Automated tax-loss harvesting: is it worth the robo fees?
Automated tax-loss harvesting is one of the most persuasive features in modern wealth management. The promise is simple: when an investment falls, the software sells it, captures the loss, and replaces it with a similar holding.
Jocelyn Davenport·Updated: August 31, 2026·15 min read
You remain invested, but the loss may offset capital gains elsewhere—or reduce up to $3,000 of ordinary income on a U.S. tax return each year.
The catch is that the tax benefit is not the same thing as free investment performance. It is available only in taxable accounts, it depends on your tax situation, and it often works by moving taxes into the future rather than making them disappear. Meanwhile, the robo-advisor still charges its management fee every year, whether the market gives the algorithm something useful to harvest or not.
So, is automated tax-loss harvesting worth the robo-advisor fee? Sometimes. But the answer usually depends less on the sophistication of the software than on the unglamorous details of your portfolio: its size, age, turnover, tax bracket, cash flows, and accumulated gains.
The math behind the marketing: what does the 1.175% claim mean?
The headline number comes from Wealthfront, which says its median tax-loss harvesting benefit is equivalent to 1.175% annually. The comparison is designed to be memorable: 1.175% is 4.7 times the company’s 0.25% annual management fee.
That framing makes the feature look like an obvious bargain. Pay a quarter of a percentage point, receive a tax benefit several times larger, and let the software handle the paperwork. The user journey is satisfyingly short.
But the number needs to be placed in the right mental folder. A claimed tax-loss harvesting benefit is not necessarily a permanent, after-tax increase in wealth. It may include tax deferral, depends on the investor’s circumstances, and can be much higher in some market periods than others. A benefit observed in a particular group of accounts is not a universal annual coupon attached to every automated portfolio.
Independent estimates are more restrained. Industry analyses commonly place the long-term benefit of tax-loss harvesting somewhere around 0.2% to 0.4% per year. That range is still meaningful, particularly for a large taxable portfolio held for many years. But it is no longer an automatic argument that the strategy pays for a 0.25% robo fee several times over.
The difference is a classic example of choice architecture. Platforms foreground the most attractive version of the outcome, while the cost appears as a small recurring percentage that feels abstract. The user sees the potential tax alpha; the platform fee quietly compounds in the opposite direction.
A useful way to think about the calculation is:
| Component | What it means for the investor |
|---|---|
| Robo-advisor management fee | Commonly about 0.25% annually at major platforms such as Betterment and Wealthfront |
| Fund expense ratios | Often add roughly 0.05%–0.15% annually |
| All-in recurring cost | Commonly around 0.30%–0.45% before considering taxes |
| Potential long-term TLH benefit | Often estimated around 0.2%–0.4% annually, but highly dependent on circumstances |
| Taxable-account requirement | The strategy does not create this tax benefit inside an IRA or 401(k) |
| Timing of the benefit | Losses may offset gains now, while the tax cost can reappear when replacement assets are eventually sold |
This is not a clean contest between 1.175% and 0.25%. It is a comparison between a variable, situation-dependent tax outcome and a predictable annual expense.
The robo fee is certain. The tax benefit is conditional. Any serious comparison has to start there.
Why tax-loss harvesting benefits decay as portfolios mature
Tax-loss harvesting tends to be most productive when a portfolio is young, receiving regular deposits, and moving through volatile markets. New contributions create fresh lots with different purchase prices. Market declines can push some of those lots below their cost basis, giving the software something to sell and replace.
The early years can therefore produce excess returns in the form of tax savings or deferral. Historical simulations have found that the effect can be substantial under favorable conditions. A 2020 CFA Institute study examining the period from 1926 to 2018 reported average annual outperformance of 1.08% compared with a buy-and-hold approach. After transaction costs and the application of wash-sale rules, the figure fell to 0.82%.
That is useful evidence that tax-loss harvesting can matter. It is not evidence that every investor should expect 0.82% each year.
The strategy’s economics change as the portfolio ages. Assets that have appreciated for a long time may carry large unrealized gains. Selling them at a loss becomes less common, especially if the portfolio is broadly diversified and the market has generally moved upward. The account may still be well managed, but there are fewer tax losses available to harvest.
This is where long-term investors can experience a form of benefit decay:
1. The portfolio starts with many low-cost opportunities. Early deposits have not yet accumulated substantial gains, and market volatility can create losses across several tax lots.
2. Replacement holdings recover. After a harvested loss, the replacement investment may appreciate. The loss has been used for tax purposes, but the new holding now has a lower basis than the original position would have had.
3. The portfolio accumulates embedded gains. As assets rise over time, selling appreciated positions can create future capital gains. The earlier tax benefit may eventually be balanced by a larger taxable gain when the investment is sold.
4. The supply of harvestable losses narrows. A mature portfolio with deep unrealized appreciation may produce fewer useful losses, particularly during long periods when markets rise with limited interruptions.
5. The investor’s tax horizon matters. If you intend to sell assets soon, move to cash, or leave the portfolio to heirs, the value of deferring taxes may differ from the value for someone who expects to remain invested indefinitely.
During volatile markets, early portfolio years may produce excess returns in the neighborhood of 0.3% to 0.9%, according to estimates in the research. In a mature, highly appreciated portfolio, the benefit can fall to 0.1% per year or less.
That last figure is not a failure of the algorithm. It is a reminder that tax-loss harvesting cannot manufacture losses from an account that mostly contains appreciated assets. No amount of elegant interface design changes the tax lots underneath it.
The hidden drag: wash-sale rules and the cost of staying invested
The attractive version of tax-loss harvesting sounds frictionless. Sell one fund, buy another, preserve market exposure, collect the tax loss. In practice, the replacement security has to be chosen carefully.
Under U.S. wash-sale rules, a loss can be disallowed if you buy the same or a substantially identical security within the relevant period around the sale. The point of the rule is to prevent investors from claiming a tax loss while effectively holding the same position. Automated platforms therefore tend to replace a sold ETF or fund with another investment that is similar enough to maintain the portfolio’s intended exposure, but different enough to avoid being treated as substantially identical.
That creates a trade-off. A replacement fund may track a slightly different index, have a different portfolio composition, or behave differently in a sharp market move. The difference may be small, but it is not always zero. Tax management introduces another layer into the investment decision, and another place where the user has to trust the platform’s assumptions.
The larger issue is that wash-sale tracking does not stop at the robo-advisor’s account. The rule can involve purchases in other accounts, including accounts held at another brokerage or, in some circumstances, purchases made by a spouse. An algorithm operating inside one platform may not have a complete view of all your transactions.
This is one of the least visible sources of cognitive load in automated investing. The product appears to take responsibility for the strategy, but the household may still be responsible for the data the software cannot see. A user who holds a broad-market ETF in a separate brokerage account and automatically reinvests dividends may unintentionally interfere with a harvest in the robo account.
The same problem can arise through recurring purchases. A replacement position may be sold at a loss in one account while an automatic contribution buys an equivalent or substantially identical investment elsewhere. The portfolio remains tidy on the dashboard. The tax treatment may be less tidy behind the scenes.
There are also transaction-related realities. Many large ETFs trade cheaply, and robo-advisors can automate the process efficiently. But an automated trade is not costless in every sense. Bid-ask spreads, market movement between orders, fund differences, and the possibility of temporary exposure changes all matter. The relevant question is not whether each trade has an explicit commission. It is whether the full strategy produces enough net value after its practical frictions.
Tax-loss harvesting also does not eliminate taxes. It generally lowers the cost basis of the replacement investment. If that replacement asset is sold later at a higher gain, the deferred tax may come due then. The strategy can still be valuable: deferral may have real economic worth, especially when the investor can keep the money invested for years. But deferral is not the same as permanent exemption.
That distinction is easy to lose in product marketing because the immediate event is concrete—a loss appears on a tax report—while the future liability is somewhere beyond the current screen. Behavioral finance has a name for our tendency to value a visible present benefit more heavily than a distant cost. Software interfaces are very good at presenting the visible part.
Beyond the fee: how automated platforms compare
The standard robo-advisor model combines portfolio construction, automatic rebalancing, tax management, and a digital planning layer. Betterment and Wealthfront commonly charge a 0.25% annual management fee. Once underlying fund expenses are included, total recurring costs often fall in the 0.30%–0.45% range.
Schwab Intelligent Portfolios takes a different approach on the advisory fee. It provides automated tax-loss harvesting without charging the same standard management fee, but its automated tax-loss harvesting service requires a minimum balance of $50,000.
That difference matters because account size changes the practical value of automation. A 0.25% fee on a small taxable balance may be modest in dollar terms, but the potential tax benefit may also be modest. A larger account can generate more meaningful tax losses and capital gains, yet it may also have more complicated holdings and more cross-account activity.
The platform decision is therefore not simply about finding the cheapest advertised tax feature. It is about identifying which part of the service you are actually buying.
| Investor situation | What automated TLH may offer | Main limitation |
|---|---|---|
| A new taxable portfolio with regular deposits | More frequent opportunities to harvest losses across fresh tax lots | Benefits depend on market volatility and the investor’s tax rate |
| A mature portfolio with large unrealized gains | Occasional tax deferral during market declines | Fewer harvestable losses; the benefit may fall to 0.1% annually or less |
| An investor with realized capital gains elsewhere | Immediate usefulness because harvested losses can offset gains | The robo platform may not know about all outside transactions |
| An investor who mostly uses IRAs or 401(k)s | Little or no value from TLH in those tax-advantaged accounts | The strategy applies to taxable investment accounts |
| A hands-off investor who dislikes portfolio administration | Lower friction and automatic rebalancing | The convenience fee continues in years when TLH produces little |
| A high-balance investor comparing platforms | Potentially material tax-management value | Minimum balances, fund choices, and household-level tax coordination become more important |
The U.S. tax rules add another layer. Harvested losses can offset unlimited capital gains and up to $3,000 of ordinary income per year in taxable investment accounts. That makes the strategy especially relevant for someone with realized gains during the same tax year. But an investor with no gains and limited taxable income may not receive the same immediate value.
The $3,000 limit is often misunderstood. It does not mean every harvested loss beyond that amount disappears. Unused losses can generally carry forward under the tax rules, but the immediate offset against ordinary income is limited. The timing and size of the benefit matter.
This is why automated tax-loss harvesting is more compelling for some taxpayers than for others. If you have a substantial taxable portfolio, regular contributions, a meaningful tax bill, and a long investment horizon, a 0.25% advisory fee may be reasonable in exchange for coordinated automation. If your taxable account is small, your portfolio is already highly appreciated, or you rarely realize gains, the feature may be doing much less work than the brochure implies.
Where tax-loss harvesting fails to deliver
The strategy tends to disappoint when investors treat it as a universal enhancement rather than a tax-management tool.
The first failure is conceptual: applying taxable-account logic to tax-advantaged accounts. An IRA or 401(k) does not generally create the same tax-loss harvesting opportunity because losses inside those accounts are not used in the same way on a tax return. If most of your assets sit in retirement accounts, paying for a robo-advisor primarily because of TLH makes little sense.
The second failure is timing. A portfolio may go through years in which there are few losses worth capturing. Markets do not distribute tax opportunities according to the annual billing cycle of a robo-advisor. The 0.25% fee arrives every year. Harvesting opportunities do not.
The third is the investor’s tendency to confuse tax savings with investment returns. Selling at a loss and buying a correlated replacement does not make the underlying market risk disappear. It changes the tax position of the portfolio. The replacement investment can still fall, and the original exposure may not be replicated perfectly.
The fourth is overharvesting. More trades are not automatically better trades. If the system pursues every small paper loss without considering fund differences, future basis, household-level activity, and the investor’s broader plan, the tax dashboard may look busy without creating proportional value.
The fifth is account fragmentation. Automated TLH works most cleanly when the platform can see the relevant taxable holdings and transactions. The modern investor, however, often has a retirement account at one provider, an old brokerage account at another, an employer plan, a cash-management app, and perhaps a spouse’s portfolio. The more scattered the financial life, the more difficult it becomes for any single algorithm to maintain a complete picture.
Finally, the strategy may be less useful for investors who prioritize simplicity at the point of sale. If you plan to liquidate the portfolio soon, relocate internationally, donate assets, or leave appreciated holdings untouched for estate planning purposes, the value of current tax deferral may be different from the value assumed by a generic projection.
None of this means manual tax-loss harvesting is automatically superior. Manual harvesting requires discipline, accurate records, awareness of wash-sale rules, and the ability to maintain an appropriate portfolio while replacing the sold investment. Software can reduce operational friction and make a sensible strategy easier to follow. It just cannot turn personal tax circumstances into a standardized product outcome.
So, is automated tax-loss harvesting worth the robo-advisor fee?
For many investors, automated tax-loss harvesting is worth considering when it is part of a broader service they would already use: portfolio allocation, rebalancing, tax-aware withdrawals, and ongoing administration. In that case, TLH can be a useful additional feature rather than the sole justification for the fee.
The argument is weaker when tax-loss harvesting is the entire reason for opening the account. A realistic long-term benefit of roughly 0.2%–0.4% may cover a 0.25% advisory fee in some situations, but not consistently and not before considering fund expenses, tax brackets, wash-sale complications, and the eventual tax treatment of replacement assets. A mature portfolio may produce much less.
The right evaluation is personal, but it does not need to be complicated. Ask four questions:
- How much money do you actually hold in taxable accounts?
- Do you regularly realize capital gains that harvested losses could offset?
- Are you still contributing to the portfolio, creating new tax lots?
- Can the platform account for your activity across the rest of your household’s investment accounts?
If the answers point toward a large, active taxable portfolio and a long holding period, automation may earn its place. If they point toward a small account, mostly retirement assets, few realized gains, and several disconnected brokerages, the feature is more likely to be a polished promise than a meaningful source of value.
The deeper lesson is about trust. Financial software earns credibility not by showing the largest possible tax benefit, but by explaining when that benefit is likely to shrink, what it costs to obtain, and what future obligation it creates. Automated tax-loss harvesting can be useful precisely because it is narrow and mechanical. It captures a tax opportunity while keeping the portfolio invested.
It becomes misleading only when the narrow tool is presented as a permanent return engine. The robo fee may buy convenience, coordination, and behavioral discipline. Tax-loss harvesting may add value on top. But neither should be mistaken for a guarantee that the software will pay for itself every year.