How Much Bitcoin Should Be in Your Portfolio?

What every major house actually says

Nine named positions, most recent first. Two things matter when reading this table and are usually lost in coverage: several of the widest figures cover crypto generally rather than Bitcoin, and one of the most-quoted numbers is not a recommendation at all.

The detail, and where each number comes from

Why the published range runs from zero to forty

Most of the spread is explained by three differences that have nothing to do with conviction about Bitcoin.

• Bitcoin or crypto. Morgan Stanley's 4% and Edelman's 40% are crypto allocations. VanEck's widely quoted 6% is 3% Bitcoin and 3% ether. Reported as Bitcoin figures they overstate the Bitcoin position, sometimes by half.
• Recommendation or hindsight. ARK's 19.4% answers what would have been optimal in 2023. That is a different question from what to hold now, and it is the single most misquoted figure in this debate.
• Who is speaking. A private bank publishing guidance carries fiduciary and suitability obligations across every client at once. An advocate does not. Both can be honest and land far apart.

What the optimizer says on J.P. Morgan's own numbers

Portfolio Lab runs on J.P. Morgan's 2026 Long-Term Capital Market Assumptions: Bitcoin at 15.0% expected return and 42.5% volatility, against a 3.1% risk-free rate. Sweeping the Bitcoin weight and recording the Sharpe ratio at each step gives a peak for each starting portfolio.

The conservative portfolio has a genuine interior peak at 10.5% and gets worse above it. The balanced portfolio peaks at 22.5%. The aggressive one was still improving at 30%, where we stopped looking, which is itself a warning about how far this logic runs if you let it.

The mechanism is correlation, not the return assumption alone. Bitcoin's estimated correlation is roughly 0.35 to US large-cap equities and slightly negative to bonds, so it lowers total portfolio variance for a while even though it is individually the most volatile thing in the book. Change the 15% return assumption and the peak moves a long way, which is why we publish the assumption alongside the answer.

Why the optimizer and the banks disagree

The optimizer budgets return per unit of volatility. Institutions budget how much of the portfolio's risk any single holding is allowed to carry. Those are different constraints and they give different answers.

Here is the same balanced 60/30/10 portfolio, measured by risk rather than by capital. The middle column is Bitcoin's share of total portfolio volatility, not of the money.

Notice that our figures independently confirm J.P. Morgan's own warning. They state that a modest Bitcoin allocation contributes about two to three times its weight in risk. Running their assumptions through our engine gives 1.9 times at a 4% weight and 2.4 times at 10%. We reach the same conclusion from the same data and then draw a different line, because a Sharpe ratio has no opinion about concentration and a fiduciary does.

The house whose data we use does not recommend it

Worth stating plainly, because it is the strongest argument against the number this site produces. Portfolio Lab runs on J.P. Morgan's capital market assumptions. In February 2026 the J.P. Morgan Private Bank published its view that Bitcoin does not belong in a core allocation, that it “has not served as a reliable diversifier during volatile equity markets”, and that during equity drawdowns Bitcoin fell 93% of the time against gold's 55%.

Those are compatible positions rather than a contradiction. Their capital market assumptions are estimates of return, volatility and correlation. Their allocation guidance is a judgment about suitability, drawdown tolerance and concentration on top of those estimates. We use the first and disagree with nothing in the second. But anyone reading an optimizer output of 22% should know that the firm supplying the inputs would not put a client there.

What to actually do with this

• Size by risk share, not by capital. Decide what fraction of your portfolio's risk you are willing to give one holding, then read the weight off the table above. Most people who say “10% feels about right” have not registered that it is a quarter of their risk.
• Treat the return assumption as the main lever. The 15% figure drives almost everything. At 8% the optimizer's answer collapses toward the institutional range. Test your own number rather than inheriting ours.
• Rebalance, or the decision unmakes itself. An unrebalanced 5% position becomes a large one after a strong run, and the risk share moves faster than the weight.
• Expect the drawdowns. Bitcoin has had several declines beyond 75%. A position you would abandon at the bottom is worse than no position.
• Hold less in drawdown than in accumulation. Once you are withdrawing, a bad early sequence forces selling into weakness. See how much Bitcoin to retire on.

What this analysis cannot tell you

Every figure here rests on forward-looking estimates that are, unavoidably, opinions with decimal places. Bitcoin has roughly fifteen years of price history, much of it in conditions that no longer apply. Correlations are unstable and tend to rise toward one in exactly the crises where diversification is supposed to help, which is the substance of J.P. Morgan's objection. Mean-variance optimization also assumes returns are normally distributed, which for Bitcoin they emphatically are not.

None of that makes the exercise useless. It makes the sensitivity analysis more important than the point estimate, which is why the tool lets you change the assumptions rather than only showing you ours.

Institutional positions reviewed August 16, 2026 and re-checked quarterly; each row links to its source. Portfolio figures computed from J.P. Morgan 2026 Long-Term Capital Market Assumptions on the same date. This is analysis, not investment advice.