Should I hold gold or bitcoin?

Gold

Too close to call Optional, small, and instead of cash.

Since 1975, the first year Americans could own it freely and the price was set by a market, gold has earned 1.75% a year over cash. That is a thin return for an asset that swings 16% a year and once fell 91% behind cash from its peak. What it has going for it is when it pays. Gold rose in 59% of the worst stock months since 2000, which only trend following beat. It is not a reliable hedge inside a crash, though: it fell 30% peak to trough during 2008, 21% of that in eleven days, and 12% in nine days in March 2020. It is protection with a lag.

Whether it adds return to a portfolio depends entirely on how you pay for it. Sell 10% of your stocks to buy gold and the portfolio grows 0.4 points a year slower. Add 10% gold on top of the stocks through a fund that borrows, and it grows 0.2 points faster. Neither number can be told from zero, and the sign flips on the funding alone. Inside the portfolio on this site, a 10% addition on top came out 0.35 points a year ahead of the portfolio without it, with a range between minus 0.2 and plus 0.9. Too close to call.

The one comparison that matters for a crash: in the worst tenth of stock months since 1926, swapping 10% of stocks into gold would have added 0.92% in the average such month. Swapping the same 10% into Treasury bills would have added 0.92%. Gold does the job of cash in a fall, at 16% a year of swings instead of none.

That is why the recommendation is small and in place of cash. If you would hold 5% cash anyway, gold is a fair substitute. Two things to know before buying now. Gold's price after inflation is higher than in all but 1.5% of months since 1975, above its 1980 and 2011 peaks. And a gold fund that borrows to hold gold on top of stocks, GDE, pays out taxable income worth 1.5 points a year, the second-largest tax cost of any fund I priced.

Which gold fund

Four funds hold bars in a vault and differ only in fee. IAUM charges about 0.09%, GLDM 0.10%, IAU 0.25% and GLD 0.40%, so the cheapest is the choice, and the famous ticker costs four times as much for the same metal. One tax point most guides miss: these funds are legally a slice of the metal, and metal is a collectible, so a gain in a taxable account is taxed at up to 28%, a worse rate than stocks get. Hold gold in a retirement account. GDE avoids the collectibles rate because it is an ordinary fund, at the price of the payout above.

Commodities

Broad commodity funds such as PDBC and DBC are sold as inflation protection, and that part is true. Commodities gained 144% across 1973 and 1974 and 10.5% in 2022. They are not crash protection. In the worst tenth of stock months since 1926 they lost 1.8% on average and rose in only 36% of those months, because a recession that sinks stocks sinks demand for oil and copper too. Silver I have never measured. Before buying any commodity fund, check whether it sends a K-1 tax form, which some do; the ones that hold their contracts through a subsidiary send an ordinary 1099, and that difference decides which account can hold it at all.

Bitcoin

No Not as protection. Zero to 2% as a bet, in a taxable account, and zero is fine.

Bitcoin has the best return for its risk of anything I measured. Between 2015 and mid-2026 it earned 60% a year over cash with swings of 71% a year. If that were the whole story it would be in every portfolio.

The problem is when it falls. Bitcoin moves about one and a half times as far as the stock market, in both directions, and its month-to-month link to stocks has tightened from 0.34 over the full period to 0.53 over the last seven years. It rose in 2 of the 12 worst stock months of the past decade. Between February and April 2025 it lost 28% while the S&P 500 lost 19%, and they hit bottom on the same day. In the first half of 2026 it lost 53% while the S&P 500 lost 9%.

Put that into a portfolio and the result is the one that decides it. Bitcoin is the only holding I tested that made the portfolio's worst fall deeper at every weight. A 10% position took a portfolio's worst fall over that period from 24.8% to 29.2%. Everything else I tested, including things that lost money, at least did not do that. Over the five years to August 2026 bitcoin returned 56% against 73% for the S&P 500, with about four times the swings.

Two smaller points. A fund that holds bitcoin on top of stocks by borrowing, RSSX, beat the same bitcoin bought by selling stocks by 0.4 points a year, which sounds like a finding and is only the stock return on the money not sold. And ether funds that promise staking income mostly do not deliver it: ETHE's 2.5% fee is larger than what staking pays, and ETHB nets about 1.9% a year, taxed as ordinary income.

If you want the bet, hold 1 to 2% in a taxable account, call it speculation, and expect it to halve now and then, as it did in the first half of 2026. BlackRock's own guidance for the fund it sells is 1 to 2%. Zero is also a good answer.

If you do buy, the spot funds all hold the same coins and differ only in fee. IBIT and FBTC charge 0.25%, BITB 0.20%, and Grayscale's mini fund, ticker BTC, 0.15%, while GBTC still charges 1.50% for the same thing. Hold it in a taxable account, where a loss can at least be set against other gains.