What actually protects you in a crash?

Over 1990 to 2025 a plain world stock index took $10,000 down to $4,730 at its worst; a 60/40 mix, 60% stocks and 40% US government bonds, fell to $7,280; the cautious portfolio, with 40% in inflation-protected government bonds, fell to $8,190. These are simulated portfolios built from index data, not fund records. Protection has a price. In the separate 1929–2025 study, moving 10% of the money from stocks to cash cost about 0.77% a year of growth. That is what this history gave up, not a fixed future price for protection. The table below asks how often each holding rose in bad months. It cannot tell you what a small options hedge would do inside a whole portfolio.

Only trend following clearly beats cash in the worst months, and it is held on top of stocks in the Plus trend portfolio. Long bonds add two hundredths of a percent a month over cash, and gold matches that. Commodities and bitcoin do worse than cash. The windows differ, so read the start years.

Long US government bonds

They paid in every growth crash and lost in all three inflation shocks. In the worst stock months since 1926 they added 0.94% back against 0.92% for cash. That sliver cost swings of 8.4% a year, a worst fall of their own of −59.1% against cash, and a 40% loss through the late 1970s. Adding 20% of borrowed long bonds on top of the portfolio read +0.34% a year in the separate 1929–2025 study, inside a range that includes zero, and all of it was earned between 1981 and 2020. Since 2020 stocks and bonds have mostly fallen together. Not added.

Inflation-protected bonds

Inflation-protected and plain US government bonds are one idea: eighteen funds move together at +0.76 to +0.85 out of 1. Hold the inflation-protected kind because your future spending rises with prices, and take any crash protection from holding fewer stocks. A ten-year inflation-protected bond pays 2.44% a year above inflation, the most since 2008. Which fund: SCHP and TIP hold the same bonds and SCHP costs a sixth as much; the cautious portfolio holds 40% of its money in it and says which account to hold it in.

Tail-risk funds

TAIL holds government bonds and buys put options that can pay when stocks fall. Its bonds can lose value as rates rise, which eats into what the options pay. CAOS mixes options with a stock holding it changes over time, so it is built differently. Their TAIL and CAOS prospectuses describe those choices.

In our exploratory test, a small TAIL holding softened the February–March 2020 crash against holding the same amount in cash-like government bills. Over the full period below, the bills ended with more money and a slightly smaller worst month-end fall. What a hedge does in one crash and what it costs to hold for years are two separate questions.

This is a cumulative difference, not a yearly fee. Resetting the holdings quarterly instead of yearly did not change the conclusion. CAOS was close to bills in its shorter April 2023–March 2026 equity test; that window contains neither the 2020 crash nor the 2022 inflation shock. Results for a small holding funded from bonds in Cautious were also close to bills for CAOS. TAIL gave up more growth.

Neither fund earns a place here on these tests. A different size, a different spending need, or a rule for spending the gains could change that. Monthly returns cannot measure what a hedge does at the low point of a single day, and CAOS changed how much stock it held during its record.

Buffer funds

A buffer fund caps your gains for a year in exchange for a cushion on your losses: a typical one charges 0.79% for a 9% cushion against an 18% cap, and measures itself against the S&P 500 without its dividends. Priced on 1,183 twelve-month windows since 1926, the cap-and-cushion package is worth −2.4% to −4.1% a year, and the funds' own record of −4.1% a year lands inside that range. You pay for the cushion with the gains, every year, whether or not the fall comes.

Catastrophe bonds

A catastrophe bond is a loan to an insurer that is written off after a named hurricane or earthquake, the only holding here whose losses come from weather rather than markets. The funds returned 3.31% a year, 2017 to 2025, after fees, about 1% a year over cash, and they can lose 10% in a week, as they did after Hurricane Ian in September 2022. The price is the problem: the bonds paid 2.2 times the expected loss, mid-2026 against 4.9 times in January 2023, the thinnest on record. Hold 0% to 3% in a retirement account, or wait; the question reopens at 3.5 times, a number published weekly.

Cash

Cash is the surprise. Moving 10% of stocks into cash-like government bills added 0.92% back in the average worst stock month since 1926, as much as gold and almost as much as long bonds, for no fee and with no fall of its own. A three-month government bill paid 3.88% in August 2026, about 0.2% a year above the past year's inflation. In an ordinary brokerage account, BOXX (0.19%) pays a cash-like return as a rise in price, so the tax falls when you sell. Cash cost about 0.77% a year of growth in that 1929–2025 study: the cheapest protection, and not a free one.

Things that look like protection

Corporate bonds are government bonds plus a slice of stock risk, and they move with government bonds at 0.83 out of 1. Strip out the interest-rate risk and what is left follows them at only +0.02 out of 1, but that leftover lost 13% in 2008 while government bonds gained 13%. Dividend funds move with the market at 0.84 out of 1 and property funds at 0.82. What a bond fund pays out is taxed, so which account it belongs in depends on the fund and on your own tax rates.