Is direct indexing worth it?
It costs a long-term investor money at any fee, including a fee of zero, because the tax losses it harvests have almost nowhere to go.
It also more than doubles how locked into the account you end up, which is the reverse of what it promises.
Direct indexing replaces your index fund with the several hundred shares inside it, held in your own name. A manager watches them, sells the ones sitting at a loss, books the loss for tax, and buys something similar so your exposure barely changes. The pitch is that you keep the index and get a stream of tax deductions for free.
We simulated it over thirty years against the honest alternative: the same account holding one total-market fund and never selling. At a 0.09% fee it cost 0.092% a year, on the median of 400 simulated paths, and it was negative on every single one of them. At a fee of zero it still cost 0.002% a year.
That second number is the one that matters. Strip the fee out entirely and the strategy still earns this investor nothing. No cheaper provider fixes it, because the fee was never the problem.
Why the losses have nowhere to go
A capital loss has exactly two places it can be used. It can offset capital gains you realised somewhere else, with no limit. Or it can be deducted against ordinary income, capped at $3,000 a year.
That $3,000 has not moved since the 1976 and 1977 amendments that set it, and it was never linked to inflation. Had it kept pace it would be several times larger today.
Now put those two facts against the strategy. The plan is to hold the account and never sell, so there are no realised gains to offset. The only outlet is $3,000 a year. Meanwhile a $1m account harvesting at a steady 4% a year generates about $40,000 of losses annually. Over thirty years, 0.2% of every dollar of loss harvested ever produces a tax saving. The other 99.8% sits in a queue that grows forever.
And the queue does not survive you. If the account passes to your heirs, the unrealised gain is wiped clean, which is the event that makes the early savings permanent. That same event destroys the unused losses. Harvest into a queue you never use and it is worth nothing, twice over.
There is also a decay nobody advertises. In the first year a contributing account harvests about 18.8% of its value in losses. By year ten it is 4.8%, and by year thirty 4.0%. Without contributions it falls from 18.6% to 0.9%. Every headline you have read is a year-one number.
It also locks you in
After thirty years the median harvested account has 55.7% of its value sitting as untaxed gain, against 26.7% for the same account holding a fund and never selling. Harvesting more than doubles the lock-in it exists to exploit, because every harvest deliberately drives the cost base down while the account keeps growing.
That is the price of ever changing your mind. A fund holder can leave by selling one line. A direct-index holder owns several hundred separately tracked purchases that cannot be moved to another manager’s model or turned back into a fund without realising the gains. Walking away with ten years to go costs about 1.42% a year at the top rate, compared with seeing it through.
When it does pay
The honest condition is a stream of realised capital gains from somewhere else in your life: restricted stock vesting, a business sale, a concentrated holding being unwound. Break-even sits at outside realised profits of about 3.0% of the account each year if you will eventually liquidate, or 1.2% if you will hold to the end. At 5% a year of outside gains it produced 0.242% a year more than the never-selling comparison, and then it is simply the right answer.
Even then, the rival is not nothing. Selling your total-market fund at a loss and buying a different sponsor’s total-market fund does the same job for no fee and two decisions a year. Direct indexing only overtakes that route once outside realised profits exceed about 2.48% of the account annually.
One trap worth knowing whichever route you take. If you sell at a loss and something substantially identical is bought within thirty days either side, the loss is disallowed. Normally it is added to the replacement’s cost base, so you get it back later. If the replacement was bought inside your own retirement account, you lose the deduction and get nothing back. Automatic payroll purchases inside a workplace plan are the common way people trip this, and the check has to cover every account in the household.
What would change our mind
Indexing the $3,000 limit to inflation, or raising it, would change the arithmetic immediately. So would a realistic account of how much a direct-indexed portfolio drifts from its index, which no provider we approached would supply.
Where these numbers come from
- Our own thirty-year simulation of 400 market paths, using industry-standard dispersion between individual shares measured from Kenneth French’s industry data, 1996 to 2026.
- The Internal Revenue Code sections governing capital loss deduction and carryforward, wash sales, and the treatment of assets at death, read August 2026.
- Published fee schedules for direct-indexing providers, read August 2026.