Skip to content
Kelly Portfolios
Search

The answer

Do the certain things first

Cut your fees. Use whatever tax-advantaged accounts you actually have. Hold each fund in the account that taxes it least.

Against the portfolio a normal person would otherwise be holding, those three moves are worth about 1.09% a year. On a hundred thousand dollars that is roughly a thousand dollars a year, for an afternoon of paperwork. They come out of fund contracts and tax statute rather than out of a forecast, so nobody has to be right about markets for them to pay. And you would know inside about 3.5 months whether you got them right.

Everything else on this site is a bet of some size. This part is arithmetic.

If you already own one cheap index fund in one account and never touch it, your number is close to zero and we have nothing to sell you. Most people sit between the two, which is why the honest outer bound on the package runs 0.04% to 2.70% a year.

Three yardsticks, never added

Fees, taxes and account placement, against the portfolio you already own
1.09% a year

Assembled from fund filings and tax statute rather than from a backtest, and measured against the portfolio you would otherwise have held. It was 0.89% before the ledger was revised. The largest single line is the capital-gain distribution an active mutual fund makes and an exchange-traded fund does not.

Source-reproduced as of 2026-08-17 Structural and tax-aware edges
Defensible edge over a cheap index fund
0.054% a year
Interval −0.92% to +0.83%

A coin flip. The range crosses zero, so we cannot even say the sign is positive. This answers a different question from the 1.09% measured against the portfolio you already own, and the two are never added.

Unresolved as of 2026-08-17 Where outperformance can come from
Gap between what funds returned and what investors in those funds actually earned
0.15% a year
Interval 0.05% to 0.60%

Morningstar, ten years to 2024. A third yardstick, and the answer to a third question. It may be shown beside the other two and never added to them.

Exploratory as of 2026-08-17 The behaviour gap is a different benchmark

Better than what

Better than what, exactly

Any claim about a portfolio has to finish the sentence "better than what". Three endings make sense, and they are three separate questions with three unrelated answers.

Against the portfolio you already hold, the answer is the number above, and it settles fast because only 0.46% of wander sits around it. Against a cheap total-market index fund, it is 0.054% a year, which is about five dollars a year on ten thousand and too small for us to tell from zero. Against what a typical fund investor earns after buying late and selling into falls, Morningstar's ten-year gap is 0.15% a year.

Those three never add. Adding them is the most common way a result here has turned out wrong, and we know because we did it to ourselves. An old headline of ours for the account-placement decision summed lines measured against three different yardsticks and booked an avoided cost as though it were a saving. The defensible figure, against something a real investor could build, is +2 to +7 basis points a year. The research code now raises an error rather than adding across yardsticks.

The bet

What counts as a bet, and how big

Here is the part that costs us something to write.

Add up everything this project has measured that might beat a plain cheap index fund, keep only what survives its own error bars, and what is left is 0.054% a year. The wander around it is 3.13% a year. In a normal year you would land within about three percentage points either side of the index, for no reason except that you own different things. Call it sixty times as much noise as signal. Over thirty years your chance of finishing ahead of the index is 54 in 100.

So it is a coin flip. We can't tell you this beats a cheap index fund.

That does not mean a cheap index fund cannot be beaten. It means we cannot show you that we can beat one, on the data anyone has. A site telling you otherwise is either measuring against something easier or has not checked how long its own claim would take to prove.

Four months against thirty-one years, on the same rule applied twice. That gap is the whole reason to take the fees and the tax before you take a view on anything.

Next

Where to go from here

The checklist puts the whole thing in order, and every step says up front who it is for. The placement tool runs the account-ordering arithmetic at your own tax rates and your own account sizes, which is the only form of that answer worth acting on.

The portfolio is one concrete construction, with the parts that are bets named as bets. Stacking covers why owning more good ideas stops helping sooner than you would think, and why how you pay for an addition matters more than what you add. What doesn't work covers market timing, crypto as ballast, crash insurance and direct indexing, each turned down on a measurement and each with the number that would reopen it.

Sources

Where this comes from

Fee, tax and account figures are read from fund filings and tax statute rather than from a backtest, checked on 2026-08-23. The comparison against a cheap index fund comes from where outperformance can come from, and the wander in it is assumed rather than measured, so every probability here is a best case. How sure we are lists every error this project has caught in itself.