How much bitcoin would be too much to own?

Imagine if you had most of your net worth in Bitcoin. How might this have happened? Maybe you are a Bitcoin enthusiast who has been stacking and "hodling" for a long time. Or perhaps you recently remembered a forgotten password that unlocked a bitcoin trove. In either case, the assumption is that this is a large holding as a percentage of your net worth and you have a low cost basis, which would mean substantial capital gains taxes if you sold. What should you do?

Most people in this position resolve the question with a feeling. The bull feels that selling would be a betrayal of the thesis that made him rich. The accidental holder feels paralyzed by the tax bill. Neither feeling is a good answer. There is, however, a framework that produces an actual number, and it has been sitting in the mathematical literature since 1956. Before we get to it, we need to be honest about what Bitcoin is, and about the best case that can be made for it.

Not a First-Class Asset
I have a general rule for windfalls: invest them in first-class assets for the benefit of future generations, and consume only the income (at most). First-class assets are productive. A business generates earnings. Timberland grows wood. Farmland yields crops. An apartment building collects rent. You can hold these assets forever, live on their output, and pass the principal on intact. The income is a signal that the asset is real and not just a speculation.

Bitcoin produces no income. It pays no dividend, collects no rent, earns no royalty. Whatever you eventually get out of it must come entirely from a future buyer. This does not make Bitcoin worthless but it disqualifies it from the category of first-class assets you build a family fortune on, and it puts it squarely in the category of speculation. Speculation is a legitimate activity, but it obeys different rules, and the most important rule of speculation is position sizing.

The Best Case for Bitcoin
Let me give the bulls their due, because the strongest argument for Bitcoin is genuinely strong, and it was prescient of Murray Stahl to argue this at least as far back as 2019. It goes like this: add up the market value of garbage currencies like the Brazilian real, the Iranian rial, and the Russian ruble (just a few bad currencies that happen to start with "r") and you arrive at something well over $2 trillion of combined value. Why couldn't a new, better currency (with an algorithmically fixed supply and the ability to transfer virtually) rival any one of those currencies, or even all three together?

That argument did not require Bitcoin to dethrone the dollar. It did not require hyperbitcoinization, the orange-pilling of central banks, or the citadel. It required only that Bitcoin function as a more trustworthy store of value than the monetary product of a few chronically mismanaged states. For a citizen of Buenos Aires, Tehran, or Moscow, that proposition is no longer all that speculative.

So the bull case has a floor of plausibility. The question is its ceiling.

Bitcoin Already Worked
The bulls were right - past tense. Over fifteen years, Bitcoin grew from a hobbyist experiment worth essentially nothing to a market capitalization of about $1.2 trillion, approaching the entire value of euro banknotes in physical circulation. That is an extraordinary achievement in monetary entrepreneurship. And the contrarians who bought at $10 billion, or $100 billion, or $500 billion have been handsomely paid for being right.

If you study technological adoption patterns (see Vaclav Smil, or Crossing the Chasm), the penetration rate for a new innovation follows a logistic growth, or "S," curve. The early phase is explosive: that was the contrarians' decade, when Bitcoin's share of global physical currency went from essentially zero in 2011 to roughly 4 percent by 2017, a four-thousand-fold increase in penetration. Then growth decelerates as the addressable market saturates. From 2017 to today, Bitcoin's share of world M0 has grown only about four-fold, despite the same number of years. Plotted on a multi-year moving average (to smooth through the cycles), the trajectory has the unmistakable shape of a logistic curve: near zero through about 2014, a steep climb through 2021, and a clear deceleration since.

At roughly 15 percent of world physical currency today, Bitcoin is no longer in its hypergrowth phase. It is in the part of the curve where the rate of change falls and the asymptote becomes visible.

"Bitcoin worked" and "owning Bitcoin will keep working" are therefore different propositions. The contrarian trade that took Bitcoin from a curiosity to a major monetary unit has already been made; it is no longer on offer. The question now is whether the next move, from this side of the S-curve where saturation dynamics dominate, represents the same kind of opportunity. It doesn't. 

Being contrarian is not, in itself, a strategy. Being contrarian about the right thing at the right time is. Having been right about Bitcoin in 2013 does not generate forward edge in 2025, any more than having been right about Apple in 2003 tells you what to do with AAPL today. The Kelly framework, which we'll come to in a moment, is forward-looking by construction. It does not ask whether Bitcoin was a good bet a decade ago. It asks whether the bet currently available, at current size and against current failure probabilities, is one worth taking, and at what fraction. 

The Ceiling Is Smaller Than You Think
The S-curve raises a structural question the bull case usually papers over: an asymptote against what? Enthusiasts routinely make a category error here, and it is worth dwelling on because the error inflates the apparent upside by an order of magnitude.

The popular version of the moonshot thesis compares Bitcoin to the global money supply, which might be roughly $123 trillion by the M2 measure or even $150 trillion by broader gauges. Against numbers like those, Bitcoin's current market capitalization of roughly $1.2 trillion looks small, and the implied possible upside looks like a hundred-bagger.

But M2 is the wrong yardstick. The bulk of M2 consists of savings deposits, time deposits, and money market funds, all of which are instruments that pay their holders interest. People park money in a savings account because the bank compensates them for doing so. Bitcoin pays nothing. It cannot compete for interest-seeking money, and no mechanism exists by which it ever would. Comparing Bitcoin to M2 is comparing it to a market it cannot address.

The honest comparison is physical currency, the narrowest measure of money, M0: the banknotes and coins of the world. Cash pays no interest either. People hold it anyway, for liquidity, portability, privacy, and crisis insurance. These are precisely the virtues Bitcoin claims for itself. Cash is the incumbent Bitcoin is actually running against.

All the physical currency in the world amounts to something like $8 trillion. 

That is the ceiling. And against an $8 trillion ceiling, Bitcoin at $1.2 trillion has already captured roughly 15 percent of its total addressable market. This is a genuinely astonishing achievement for fifteen-odd years of work. More astonishing than the bulls themselves seem to realize when they reach for the M2 comparison. But it carries an uncomfortable implication: the remaining upside, in the best plausible case in which Bitcoin equals every banknote and coin on Earth, is about 6.7x. 

The Distribution of Outcomes
A 6.7x ceiling would be perfectly attractive if it were the likely or expected outcome. It is neither. It is the far tail of a distribution that includes zero. Curtis Yarvin framed the long run for Bitcoin very succinctly: "One day, everything in the world economy will be priced in Bitcoin. Or, one day, Bitcoin will be regarded as the biggest bubble that ever was. There are stable equilibria between these outcomes, but they are unusual and structurally exotic." 

He put forward three failure modes: Bitcoin could be actively killed by its enemies; it could lose its energy source; or it could be outcompeted by another candidate monetary standard.

None of these is far-fetched. States have banned more popular things than Bitcoin. Proof-of-work rests on an energy-economic equilibrium that regulation or technology could upend. And monetary standards are winner-take-most competitions in which the eventual winner (a central bank digital currency, a successor protocol, something unimagined) need not exist yet.

There is also now experimental evidence bearing on the upside scenario, and it is not encouraging. In 2021, El Salvador became the first country to make Bitcoin required tender, deploying both subsidies (a $30 signing bonus, free transactions, zero capital gains tax) and coercion (businesses were required by law to accept it). After four years, the experiment was dead: crypto remittances had shriveled from 4.5 percent of the total to under 1 percent, and by 2024 just one in two hundred Salvadorans was paying with bitcoin as often as every other week. 

JP Koning, the monetary economist who chronicled the experiment at his blog Moneyness, identified the structural reason it failed. There is a very big hurdle that prevented El Salvador's one-two punch of subsidies and coercion from working: Bitcoin is intrinsically ill-suited to perform as money. The stuff is innately volatile, so risk-shy individuals don't dare hold it or use it for payments. Risk-seekers can tolerate the volatility, but they expect to be rewarded by a dramatic price rise, so they refuse to spend their bitcoins, because they could miss out on the jump. The net result is that no one, neither society's risk-seekers nor its risk-avoiders, ends up paying with bitcoins.

This matters enormously for our probability estimates. The full moonshot (Bitcoin displacing the world's physical currency) requires Bitcoin to actually function as circulating money. A national government just spent four years trying to force exactly that outcome, under maximally favorable conditions, and failed. A rational holder should weight the M0-parity scenario accordingly: it remains possible, but the best natural experiment we have gotten so far argues against it.

So the shape of the bet is: a meaningful chance of losing everything, a meaningful chance of stagnation, and a tail chance of making roughly five to seven times your money. How much of your net worth do you put on a bet shaped like that?

Kelly Tells You the Number
In 1956, John Kelly of Bell Labs derived the formula for exactly this problem: given a bet with known odds and probabilities, what fraction of your bankroll maximizes the long-run growth of your wealth? Bet less than the Kelly fraction and you leave growth on the table. Bet more and volatility itself destroys you. Overbetting a favorable gamble, repeated over time, grinds even a winning hand down to nothing. The Kelly fraction is where aggression and survival balance.

For an asset with several possible outcomes, the procedure is to choose the fraction *f* of your portfolio that maximizes the expected logarithm of wealth:

E[ln(W)] = Σ pᵢ · ln(1 + rᵢ · f)

where pᵢ and rᵢ are the probability and return of each scenario. Solve for the *f* where the derivative equals zero, and you have your allocation. The inputs are estimates, but the discipline of writing them down is most of the value of the exercise. Let us write down three honest sets.

The skeptic thinks Bitcoin is a doomed experiment: a 65% chance it goes to zero, a 28% chance it merely treads water, a 7% chance it doubles by displacing one weak currency. Expected value is sharply negative, and the Kelly fraction comes out negative as well. The formula is telling you the bet is unfavorable: the correct allocation is zero. If these are your probabilities, you should own no Bitcoin at all.

The pragmatic bull believes Bitcoin survives and grows but stops short of conquest: a 45% chance of failure, a 35% chance of stagnation, a 15% chance it displaces all three weak "r" currencies (about $6 trillion, a 5x from here), and a 5% chance it reaches full M0 parity at $8 trillion (a 6.7x). Solving the optimization yields a Kelly fraction of roughly 15% of net worth, and at half-Kelly, the standard discount practitioners apply because real-world probability estimates are soft, about 7%.

The true believer is maximally, but still rationally, bullish: only a 30% chance of failure, a 25% chance of stagnation, a 25% chance of the $6 trillion outcome, and a full 20% probability that Bitcoin swallows the world's physical currency entirely (a probability the El Salvador evidence suggests is generous). Even granting these aggressive inputs, the Kelly fraction is roughly 50%, and half-Kelly says about 25%. Note what those numbers require: a one-in-five chance Bitcoin wins outright against every banknote and coin on Earth, and only a three-in-ten chance of total failure — assumptions that strain against the experimental record.

The defensible range is therefore wide but shaped. The skeptic case gives zero. The pragmatic bull, applying the standard half-Kelly buffer, gives about 7 percent. Only the true believer, with assumptions that don't really survive scrutiny, gets to roughly a quarter of net worth at half-Kelly. To justify the very large or "all-in" allocations of the biggest enthusiasts, you would need probability estimates so lopsided (near-certainty of success, near-zero chance of failure) that writing them down would embarrass you. That is the point of writing them down.

A Second Opinion: The Market Portfolio
Suppose you distrust the whole exercise of assigning probabilities. Fair enough, there is another framework, that requires no views at all.

If you believe markets are efficient, the portfolio you should hold is the global market portfolio: all the asset holdings in the world, weighted by market capitalization. This is the allocation of the investor with no edge and no opinions. What does it contain? World stock market capitalization is roughly $150 trillion. The world fixed income (bond) market is about equal. Global real estate is perhaps $400 trillion. Cryptocurrency, in its entirety (every coin and token on CoinMarketCap) is a couple of trillion, of which Bitcoin's share (its "dominance," in the argot) is about 58 percent.

Run the division and Bitcoin's weight in the total market portfolio comes to roughly 20 basis points. Granted, world real estate in its entirety is not an investable asset class, so one might reasonably narrow the universe. But even against world stocks and bonds together, Bitcoin's weight is well under half a percent. (And against stocks alone, still under 1 percent.)

Two entirely independent frameworks: one (Kelly) that assumes you do have views and rewards you for acting on them; one (the market portfolio) that assumes you have none and tells you to hold assets in proportion to their market valuations. The first, fed honest pragmatic-bull inputs and applying the standard half-Kelly buffer, says about 7 percent. The second says 0.2 to 1 percent. The defensible range for a Bitcoin allocation, then, runs from a fifth of a percentage point to perhaps 25 percent, with where you sit in that range determined by how confident you are that you see something the market doesn't and that you can assign probabilities the El Salvador evidence does not refute. What neither framework produces, under any inputs whatsoever, is a number anywhere near "most of my net worth."

"But the Taxes"
Now return to our holder with the low cost basis. Selling means realizing an enormous gain and handing a quarter or a third of it to the government. But the tax argument, examined closely, is an argument about which loss you prefer. Bitcoin has declined 80 percent or more from its peak multiple times in its short history. If it does so again, or if one of Yarvin's failure modes materializes, the holder who refused to pay a 25% tax will have donated far more than that to the market. Taxes are the price of having won. Drawdowns are the price of refusing to admit it.

One way to translate this thinking into action would be a target allocation consistent with one's probability estimates. Sell down to it over a multi-year horizon, into strength where possible, spreading the tax burden across tax years, and redeploy the proceeds into first-class assets.

The Answer
How much Bitcoin would be too much to own? Under the framework above, allocations beyond roughly a quarter of net worth require probability assumptions that the El Salvador evidence directly undermines. A defensible bull case, sized prudently at half-Kelly, puts the ceiling closer to 7 to 10 percent. The efficient-markets baseline says the neutral weight is closer to a fifth of a percent. The asset produces no income, faces live existential risks, has demonstrably failed its one national-scale audition as actual money, and has already captured something like 15 percent of its total addressable market. The glory days of hundredfold upside are history.

The Kelly Criterion gives you a framework. The framework says: even if you love Bitcoin, keep it to a fraction of your portfolio. Exactly which fraction depends on your honest assessment of the probabilities. But "most of my net worth" is not a defensible answer under a reasonable set of assumptions.

[This essay is a framework discussion, not personalized investment advice. Probability estimates are illustrative; readers should consult their own advisors regarding their specific circumstances and tax situations.]