What "Tax Alpha" Actually Means, and Why 1% a Year Is the Wrong Number

“Tax alpha” is one of the few investing terms that manages to sound rigorous while meaning almost nothing on its own. A provider advertises 1% a year, or 1.8%, and the number lands in your head next to your expected return, where it looks like a permanent upgrade. Seven percent becomes eight.

That is not what it measures. This guide takes the term apart: where the headline figures come from, what they assume, and why the same harvested loss can be worth fifty times more to one investor than to another. The short version of our position, which matches what the research actually supports: tax-loss harvesting is worth doing when it is cheap and easy, and it is not a source of return you can add to a projection.

Four different numbers wearing the same name

Most confusion about tax alpha comes from four quantities that get used interchangeably and differ by an order of magnitude:

• Losses harvested. The dollar amount of losses you realized. A portfolio that harvests $100,000 of losses has not earned $100,000. It has created a tax asset whose worth depends on whether, when, and at what rate you can use it.
• Tax saved this year. Losses multiplied by your rate. This is the number in the app’s year-end summary, and it is almost always the biggest of the four.
• “Tax alpha.” An annualized percentage that different providers compute differently: before or after fees, before or after the eventual tax bill, with or without assumed new contributions, at whatever tax rate the model chose.
• Extra spendable wealth at the end. The only one that pays for anything. It is the smallest, the hardest to compute, and the one nobody advertises.

What a harvested loss actually does

Buy a fund for $100,000. It falls to $80,000. You sell, book a $20,000 loss, and immediately buy something similar but not substantially identical, so you stay invested.

At a 23.8% rate, that loss offsets gains and saves you $4,760 today. Here is the part the pitch skips: your replacement position now has an $80,000 basis rather than $100,000. If it later climbs to $120,000 and you sell, your taxable gain is $40,000 instead of $20,000. The tax you skipped comes back.

This is why the honest description is deferral rather than elimination. Wealthfront says so in its own methodology paper: tax-loss harvesting “can be used to defer tax liabilities, not avoid them. As a result it has very little value if applied over a short investment horizon (i.e. less than two years), but can be extremely valuable if executed over a long period of time.”¹

Deferral is genuinely valuable, and the case against harvesting is as wrong as the case for treating it as free money. The $4,760 stays invested and compounds for you rather than for the Treasury. Think of it as an interest-free loan whose principal comes due when you sell, whose term you control, and which is sometimes forgiven entirely.

The same loss, priced under every exit

Here is that $20,000 loss at a 23.8% rate, with the deferred tax reinvested at 5% after tax. The only thing changing is how the story ends.

From $238 to $12,630 on an identical harvested loss: a spread of roughly fifty times, driven entirely by the exit. The $4,760 of “tax saved” that gets reported to you is not any of these figures.

That spread is the reason a single annualized percentage cannot describe this strategy. The published numbers are honestly computed; they simply depend on facts about you that no headline figure can carry.

Where 1% comes from, and what it assumes

The figure behind most marketing traces to one paper. Chaudhuri, Burnham and Lo tested harvesting on the 500 largest US companies from 1926 to 2018 and reported average tax alpha of 1.08% a year.² Three things about that number are easy to miss and are stated by the authors themselves:

• It is measured before transaction costs.
• It assumes a 35% short-term and 15% long-term rate, so much of the gain is rate arbitrage available only to investors generating short-term gains.
• When the authors impose the wash-sale rule, the same strategy’s alpha falls to 0.82%. That is a quarter of the benefit removed by a rule that applies to everyone.

A year later, the same journal published the rebuttal. Khang, Paradise and Dickson open by naming the convention directly: “In the tax-loss harvesting literature, a typical investor is assumed to have an unlimited supply of offsetting capital gains and can earn annualized tax alpha on the order of 100 bps.” Using investor-level data, they find that investor profiles drive roughly 60% of the variation in harvesting outcomes.³ The benefit is mostly a fact about the investor rather than the strategy.

Vanguard’s later research puts the range at 0.47% to 1.27% and adds a scaling caveat worth reading twice: the figure “must be scaled by the size of taxable equity assets relative to the size of the entire portfolio. That is, if an investor’s TLH alpha is 1%, but the applicable assets (taxable equity) reflect only 25% of the portfolio, the expected impact to the portfolio would be only 0.25%.”⁴ Most people hold a large share of their wealth in retirement accounts, where harvesting does nothing at all.

Run the claim backwards

A useful sanity check is to ask what 1% a year would require. On a $100,000 taxable portfolio, 1% is $1,000 of benefit. At a 23.8% rate, producing $1,000 of tax value needs $4,202 of harvested losses, which is 4.2% of the portfolio, realized as losses, every single year, with gains available to absorb all of it.

In a bad year, or in a young account still near its purchase prices, that is entirely achievable. As a twenty-year average, in a portfolio that mostly goes up, it is not. This is the mechanism the industry calls ossification, and Vanguard describes it plainly: harvesting pushes your cost basis down, which combined with rising markets “makes those assets from the reinvested proceeds harder to harvest in the future, which can manifest in dwindling TLH alpha over the investment horizon.”⁴

The same paper is blunter than most marketing: “It is important to note that TLH is not a free lunch. There are risks for any investor engaging in TLH and it can even subtract value in certain circumstances.” That is the fund company selling the service.

Where the benefit is genuinely large

None of this makes harvesting pointless. It makes the benefit conditional, and the conditions are knowable. Using the same $20,000 loss over twenty years, the value swings on which rate you deduct at and which you repay at:

The best case by a wide margin is offsetting a short-term gain, because you are deducting at ordinary-income rates and repaying at capital-gains rates. That is exactly the situation a buy-and-hold index investor is least likely to be in, and exactly the assumption the headline study makes. The gap between the study and the reader lives right there.

So the investors who get the most from harvesting are the ones with recurring realized gains, high rates, long horizons, ongoing contributions that keep creating fresh cost basis, and a charitable or estate plan that means the deferred tax may never be paid. The investors who get the least hold mostly retirement accounts, sit in the 0% capital-gains bracket, have no gains to offset, or plan to sell soon.

What this means for a paid service

Once you accept that the benefit is conditional, the question about direct indexing or a managed harvesting service becomes arithmetic rather than philosophy. A fee has to be beaten by incrementalbenefit, meaning benefit above what you would have captured harvesting a couple of broad funds yourself for free.

On a $500,000 taxable portfolio, a 0.40% fee costs $2,000 a year. At a 23.8% deduction rate, the service has to find roughly $8,400 of extra usable losses annually, every year, purely to break even against doing it yourself. It might well do that. Ask for that number specifically, because gross harvested losses do not answer the question.

We work through when that math favors a provider in Do You Actually Need Direct Indexing? and the leveraged version in Tax-Aware Long-Short. Worth noting that even the sellers concede the shape of the problem: AQR researchers report that losses from direct-index harvesting “taper off within the first few years,” reaching a maximum near 30% of initial capital, and that full liquidation typically means realizing most of the gains you deferred.⁵

Price your own loss

Enter a loss, your rates, and how you expect the position to end. The chart shows every exit side by side, which is the fastest way to see why one percentage cannot cover them all. The second panel converts a fee into the extra losses a service must find to justify it.

What we recommend

Harvest losses when it is cheap, quick, and does not change what you own in any way that matters. Two or three broad index funds with sensible replacements will capture a meaningful share of the available benefit at close to zero cost and about ten minutes a year. Our tax-loss harvesting guide covers the mechanics, including the wash-sale traps that catch people with automatic dividend reinvestment.

Do not put tax alpha in a spreadsheet as a return assumption. It is not an expected return, it does not compound reliably, it shrinks as your account ages, it applies only to the taxable slice of your wealth, and a good deal of it comes back when you sell. Treat it as maintenance that occasionally pays well, in the same category as rebalancing thoughtfully or choosing the right account for the right asset.

Keep the ordering straight, too. How much you save, what you own, what you pay in fees, and whether you stay invested through bad years each matter more than harvesting ever will. Tax work belongs in the optimization layer, and the optimization layer is worth very little if the foundation underneath it is wrong.

How Summitward helps

The tax-loss harvesting page is a display-only analyzer rather than a trading tool. You enter your tax lots and it computes unrealized gains and losses per lot, separates short-term from long-term, applies a materiality threshold so you are not chasing trivial harvests, estimates both the current-year tax reduction and the deferred liability you are creating, models the netting order through the $3,000 ordinary-income offset and any carryforward, and flags basic wash-sale risk. It does not connect to your brokerage or place trades. Showing the deferred liability alongside the savings is the point: those two numbers belong on the same screen.

Frequently asked questions

Is tax-loss harvesting worth doing at all?

Usually yes, when it costs you almost nothing. The expected benefit is positive for most taxable investors with losses available, and the effort for a simple two-fund swap is small. What it is not is a strategy worth reorganizing a portfolio around, paying a large fee for without checking the arithmetic, or counting on in a retirement projection.

Why do providers quote such different numbers?

Because they are measuring different things and rarely say which. One firm may report first-year tax savings as a percentage of assets, another the long-run after-tax return difference against an unmanaged portfolio, another a figure before its own fee. Ask what the number is net of, what it assumes about your gains and rates, and whether it accounts for the tax due when you eventually sell.

Does harvesting more losses mean more money?

Not by itself. A loss you cannot use this year offsets up to $3,000 of ordinary income and carries forward, which is worth something but much less than offsetting a gain today. Harvested losses are an input. Usable losses, at a high enough rate, with a distant or forgiven repayment, are the output.

Is the tax I defer ever forgiven completely?

Under current law it can be. Donating appreciated shares to charity, or holding them until the basis is adjusted at death, can mean the deferred gain is never taxed to you. Those two paths are where harvesting comes closest to its marketing, which is also why they are worth planning for deliberately rather than assuming. Both have conditions, and the rules move: a new 0.5% floor on charitable deductions took effect for tax years beginning after 2025, which changes the arithmetic of the giving route without changing the capital-gains avoidance itself.⁶ Our guide to donating appreciated stock covers the current rules.

Key takeaways

• Harvesting defers tax rather than erasing it: the loss cuts your basis by the same amount, so the bill usually returns when you sell.
• One $20,000 loss is worth $238 or $12,630 depending only on the exit, a spread of roughly fifty times that no single percentage can capture.
• The 1.08% headline figure is before trading costs, assumes 35% short-term and 15% long-term rates, and falls to 0.82% once the wash-sale rule is enforced.
• Investor circumstances drive about 60% of the variation in outcomes, and Vanguard’s range is 0.47% to 1.27%, scaled down further by how much of your wealth is taxable equity.
• Reaching 1% a year on a $100,000 portfolio requires harvesting $4,202 of usable losses annually, which gets harder every year as basis falls.
• A paid service must beat its fee with losses beyond what free fund-level harvesting already finds, not with gross harvested losses.

Related guides

• Tax-Loss Harvesting: How to Turn Portfolio Losses Into Tax Savings the mechanics, netting rules, and wash-sale traps.
• Do You Actually Need Direct Indexing? when a paid harvesting service clears its fee, and when it does not.
• Tax-Aware Long-Short the leveraged version, and the risks that come with it.
• How to Sell Tax-Efficiently: SpecID and Tax Lots choosing which lots to sell, which often matters more than harvesting.
• Donate Appreciated Stock or Use a DAF? the exit that turns deferral into avoidance.
• Tax-Aware Decumulation how to pay the deferred bill at the lowest rate you can arrange.

Sources

1. Wealthfront. Tax-Loss Harvesting white paper.
2. Chaudhuri, S. E., Burnham, T. C., & Lo, A. W. (2020). An Empirical Evaluation of Tax-Loss-Harvesting Alpha. Financial Analysts Journal 76(3), 99–108. 1.08% before transaction costs; 0.82% with the wash-sale rule imposed; 500 largest US companies, 1926–2018; 35% short-term and 15% long-term rates assumed.
3. Khang, K., Paradise, T., & Dickson, J. (2021). Tax-Loss Harvesting: An Individual Investor’s Perspective. Financial Analysts Journal 77(4), 128–150.
4. Paradise, T., Khang, K., & Dickson, J. M. (2024). Tax-loss harvesting: Why a personalized approach is important, Vanguard research.
5. Liberman, J., Krasner, S., Sosner, N., & Freitas, P. P. M. (2023). Beyond Direct Indexing: Dynamic Direct Long-Short Investing. The Journal of Beta Investment Strategies 14(3), 10–41.
6. 26 U.S.C. § 170(b)(1)(I), added by Pub. L. 119-21 (2025), imposing a 0.5% of contribution base floor on individual charitable deductions for taxable years beginning after December 31, 2025. See also § 1211(b) for the $3,000 ordinary-income offset, unchanged and not indexed for inflation, and § 1014 for basis adjustment at death.

Editor’s note

Educational content, not tax advice. Every figure in the exit tables is computed under stated assumptions (a 23.8% rate unless noted, the deferred tax reinvested at 5% after tax, and the harvested loss fully usable against gains in the year realized) and your situation will differ. Rules on the wash sale, the ordinary-income offset limit, charitable deductions, and basis adjustment at death are current law and subject to change. Talk to a tax professional before acting. Citations verified against publisher metadata and the referenced research papers as of July 2026.