The Safest Thing You Own

Executive summary · free edition · September 2026

Complete in itself. The finding is here, not held back to make you buy the report.

We'll start by agreeing with the client

Over one year, holding cash has usually been the right call. From the start of an easing cycle, rolling cash beat locking in a matched-maturity Treasury nine times out of sixteen since 1982. Including this one: from September 2024, $100 in cash became $104.59 against $103.97 for the note.

That is true, it is checkable, and any research that skips past it deserves to be ignored.

Now the part that changes the conversation

When cash won, it won by almost nothing — a median of under two tenths of a percentage point over one year. When duration won, it won by a lot. At ten years, the worst outcome duration ever produced still left cash 8.70 points behind per $100.

In fourteen completed ten-year windows since 1982, rolling cash has never once come out ahead. The one-year answer and the ten-year answer are different, and the question is usually asked in one-year language about a ten-year horizon.

What we removed from our own argument

The usual pitch for duration is "lock in now, before rates fall." The market is not pricing rates falling. As of 4 September the curve prices short rates about half a point higher a year from now.

And the curve is not a forecast either. Tested against its own history, the market-implied forward lost to simply assuming rates stay where they are, at every horizon we tested, and called the direction of five-year moves correctly 47.3% of the time — worse than a coin.

So the case here rests on a forty-year record, not on a view about rates. We would rather say that than lean on something that does not hold.

The second finding, which is about the products

Nineteen funds, all sold into the same slot in a portfolio. Their exposure to mortgage securities — where the return depends on how fast homeowners refinance, not on the level of rates — runs from 18.9% to 89.1%.

A factor of nearly five, decided by a fund choice the client never saw. One $230 billion fund is 89% mortgage, and its name contains neither the word "bond" nor the word "mortgage".

The specialist funds are the honest ones. The pure mortgage funds are 98% mortgage, exactly as labelled. All of the dispersion sits among the products that sound diversified.

Where the mechanism is easiest to see

Between 2022 and 2024 the Federal Reserve planned to let up to $820 billion of mortgage bonds run off its balance sheet. It received about $450 billion.

Roughly $370 billion of principal simply never arrived, because the loans underneath were too cheap to refinance. Half of all American mortgage debt carries a rate below 4%, against a market rate near 6.7%.

What we are careful not to say

• A core bond fund is not "negatively convex". Its convexity is positive. The mortgage portion strips out about a third of it. The accurate phrase is less convex.
• "Money keeps pouring into cash" is not accurate. Two of the last several months saw outflows.
• Nothing here is a recommendation. Every figure is what happened historically or what the market currently prices. This report contains no statement about what anyone should hold.

What would prove us wrong

The full report names nine markers, re-scored quarterly in public. The one that matters: a single completed ten-year window in which cash finishes ahead breaks the central claim. We would print that in the quarter it happened.

Finnacl Research publishes impersonal research, identical for every reader and tailored to no one. It is not individualised investment advice, and Finnacl Research is not an investment adviser. The author holds no position in any fund or security named. Nothing here is an offer to buy or sell. As-of dates are printed per figure in the full report; markets move.