Direct indexing tax loss harvesting: how much tax alpha this strategy really delivers.
Providers market 1% to 2% a year. Wealthfront's own published research says 0.18% to 0.44%. Here is what the numbers actually say, what a harvested loss is worth, and why the benefit fades. Then a tool for modeling the basket yourself.
Educational only · Never places a trade
In short
Direct indexing tax loss harvesting means owning an index as individual stocks so you can sell the fallen ones, book the loss, and replace them with something similar. An index fund cannot do that because winners and losers net out inside the wrapper. The benefit is real but smaller than advertised: Vanguard tells advisors it aims for up to 1% to 2% or more a year in after-tax alpha for clients who regularly realize large capital gains, while Wealthfront's own whitepaper, backtesting February 2015 to December 2025, estimates an annual after-tax benefit of 0.18% to 0.44% of account value on the US stocks portion. Harvested losses offset capital gains first, then only $3,000 of ordinary income a year. The benefit is front-loaded and decays as cost basis falls, while the 0.09% to 0.40% fee does not. It is worth paying for when you reliably realize gains, and close to pointless when you do not.
Last updated August 2026
Tax alpha
What the published research says tax alpha is worth
Four sources, four very different numbers, and the differences are the interesting part. Every figure below comes from the organization's own published material rather than a comparison blog.
| Source | Stated benefit | Period studied | What it actually means |
|---|---|---|---|
| Vanguard Personalized Indexing | Up to 1% to 2% or more a year in after-tax alpha | 2022 simulation, data as of September 2021 | Stated as a target for "certain clients", specifically those whose portfolios regularly realize large capital gains. It is not a figure for a typical buy and hold investor. |
| Wealthfront US Direct Indexing | 0.18% to 0.44% a year of account value on the US stocks portion | Backtest, February 2015 to December 2025 | Wealthfront's own whitepaper. It comes from a 1.01% harvesting yield advantage over an ETF-only approach, converted at marginal tax rates of 18% to 44%. This is the most conservative number any provider publishes about itself. |
| J.P. Morgan Asset Management | About 0.30% a year, from scanning daily instead of monthly | 16 scenarios, 2018 to 2021 | Not the total benefit. This is the incremental gain from harvesting frequency alone, which is a useful way to see how much of the outcome is operational rather than structural. |
| Elm Wealth (skeptical view) | No annual figure given. Fees "can completely eliminate" the benefit | Analytical, no stated backtest window | Also finds that harvesting with sector ETFs captures roughly 70% of the losses full direct indexing could generate, at a small fraction of the fee. |
The gap between rows one and two is the most useful thing on this page, and it is not a contradiction. Vanguard's figure is a target for a specific client: someone whose portfolio throws off large realized capital gains year after year, so every harvested loss immediately cancels tax at the top rate. Wealthfront's figure is a backtested estimate across its actual client base at ordinary marginal rates. If you are the first investor, the high number is plausible. If you are a salaried professional with a taxable brokerage account and no regular gains, the low number is the one to plan against, and even that assumes you have something for the losses to offset.
Wealthfront's arithmetic is worth spelling out because it is unusually transparent for a provider writing about itself. Its US Direct Indexing accounts produced an annualized harvesting yield of 3.61% between February 2015 and December 2025, against 2.60% for a comparable ETF-only tax loss harvesting approach. The advantage attributable to holding individual stocks was therefore about 1.01% of harvested losses a year, not 1.01% of return. Converting harvested losses into money depends on your tax rate, and at marginal rates of 18% to 44% that 1.01% is worth an estimated 0.18% to 0.44% of account value annually. Confusing a harvesting yield with a return is the single most common way this strategy gets oversold.
The skeptical view deserves equal space. Elm Wealth's analysis found that a tax loss harvesting program run with sector ETFs captures roughly 70% of the capital losses a full direct indexing program could generate, and concluded that direct indexing fees "can completely eliminate the expected risk-adjusted benefit of Tax-Loss Harvesting for many investors." That is a fair reading of the numbers in the table. If a cheaper structure delivers most of the losses, the case for paying 0.40% rests on the remaining 30%, and that is a thin margin to defend. It gets much stronger at 0.09%, which is where the retail market has moved. Our full fee comparison across every provider is on direct indexing platforms.
Tax loss harvesting limit
What a harvested loss is actually worth
A harvested loss is not cash and it is not a refund. It is an offset, and its value depends entirely on what you have to offset it against.
| What the loss offsets | Rate it saves | Annual limit | Why it matters |
|---|---|---|---|
| Realized long-term capital gains | Up to 23.8%, the 20% top rate plus the 3.8% net investment income tax | No annual limit | This is where harvested losses do their real work. No gains means no work to do. |
| Realized short-term capital gains | Your ordinary rate, up to 40.8% including the 3.8% surtax | No annual limit | The highest value use of a loss, which is why harvesting matters most to people who trade or exercise equity comp. |
| Ordinary income, after gains are used up | Your ordinary income rate | $3,000 a year, $1,500 married filing separately | The cap that quietly deflates most of the marketing. A $60,000 loss pile with no gains behind it takes 20 years to use. |
| Nothing this year | Nothing yet | Carries forward indefinitely | Unused losses never expire, but they also never earn anything while they wait, and you keep paying the fee. |
The $3,000 line does more damage to the sales pitch than anything else in this article. Under IRS Topic 409 a net capital loss offsets ordinary income at only $3,000 a year, or $1,500 if you file married filing separately, with the remainder carried forward indefinitely. So picture the case the marketing never shows you: a $200,000 account harvests $10,000 of losses in a good year, and the owner has no capital gains anywhere. That $10,000 releases at $3,000, $3,000, $3,000 and $1,000 over four years, saving perhaps $3,500 in total at a 35% rate, while the account paid 0.40% a year, or $800 a year, for the privilege. Over those four years the fee premium roughly matched the benefit.
Now change one fact. The same investor sells a rental property and realizes a $150,000 long-term gain. That same $10,000 of harvested losses now offsets gain taxed at up to 23.8%, saving $2,380 in a single year with no cap in the way. Nothing about the portfolio changed. The value of the strategy moved by an order of magnitude because of something happening elsewhere in the tax return. This is why "should I use direct indexing" is a question about your whole financial picture rather than about the product.
One more correction worth making, because it changes how much the whole exercise is worth. Harvesting mostly defers tax rather than eliminating it. When you sell a low-basis replacement share later, the gain you avoided earlier comes back. What you gained was the use of that money in the meantime, which is genuinely valuable but is not the same as a permanent saving. It becomes permanent in three situations: the deferred gain is eventually offset by other losses, the appreciated shares are donated to charity, or they receive a step-up in basis at death. Treating deferral as pure savings is the most common overestimate in this whole category.
Harvest depletion
Why tax loss harvesting stops working after a few years
Every harvest lowers your cost basis. You sell a position that is underwater, take the loss, and buy a replacement at today's price. Do that repeatedly through a market that rises over time and you gradually convert a portfolio of mixed lots into a portfolio where nearly everything is worth more than you paid for it. At that point there is nothing left to sell at a loss, and the account has become an ordinary appreciated portfolio that happens to hold 300 positions instead of one fund.
The shape of this is consistent across every provider that discusses it. Harvesting is heaviest in year one, when a portfolio funded with cash has lots sitting right at their purchase price and any dip puts them underwater. It stays useful for the first few years, particularly through volatile stretches. Then it thins out. In the later years most of the remaining harvest comes from newly purchased lots: reinvested dividends, fresh contributions and shares received through corporate actions, all of which enter at current prices and can therefore fall below them.
The fee does not decay. That asymmetry is the whole problem. A 0.40% account costs the same in year eight as it did in year one, while the thing you are paying for has quietly shrunk. Anyone evaluating a long holding period should assume the benefit is front-loaded and the cost is flat, then decide whether the early years justify the later ones.
Harvesting frequency partly offsets this, and it is measurable. J.P. Morgan Asset Management found that scanning daily rather than monthly delivered on average about 30 basis points of additional annualized tax alpha, across 16 scenarios between 2018 and 2021. A dip that recovers within a month is invisible to a monthly process and harvestable by a daily one. It is a fair question to ask any provider how often they actually look, because the answer is worth roughly a third of a percent a year and almost nobody advertises it.
What to do once the losses run out is a longer subject, and we walk through the options that genuinely work in what to do when you have no losses left to harvest.
Wealthfront, Fidelity, Schwab, Frec and Parametric tax loss harvesting
Who harvests losses for you, and what it costs
Minimums and fees verified against each provider's own published pages in August 2026. Where a firm publishes nothing we have left it blank rather than repeating a number from a comparison site.
| Provider | Minimum | Annual fee | Harvests losses? | Best for |
|---|---|---|---|---|
| Wealthfront S&P 500 Direct | $5,000 | 0.09% | Yes, at the individual stock level | The cheapest managed harvesting available to an individual |
| Wealthfront US Direct Indexing | $100,000 | 0.25% advisory, harvesting included | Yes, at the individual stock level | The tier the published 0.18% to 0.44% research describes |
| Fidelity Managed FidFolios | $5,000 | 0.40% index tracking, 0.70% active | Yes, on the sampled basket | People already custodying at Fidelity who want one relationship |
| Frec Classic | $20,000 to $50,000 | 0.09% to 0.35% | Yes, daily scanning across 25 index choices | Low ongoing cost with index choice beyond the S&P 500 |
| Schwab Personalized Indexing | $100,000 | 0.40%, 0.35% above $2M | Yes, on the sampled basket | Six-figure accounts that also want values or sector screens |
| Parametric Custom Core | Not published, advisor only | Negotiated | Yes, this is the institutional standard | Advisor-led households with complicated tax situations |
| Aperio (BlackRock) | Not published, advisor only | Negotiated | Yes, alongside deep screening | High-net-worth clients with specific exclusion lists |
| Vanguard Personalized Indexing | Not published, advisor only | Not published | Yes, this is the product's stated purpose | Clients of an advisor already on the Vanguard platform |
| Betterment | Not applicable | 0.25% Automated, 0.65% Premium | On ETF portfolios, not at the stock level | Retail direct indexing announced for 2026, still not shipped |
| A plain S&P 500 ETF | One share | 0.03% expense ratio | No. Winners and losers net out inside the fund | Almost everyone with no meaningful gains to offset |
| Indexes (this site) | No minimum | From $12 a month | No. We are software, we place no trades | Modeling and backtesting the construction before you commit |
Two things in that table decide most outcomes. The first is the fee spread: 0.09% at Wealthfront S&P 500 Direct and the cheaper Frec Classic indexes against 0.40% at Fidelity and Schwab. On a $100,000 account that is a difference of about $310 a year, which is a large fraction of the conservative published benefit estimate. The second is that Betterment, despite winning awards for tax loss harvesting on ETF portfolios, still does not offer stock-level direct indexing to retail investors as of August 2026, which we re-checked against its pricing page. Per-provider detail sits on our Wealthfront direct indexing, Fidelity direct indexing, Frec, Schwab Personalized Indexing and Betterment pages, and the two cheapest are compared directly in Frec vs Wealthfront direct indexing.
Tax loss harvesting rules
Three rules that quietly delete the benefit
The wash sale rule
Under IRC Section 1091, buying the same or a substantially identical security within 30 days before or after selling at a loss disallows that loss for now. Note it runs in both directions, so the window is 61 days in total, not 30. The disallowed loss is added to the basis of the replacement shares, so in a taxable account you get it back eventually.
It follows you across accounts
The window is not per account. If you sell at a loss in your brokerage account and your spouse's account buys the same security inside the window, it is a wash sale. Automatic dividend reinvestment and scheduled 401(k) contributions cause most of these, because nobody is thinking about them. No platform can see the accounts it does not custody.
Revenue Ruling 2008-5
This is the one worth memorizing. If the replacement purchase happens inside an IRA, the loss is disallowed and your IRA basis is not stepped up. An IRA has no outside basis for the deferred loss to attach to, so it is gone permanently rather than delayed. A wash sale in a taxable account costs you timing. A wash sale into an IRA costs you the deduction forever.
There is a fourth constraint that is not a rule but behaves like one: tracking error. Every time you sell a holding and buy a similar one to stay invested, your basket drifts a little further from the index it is supposed to represent. Do it a few hundred times and you own something that is no longer really the S&P 500. Providers manage this with explicit tracking risk budgets, which is also why most of them do not harvest as aggressively as they theoretically could. The full rule set, with the IRS citations, is in tax loss harvesting rules.
Tax loss harvesting strategy
Four questions that settle whether this is worth paying for
Do you realize capital gains?
This is the gate, and it is not a preference. Losses are worth money against realized gains, and only $3,000 a year against ordinary income. If you sell a business, exercise equity comp, rebalance a large taxable account or sell property, the answer is yes and everything else is detail. If you buy and hold and realize nothing, an ETF at 0.03% is the better product.
Is the account taxable?
Direct indexing belongs in a taxable account or nowhere. An IRA or 401(k) has no capital gains tax to offset, so harvesting is worth exactly nothing inside one, and a purchase there can permanently kill a loss you harvested outside it.
What fee are you actually paying?
At 0.09% the premium over an ETF is about 0.06% and the arithmetic works easily. At 0.40% the premium is about 0.37%, which sits right on top of the conservative 0.18% to 0.44% benefit estimate. Same strategy, completely different verdict, decided entirely by the fee.
How long will you hold it?
The benefit is front-loaded and the fee is flat. A five year horizon captures the productive years. A twenty five year horizon means paying full price for a long tail of thin harvests, unless you keep adding new money that brings fresh cost basis with it.
Answering those four honestly sends a lot of people to a plain index fund, which is the correct outcome rather than a failure of the strategy. Direct indexing is a tax tool, and a tax tool is only worth what it saves. The people it genuinely serves, those with large recurring realized gains in high tax brackets, tend to know exactly who they are. Whether the strategy suits you at all is covered in is direct indexing worth it, and the mechanism itself in what direct indexing is.
Being straight about this
Indexes does not harvest tax losses
What we do not do
We do not custody assets, place trades, track cost basis, scan for harvestable lots or file anything, and we are not a registered investment adviser or a tax advisor. If you want losses harvested for you, one of the providers in the table above is the right answer. Everything on this page is educational only.
What we do instead
We are the modeling layer that runs before the decision. Define the basket, set the weights and the rebalancing rule, then backtest that exact construction over real market history and track it as a named index against the S&P 500 or BTC. No minimum, no account to fund, from $12 a month.
Why that helps here
Because the tracking error question is answerable in advance. Before you pay anyone to run a sampled basket, you can see how a 30, 50 or 100 stock construction actually behaved against the full index over real history, and decide whether the tax benefit is worth the drift.
If you are considering running the harvesting yourself, be realistic about the workload rather than the concept. The concept is straightforward and the fractional share support at every major broker makes a 30 to 50 name basket practical. The work is the part people underestimate: tracking basis lot by lot, scanning on a schedule, choosing replacements that are not substantially identical, and keeping wash sales clear across every account you and your spouse control anywhere, including IRAs and automatic dividend reinvestment. That operational load is exactly what the 0.09% to 0.40% pays for. Doing direct indexing yourself walks through it in detail.
Questions
Direct indexing tax loss harvesting, answered
How does direct indexing tax loss harvesting work?
You own the individual stocks of an index instead of a fund that tracks it, so you can sell the specific holdings that have fallen below their purchase price, book the loss on your tax return, and buy something similar to keep your market exposure roughly intact. An index fund cannot do this because the winners and losers net out inside the wrapper. The loss offsets your capital gains first, then up to $3,000 of ordinary income a year.
How much tax alpha does direct indexing actually generate?
Less than the marketing implies. Vanguard tells advisors it aims for up to 1% to 2% or more annually for clients who regularly realize large capital gains. Wealthfront's own whitepaper, backtesting February 2015 to December 2025, puts the estimated annual after-tax benefit at 0.18% to 0.44% of account value on the US stocks portion. Both can be true, because they describe different investors. The high number assumes you have large gains to offset every year. Most people do not.
Is the tax alpha from direct indexing worth the higher advisory fee?
It is worth it when you reliably realize capital gains, and close to pointless when you do not. Compare the two numbers directly: paying 0.40% instead of 0.03% for an ETF costs about 0.37% a year, and the conservative published benefit estimate is 0.18% to 0.44%. On those figures the strategy roughly breaks even before you account for having gains to offset. At 0.09% the premium is about 0.06% and the arithmetic gets much friendlier. Fee level, not the strategy, decides this.
What is the minimum balance for direct indexing to be worth it?
The account minimums are now as low as $5,000 at Wealthfront S&P 500 Direct and Fidelity Managed FidFolios, but a minimum you can meet is not the same as a balance worth it. The real gate is whether you have realized gains to offset. Someone with $30,000 and a large gain from selling a rental property gets immediate value. Someone with $300,000 in a buy and hold portfolio and no gains for years gets a slow drip against the $3,000 ordinary income cap while paying the fee the whole time.
What are the tax loss harvesting rules?
Three matter most. Under IRC Section 1091 you cannot buy the same or a substantially identical security within 30 days before or after the sale, which is a 61 day window in total. The window applies across every account you control, including your spouse's. And under Revenue Ruling 2008-5, if the replacement purchase happens inside an IRA, the loss is disallowed and your IRA basis is not stepped up, so the deduction is lost permanently rather than deferred.
Why does tax loss harvesting stop working after a few years?
Because harvesting lowers your cost basis every time you do it. You sell a position below cost, book the loss, and buy a replacement at today's price. Repeat that through a rising market and eventually almost every lot you hold is worth more than you paid, so there is nothing left to sell at a loss. The benefit is front-loaded into the early years, especially in volatile markets, while the fee stays exactly the same. This is the part providers rarely put in the pitch.
Are there ways to get tax alpha when the cost basis drops too low?
The honest answers are limited. Adding new cash gives the account fresh lots at current prices, which is the cleanest fix and the one most providers suggest. Reinvested dividends do a small version of the same thing continuously. Some platforms offer long short or leveraged structures that manufacture more harvestable positions, at meaningfully higher fees and complexity. And charitable giving of appreciated low-basis shares, or holding them for a step-up at death, addresses the accumulated gain rather than generating more losses. What does not work is waiting for the account to reset on its own.
Does tax loss harvesting actually save you money or just defer it?
Mostly it defers. Selling low-basis replacement shares later realizes the gain you avoided earlier, so a large part of the benefit is the time value of paying tax later rather than paying less tax overall. It becomes a permanent saving in specific cases: when the deferred gain is eventually offset by other losses, when appreciated shares are donated to charity, or when they receive a step-up in basis at death. Treating deferral as though it were pure savings is the most common way people overestimate this strategy.
Can I do tax loss harvesting myself without a platform?
Yes, on a basket of a few dozen names. Fractional shares make a 30 to 50 stock construction practical at any zero-commission broker, and the loss harvesting opportunity is real. What you take on is the work: tracking cost basis lot by lot, scanning for underwater positions on a schedule, choosing replacements that are not substantially identical, and keeping wash sales clear across every account you and your spouse hold anywhere, including IRAs and automatic dividend reinvestment. That last one causes most DIY mistakes.
Is there more harvesting opportunity in the Russell 1000 than the S&P 500?
In principle yes, because a broader index holds more names, more dispersion between them and more small and mid cap volatility, which produces more positions below cost in any given period. In practice the difference is smaller than it sounds, because the added names carry less weight, and tracking a wider index with a sampled basket introduces more tracking error. Frec offers the widest published index menu among retail platforms if you want to make this choice deliberately.
Does tax loss harvesting work in an IRA or 401(k)?
No, and paying for it there is a straightforward mistake. Tax-advantaged accounts have no capital gains tax to offset, so a harvested loss is worth nothing inside them. Worse, a purchase inside your IRA can disallow a loss you harvested in your taxable account under Revenue Ruling 2008-5, permanently. Direct indexing belongs in a taxable account or nowhere.
What is tax loss harvesting software?
It generally means one of two things. Managed platforms like Wealthfront, Frec, Fidelity and Schwab run the harvesting for you inside an account they custody, which is a service rather than software. Separately there are portfolio and tax tools that flag harvesting candidates and track basis while you place your own trades. Indexes is neither: we model and backtest the construction of a weighted basket, and we place no trades and track no cost basis.
Go deeper
See how the basket behaves before you pay for the tax management
Build the weighted index, backtest the construction over real history, and track it against the S&P 500. No minimum, no account to link, no trades placed. Educational and informational only, not tax or investment advice.