has failed to hold since late 2025.

Broad money and central bank liquidity are not the same thing. Measured properly, using the five major central banks’ balance sheets (US, UK, China, Japan and the euro area) converted to dollars at each month’s spot rate, the relationship is much clearer.

G5 central bank liquidity stood at $26.17 trillion at the July 2026 month end. It is contracting at 3.6% annualised over three months and has grown just 1.1% over twelve.

The entire twelve-month expansion is the People's Bank of China. It is being offset by the Bank of Japan. The Federal Reserve has only contributed 0.37 points.

Anyone building a Bitcoin thesis on "global liquidity is expanding" is, on these numbers, building it on Chinese balance sheet growth.

Testing the three-month change in G5 central bank liquidity against the same change in Bitcoin over 2015 to 2026, liquidity peaks three months ahead of Bitcoin. The correlation at that peak is 0.35.

The lead is real. However, 0.35 is a modest correlation. It explains roughly 12% of the variation in Bitcoin’s three-month change, so it is a tilt, not a forecast.

The most recent reading, to the July month end, has the three-month impulse running at -3.6% annualised. On a three-month lead that maps to pressure through the autumn.

Predicting asset price movement can be a fool’s errand. So far, the only theory that seems to have nailed it in public (but without any credible explanation) is the 4-year cycle.

We have developed a set of proprietary models that attempt to take advantage of Bitcoin’s price movements. The four models - machine-learning, neural networks, trend, and market regime - work in tandem to give us a holistic view of Bitcoin’s near-term risk profile.

The weakness of this approach is that it can lag the buy-and-hold strategy if Bitcoin goes on a parabolic run (any “risk-off” model days would give up extra returns). But the big upside, given Bitcoin’s huge drawdowns and volatility, is that we tend to cap our downside when Bitcoin falls, stepping aside before drowning in sorrow.

Currently, the ML model, the Trend model, and the Regime model are in agreement on the bullish side, resulting in an overall “Long” bias. However, the neural model is Risk Off, so we are not quite at unanimous yet.

Here, we do away with the complicated Elliott Wave analysis, and focus on the simple factors that matter.

First, Bitcoin has retested and bounced from its 2021 peak of $69k. That may have just marked the completion of the latest “4-year” cycle. It still invites the obvious criticism: a 14% gain on the 2021 peak, nearly five years on, is hardly worth celebrating for an asset that has fuelled so many get-rich-quick dreams.

The key moving-average for investors to watch is Bitcoin’s 50-day MA. Over a 5-year period, the 50-day MA is the best performing strategy, ahead of buy-and-hold.

While it is a high turnover strategy, it has returned 234% against 181% for buy-and-hold, with a maximum drawdown of 26% against 53% for the asset itself.

Right now, Bitcoin is above its 50-day MA (as well as its 100-day and 200-day).

Finally, if we try to glean some patterns since last October, we see that there were three selloff phases: S1, S2, and S3. S2 matched the depth of S1, but S3 fell only about 75% as far as S2. This was a touch shallower than expected and could signal that fast-money sellers have been mostly exhausted.

In terms of rallies, there have been three phases too, R1, R2, and R3. The interesting one is R3 vs R2. Right now, R3 has matched the rally of R2, but stalling.

For those who don’t like chasing, it may be better to wait for the next pullback retest of the lows near the key MAs. The pullback could happen with the broader liquidity headwind and higher long-end Treasury yields that discount long-duration assets like gold and Bitcoin.

Most of the risks above are now well understood and priced by the market. The exception is Strategy (MSTR), which deserves its own piece. I find the recent bounce very constructive for scoping out where the cycle low could be.

Therefore, I think we might be close to the start of a new cycle.

The next retest of the lows - and that retest will come in the coming weeks - could either be the low or prove that the low is in. There is a chance that if the current low breaks, we test $50k (in the scenario that the next selloff, S4, repeats S2).

From there, it is worth watching the changes in the G5 central bank liquidity, especially outside of China for the lead signal.

DISCLAIMER: This newsletter is strictly educational. Any information or analysis in this note is not an offer to sell or the solicitation of an offer to buy any securities. Nothing in this note is intended to be investment advice and nor should it be relied upon to make investment decisions. Any opinions, analyses, or probabilities expressed in this note are those of the author as of the note's date of publication and are subject to change without notice.