Skip to content
Kelly Portfolios
Search

Does bitcoin protect a portfolio?

No: bitcoin was the only thing we tested that made a portfolio’s worst fall deeper at every weight, although it also had the best standalone return for its jumpiness of anything on the panel.

If you want to own some, own it as a speculation rather than as the thing that will hold up when shares do not.

Take the fair part first, because it is real. Over the 137 months between February 2015 and June 2026, bitcoin produced the best return for its jumpiness of anything we tested. It returned 60.03% a year above cash at a jumpiness of 71.48%. Dividing one by the other gives 1.00, on a scale where the US stock market’s own long record sits near 0.45 and anything above 1 is exceptional. Trend following came second at 0.98, and nothing else on the panel was close.

That is a genuine result and we are not going to bury it. What it is not is evidence that bitcoin protects anything.

It falls harder than it rises

We measured how bitcoin moves with the US stock market separately in up months and down months, over those same 137 months. When shares rose, bitcoin moved 1.53 times as much. When shares fell, it moved 1.62 times as much.

Be careful how much weight you put on that gap. The difference between the two is nowhere near large enough to be sure of. What the numbers say, taken together, is simpler and worse for the sales pitch: bitcoin is close to a straight line against the stock market with a slope over 1.5. Holding 2% of bitcoin is, to a first approximation, the same as raising your share holding by 3% and taking on a large amount of risk that belongs to nothing else.

What it did in the worst months

This count needs a round decade rather than the panel above, so it runs over the 120 months to August 2026. Take the twelve worst of them for US shares. Bitcoin was positive in two. One of those two was May 2019, when shares dipped 6.6% in the middle of a bitcoin bull run and bitcoin gained 62.2%. The other was March 2026, up 2.2% against a 5.1% fall in shares, and that one is a real if small point in bitcoin’s favour.

Against that: March 2020, shares down 12.5%, bitcoin down 24.8%. June 2022, shares down 8.4%, bitcoin down 37.6%. April 2022, shares down 8.8%, bitcoin down 17.2%.

The result that settles the ballast question

We tested five candidates as additions to a share portfolio at weights of 1%, 2%, 5% and 10%: trend-following, gold, corporate credit with the interest-rate risk removed, commodities, and bitcoin. Four of them reduced the portfolio’s worst fall. Bitcoin is the only one that made the worst fall deeper, and it did so at every weight tested. At 10% it took the worst fall from 24.8% to 29.2% over its own window.

Something that deepens your worst loss cannot be ballast, whatever else it might be.

The rest of the honest picture

It has also been drifting towards shares. On a scale where zero means the two move independently and one means they move in lockstep, bitcoin read 0.34 against shares over those same 137 months and 0.53 over the most recent 81 of them. It is becoming more like shares, not less.

Over the five years to 22 August 2026 bitcoin returned 56.3% against the S&P 500’s 72.8%, at roughly four times the jumpiness. In the first half of 2026 it fell 33.2% while US shares returned 9.9%.

And the enormous historical return does not survive a fair test. Measured against shares at matched risk, the gap is 0.10 points a year, against a smallest detectable effect of 15.58 points. Those 137 months contain the best returns in the asset’s history and still cannot distinguish it from the S&P 500. That is a statement about how little eleven and a half years can settle, not a verdict that it is worthless.

Where that leaves it

If you want to own some, own it as a speculation, in an amount you could watch fall by three quarters without changing any other decision, and hold it where the tax treatment suits. Zero is also a perfectly defensible answer. What you should not do is count it as the thing that will hold up when everything else does not, because the record says it does the opposite.

What would change our mind

A full stock market crisis, not just a bad month, in which bitcoin holds its value while shares fall. Or enough further history to separate its return from the stock market’s at matched risk, which on current jumpiness would take decades.

Where these numbers come from

  • Monthly bitcoin prices from the St Louis Federal Reserve’s public data service. The return, jumpiness and correlation figures are measured on the 137 months from February 2015 to June 2026; the count of worst months uses the 120 months to August 2026; and the trailing five-year and half-year returns are measured to 22 August 2026.
  • Kenneth French’s US stock market series and the S&P 500 index over matched windows.
  • Our own simulation of adding each candidate to a share portfolio at four weights, each measured over its own available history.