Can you buy insurance against a crash?

You can buy it, and on the record it has cost far more than it ever paid out; holding fewer shares does the same job for no fee.

There are funds that do nothing else, and the question was never whether they exist. It is what they cost you in the years when nothing happens, which is most years.

Start with the arithmetic, because it decides the case before any fund is named. A fund that protects you buys options, and an option is an insurance contract. The person selling it expects to make money, on average, or they would not sell it. So a strategy of continuously buying protection has a negative expected return by construction. That is not a criticism of any particular manager. It is what insurance is. The case for buying it can only ever be that the payout arrives at a moment when it is worth more to you than the premiums cost, which is a claim about timing rather than about return.

What the funds did

TAIL, the best known of them, reports in its own prospectus a loss of 8.35% a year since it launched in 2017, measured to the end of 2024, against a gain of 14.52% a year for the US stock market over that identical window. Over an identical ten-year window a similar fund, CAOS, returned 3.00% a year against the S&P 500’s 14.93%. Either way the shortfall against simply owning the index runs to twelve points a year or worse.

The calendar years are more revealing than the average. TAIL gained 6.98% in 2020, the year the S&P 500 fell 33.9% from peak to trough. It lost 13.15% in 2022, a year in which the S&P 500 returned −18.11%. A crash hedge that loses money during a bear market is a bet on the shape of the fall, not protection against falling.

Be accurate about where TAIL’s loss comes from. The fund is roughly 91% ten-year Treasury bonds and only about 5% options. A large part of that loss is the worst bond market in forty years, not option decay. That makes the product worse rather than better: you are paying 0.59% a year for a fund that is mostly Treasury bonds you could buy for 0.03%, plus a slice of the one thing with a negative expected return.

Then there is CYA. It launched in September 2021 and reported −99.16% since inception on its own sponsor’s website at the end of 2023. It took a one-for-twenty reverse share split in February 2024 and was wound up on about 14 March 2024. A total loss, on a product sold as crash protection, over a stretch that contained a bear market.

The one that gets worse when you need it

There is a family of funds that buy calm shares and sell short the excitable ones, sold on the basis that they move against the market. Measured over 1,146 months, for every 1% shares rose the trade lost 0.26%, which is exactly what it advertises. For every 1% shares fell, it lost a further 0.12%, which is the reverse of what it advertises. The swing between those two is the only statistically solid shape we found anywhere on the panel, and it points the wrong way. The mechanism explains it: the trade effectively lends borrowing power to people who cannot borrow for themselves, and a crisis is precisely when borrowing gets pulled.

The fund version charges 1.40% a year after waivers, or 1.65% before, and has returned −3.63% a year since inception against the S&P 500’s +15.41%.

The buffer funds do not escape it

A newer category promises the market with a floor under it. You give up some of the rise in exchange for a limit on the fall. Someone has now checked 102 of these funds over ten years against the obvious alternative, which is simply holding a matching mix of shares and cash. 86% of the buffer funds returned less than that mix. 70% of them fell further than it did. Not one of the 102 managed to do better than the mix on both counts at once.

That is the general principle stated as cleanly as it can be. Paying for protection usually costs more than holding less of the risky thing in the first place.

Commodities belong in the same paragraph. They are sold as the inflation protection in a portfolio, and a broad commodity fund has returned minus 2.10% a year since 2006. That is a live fund over twenty years, not a simulation.

The cheaper thing that works

Split every month since 1926 by how badly shares did and take the worst tenth. Moving 10% of a share portfolio into Treasury bills puts 0.92% back in the average such month, compared with leaving it in shares. It costs no fee, has no bad years and cannot be mistimed. Long Treasury bonds put back 0.94%, two hundredths of a point more than the bills, bought with 8.42% annual jumpiness and a 59.1% worst fall. Gold and corporate credit come in at 0.92% and 0.91%, no better than the bills. The only thing that materially beats holding cash is trend-following, at 1.08%, and it delivers that in 69% of those months rather than in one lucky outlier.

Put the two together. A 10% holding in TAIL would have cost about 2.2 points a year of portfolio growth over its life. A 10% shift into cash costs nothing and buys slightly more in the average bad month than the hedge did.

If you are not borrowing and you will not be forced to sell at the bottom, holding fewer shares beats buying protection. It converts a diffuse, uncertain future loss into no exposure at all, rather than into a certain annual bleed.

What would change our mind

Evidence that an ordinary investor sells a hedge at the bottom rather than holding it through the recovery, which is the only way an option programme pays for itself. Or a protection fund whose payout in a crisis exceeds the cost of holding it, measured over a full cycle rather than from a chosen start date.

Where these numbers come from

Fund factsheets, prospectuses and SEC filings for TAIL, CAOS, CYA and BTAL, read in August 2026.
SEC filings recording CYA’s reverse split and liquidation.
AQR’s published series for the calm-versus-excitable trade, December 1930 to May 2026.
Kenneth French’s US stock market series, 1926 to 2026, for the worst-tenth-of-months comparison.