Tax-Loss Harvesting: When Paying the Tax Is the Better Trade

Why a strategy that looks free on paper sometimes costs more than the tax it was supposed to save

May 20, 2026

Tax-loss harvesting is one of the most reflexively recommended planning moves in the wealth-management world. The pitch is clean. A position in your taxable account is down. You sell it. You buy something similar but not "substantially identical." The realized loss offsets your other gains for the year, and any unused balance carries forward indefinitely against future gains and against up to $3,000 a year of ordinary income¹. The market exposure stays roughly the same. You pocket a tax benefit at no apparent cost. It is the rare item on the planning menu that looks like a free lunch.

I am not against tax-loss harvesting. Used correctly, it is a legitimate tool. But the pitch obscures something important, and I think that obscured piece is exactly what causes the strategy to be over-used in practice. Tax-loss harvesting does not erase a tax bill. It moves it. The amount you save today shows up as a larger gain on the replacement position later, because the replacement position inherits a lower cost basis². The strategy only creates real economic value in three specific situations: when the deferral period is long, when the rate you eventually pay is lower than the rate you save today, or when the recognized gain at the back end is eliminated by the basis step-up at death under §1014³. If none of those three is true, the harvest is mostly cosmetic --- it changes the line items on your 1099 without changing your lifetime tax bill, and along the way it adds friction, complexity, and several wash-sale traps that most clients are not actively monitoring.

This memo walks through what I watch for on the harvesting side of the planning conversation. The goal is not to argue against the strategy. The goal is to lay out where it actually creates value, where it quietly destroys value, and the situations in which I would tell a client to pay the tax now rather than chase a deduction. As always, the specifics belong in a conversation with your CPA, your estate attorney, and Perissos --- this is the framework, not the prescription.

A capital loss harvested in a taxable account is governed by two interlocking sections of the Code. Section 1211(b) limits an individual taxpayer to using net capital losses against capital gains without limit, then against up to $3,000 of ordinary income per year ($1,500 if married filing separately), with the unused balance carrying forward indefinitely¹. That $3,000 figure has not been indexed for inflation since Congress raised it from $1,000 in 1976 --- almost half a century of bracket creep has happened around a deduction limit that has not moved a dollar¹. Section 1091 then layers on the wash-sale rule: if a "substantially identical" security is purchased within 30 days before or 30 days after the loss sale, the loss is disallowed in that year and instead added to the basis of the replacement security². The 30-day windows on each side create a 61-day total period in which any replacement purchase carries the risk of triggering the rule².

In ordinary cases, a wash sale is not catastrophic. The disallowed loss is preserved in the basis of the replacement security and is realized later when that security is eventually sold². The cost is timing --- not deduction. But there is one case where the wash-sale rule is genuinely punishing, and it is the case most clients do not see coming. Under Revenue Ruling 2008-5, if the substantially-identical replacement is purchased inside the taxpayer's IRA or Roth IRA --- whether by the taxpayer's direct trade, an automated rebalance, or a target-date fund inside the IRA buying the underlying index --- the loss is disallowed at the taxable account and §1091(d) does not allow the disallowed loss to be added to the IRA's basis⁴. The deduction is not deferred. It is permanently lost. The reason is structural: an IRA has no taxable basis to step up, so the carry-forward mechanism that protects an ordinary wash sale simply has nowhere to land⁴. The IRS has not formally extended that ruling to employer plans, but the logic of treating tax-exempt retirement accounts as an extension of the same taxpayer is the same, and conservative practitioners apply the same caution to 401(k)s and similar plans.

That single ruling is worth pausing on. A client who sells a broad-market index ETF in their brokerage account for a $50,000 loss, and whose IRA target-date fund happens to be purchasing more of the same underlying index on its scheduled rebalance two days later, has just permanently lost a $50,000 deduction⁴. They did not see the trade. They did not authorize it. They will not see it on a tax form. The loss simply does not appear in any future year. This is the single largest hidden cost in DIY tax-loss harvesting, and it is the first thing I look at when a client asks me whether the harvest they made last week was clean.

The same logic applies to a spouse's account². The wash-sale rule treats the household as one taxpayer, so if the harvesting spouse sells at a loss and the non-harvesting spouse's account buys substantially identical shares inside the 61-day window, the loss is disallowed regardless of which accounts are formally separate². For households where each spouse has their own brokerage and their own advisor, this happens more often than people realize --- and the harvesting spouse rarely thinks to call the non-harvesting spouse's advisor before the trade.

Set the wash-sale traps aside for a moment and assume a clean harvest. The trade goes through. The loss is realized. The deduction lands on this year's return. What actually happened?

Three things, in order. First, you sold a position at a price below your purchase price, locking in a loss. Second, you bought a replacement position --- a similar fund, a similar ETF, a similar basket of single stocks --- to keep your market exposure. Third, the replacement position now sits in your account with a cost basis equal to its current price, which is lower than the basis the original position had². That last point is the entire economic content of the trade. You did not destroy the gain. You moved the gain from being unrealized in one position to being unrealized in another position with a lower starting basis. The future tax bill, when the replacement is eventually sold, will be larger by exactly the amount of loss you just harvested.

If you run this once, the deferral is real and meaningful --- you have moved a tax bill from this year into some unknown future year, and the time value of money on the savings is yours. If you run it five or ten or twenty times over a market cycle, something more subtle happens. The basis of the broad portfolio drifts steadily downward. Each harvest you do is satisfying in the moment because the tax savings show up on this year's return. But the embedded gain in the portfolio --- the gap between fair-market value and basis --- grows in lockstep. The portfolio becomes harder and harder to sell down without recognizing meaningful tax, because by harvest fifteen the basis is a small fraction of the market value. Figure 1 illustrates the arithmetic.

This is the "basis treadmill" effect. It is not theoretical. It is the natural outcome of a strategy that lowers basis every time it triggers, run inside a portfolio that is rising over the long term. The harvests deliver real cash-flow benefits in the years they happen. They also build a future tax liability that the client now cannot get out of without writing a check. For households whose plan is to spend down the taxable account in retirement --- to fund their lifestyle by selling shares --- this matters a great deal. The "deferral" they earned in their forties shows up as a much larger embedded gain in their sixties, exactly when they need to sell.

The strategy still wins, on paper, if the eventual rate paid at sale is the same as the rate saved at harvest. The time value of money makes the deferral economically positive. But it does not win as much as the year-by-year tax-statement view suggests, and the math gets worse if the eventual sale lands the household into a higher bracket than the bracket they were in when the harvests happened. Several common situations push the eventual rate higher: a business sale year, a year with a large Roth conversion, a year with a forced concentrated-position liquidation, or a surviving-spouse year after the joint filing status converts to single. In each of those cases, the household harvested losses against the 15 percent long-term-capital-gain bracket and is now realizing the deferred gain at 20 percent plus the 3.8 percent Net Investment Income Tax^5,6. That is a 8.8-percentage-point swing in the wrong direction. Multiply by fifteen years of harvested basis and the "free deduction" turns into a real tax cost.

Figure 1: Stylized 20-year illustration of a $1,000,000 taxable account compounding at 8% per year, with 3% of market value harvested as losses each year. Cost basis falls in lockstep with each harvest while market value rises, producing the widening embedded-gain wedge --- the deferred tax liability the strategy builds quietly underneath. Illustrative only --- not historical and not a forecast.

Here is where the analysis gets uncomfortable for the harvest-everything camp. Under §1014, when an asset passes to an heir at the original owner's death, the heir's cost basis is reset to the asset's fair-market value on the date of death (or the alternate valuation date six months later, if the executor elects it)³. The unrealized gain that accumulated during the owner's lifetime is erased. The heir can sell the position the next day and owe zero capital-gains tax on the entire pre-death appreciation³.

What this means for tax-loss harvesting is structural. The "deferred gain" you created by harvesting a loss against a lower-basis replacement position --- the entire amount of the future tax bill the strategy is supposed to be hiding from --- evaporates at death, as long as the replacement position is still held when the owner dies. The household that harvests aggressively at age 50 and dies at 85 without ever selling the replacement positions has captured the deduction at the front end and never paid the offsetting gain at the back end. The strategy works exactly as advertised.

But that is not the common case. Most appreciated taxable portfolios get sold during the owner's lifetime, not held to death. Retirement spending sells them. House purchases sell them. Business reinvestment sells them. Gifts to children sell them. Each of those sales reaches into a portfolio whose basis has been ground down by years of harvesting, and the recognition that the harvesting strategy was supposed to defer arrives in full --- often into the highest bracket the household will ever be in.

The household that does not harvest aggressively, and instead lets gains compound undisturbed, ends up with a different shape of taxable account. Some lots are deeply appreciated. Some lots, by accident of when they were bought and what the market did, sit close to or above their original basis. Some lots may even sit at a small loss. The household has flexibility at sale time --- it can choose which specific lots to sell to manage the realized gain that year. And the portion of the portfolio that is never sold during life crosses death with a full step-up under §1014, eliminating the embedded gain entirely³. The trade-off is real: aggressive harvesting captures small certain benefits each year and surrenders a larger contingent benefit at the back end. Conservative harvesting surrenders the small certain benefits and preserves the larger contingent one. Which choice wins depends on the household's actual horizon, spending plan, and probability of selling the whole portfolio during life.

Figure 2 lays out the comparison in a stylized way: a household that harvests aggressively for thirty years versus one that does not, both holding the same underlying market exposure, looked at from a lifetime tax perspective rather than a single-year one.

Figure 2: Stylized comparison of $300,000 of harvested losses over a 20-year horizon. Path A: losses absorb gains at the 15% LTCG bracket, lowering basis on the replacement positions; those positions are later liquidated during life and the deferred gain is recognized at 20% + 3.8% NIIT (IRC Sec. 1411). Path B: no harvest; portfolio held into death; embedded gain erased by basis step-up under IRC Sec. 1014. Federal only. Illustrative.

There is a set of situations in which paying the tax now is simply the more efficient choice, and these are the situations I want clients to recognize when they hear themselves saying "let's harvest some losses."

The clearest case is the household sitting inside the 0 percent long-term-capital-gain bracket. For 2026, that bracket runs to $98,900 of taxable income for joint filers and $49,450 for single filers⁵. A retired couple living mostly off cash, Roth distributions, and modest pension or Social Security income often has taxable income well below the joint threshold. For that household, recognizing long-term capital gains is essentially free at the federal level until the threshold is hit. Harvesting losses in that environment is actively bad planning --- the loss is being used to offset gains that would have been taxed at zero anyway, and the only deduction the household captures is the $3,000 against ordinary income¹. The remainder of the harvested loss carries forward into future years where it competes against gains that may also be in the 0 percent bracket. The household has given up a permanent basis reset on the harvested position in exchange for a deduction that does not actually save federal tax. Worse, the lower basis on the replacement position now creates a future tax liability that the household would not otherwise have had.

A second case is the household with a planned charitable lump-sum --- a donor-advised fund contribution, a direct gift of stock, or a charitable remainder trust funding event⁷. When appreciated long-term stock is given directly to a qualified public charity under §170, the donor takes a fair-market-value deduction (subject to the 30 percent of AGI limit) and recognizes zero capital gain on the transfer⁷. The most-appreciated lots in the portfolio are by definition the most efficient lots to give. Harvesting losses in the years leading up to a charitable lump-sum reduces the basis of the remaining lots, which means the household ends up giving away exactly the appreciated lots it would have given away anyway --- but the harvest has saved nothing on those gifted lots (a §170 fair-market-value gift recognizes no gain in the first place) and has lowered the basis of the lots the household plans to keep. The net effect is a worse picture, not a better one. For charitable households, the rational sequence is the opposite of harvest-everything: let appreciation build, donate the most-appreciated lots, hold or step-up the rest.

A third case is the household with a multi-year sale plan. If a client is going to liquidate a $2 million embedded-gain position over five years through the 15 percent long-term-capital-gain bracket, the year-by-year recognition is already happening at a relatively low rate⁵. Harvesting offsetting losses inside that same window saves 15 percent of the offset amount today, at the cost of a lower-basis replacement position that will throw off taxable gain later. Unless the "later" year is at a meaningfully lower rate (a single-filer surviving-spouse year is unlikely to qualify --- it is usually higher), the harvest is not creating durable value. It is just shuffling timing within a band of rates the household is already paying.

A fourth case --- and this one is psychological, not statutory --- is the household whose harvesting strategy is causing them to hold a worse portfolio than they would otherwise hold. A common pattern: the household is committed to a position they would rather sell, but they refuse to sell because they "don't want to recognize the gain." Six months later the position is down and they harvest the loss --- not because the loss is meaningful but because the harvest gives them an excuse to act. They buy a "non-substantially identical" replacement that they would not otherwise have chosen. The replacement is held only because the 30-day clock has to run, and after 30 days the household either holds the inferior replacement out of inertia or switches back to the original position. The realized loss is real. So is the tracking error against what the household actually wanted to own. For positions a household genuinely wants to sell, the simpler answer is to sell, pay the tax, and own the portfolio they want. Harvesting is not a substitute for portfolio construction.

Tax-loss harvesting also has costs that do not show up on a tax return. They are easy to ignore in the year a harvest happens. They are not easy to ignore once a household has been running the strategy for a decade.

The first hidden cost is record-keeping complexity. Each harvested lot creates a new basis at a new date, and each replacement position then inherits the modified basis of any wash-sale disallowance that occurred. Over years, a portfolio that began with a handful of clean lots becomes a sprawling lot-by-lot ledger. For households whose tax returns are already complicated --- business owners, multi-state filers, large estates --- adding a year-by-year harvest schedule to the ledger increases the surface area for CPA errors at the exact moment when stakes are highest. The cost of one mis-tracked wash sale on a return can run into five-figure penalties and interest.

The second hidden cost is the friction of finding non-substantially-identical replacements. The IRS has never published a bright-line definition of "substantially identical"², and the case law and ruling history is narrow. Two S&P 500 index funds from different managers are widely treated by practitioners as substantially identical to each other; an S&P 500 fund and a Russell 1000 fund are widely treated as not substantially identical, even though they hold a heavily overlapping basket of names. The line is fuzzy, and the IRS has been clear that they will challenge close calls under §1091 if they choose to². Households that harvest aggressively spend real time and real advisory cost managing this gray zone, and a household that drifts across the line --- intentionally or otherwise --- finds out only when the deduction is denied on audit and the carry-forward they were counting on is gone.

The third hidden cost is the most often missed: the harvest can disrupt the household's actual investment plan. Tax-loss harvesting forces a sale at a market low (that is what creates the loss in the first place). The replacement is bought immediately, but it is bought at the same low price, and now the household owns a different security at the bottom of the move. If the original security and the replacement do not behave identically through the recovery --- which they will not, since by construction they are not substantially identical --- the household has introduced tracking error at the worst possible time. Over a year or two this can wash out. Over a decade of harvests it can produce a portfolio whose return diverges meaningfully from the benchmark the household is supposedly tracking. The deduction was real. So is the divergence.

I want to be clear that I am not arguing against the strategy. There are situations where tax-loss harvesting is the right answer, and I will recommend it in those situations. The point of the memo is that the situations are narrower than the typical pitch suggests.

The cleanest case is the household with a known, dated, large gain coming. A client who knows they are going to sell a business in three years, exercise a large block of nonqualified stock options in five years, or sell a concentrated stock position in seven years has a clean target for a multi-year loss-bank. Harvesting losses in the years leading up to the event creates carry-forwards that absorb the known future gain at a rate the household is unlikely to ever see otherwise. The harvest in this case is not chasing a deduction in a vacuum. It is positioning the household's basis ledger to absorb a known event at the lowest possible rate. This is structurally different from harvest-everything-every-year and produces materially better outcomes.

A second case is the household with very-long-horizon equity positions inside a taxable account that is genuinely a multi-decade compounding bucket. A harvest at year five followed by a step-up at year forty captures the deduction at the front end and erases the deferred gain at the back end under §1014³. The window has to be long enough that the time-value-of-money math works, and the household has to actually hold the replacement position into death rather than spending it down. For young clients with high-income years still ahead, that is sometimes the right picture. For pre-retirees within five to ten years of starting to spend the taxable account, it is rarely the right picture.

A third case is the household running coordinated direct-indexing or tax-managed indexing as part of a deliberate, multi-year plan to build a loss bank. This is the only case in which "harvest aggressively" actually pencils, and it pencils only because the harvests are being run with knowledge of the household's projected future bracket path, charitable plans, and estate plan. The strategy is the same on paper as the do-it-yourself harvest --- §1091 is the same statute regardless of who is doing the trading² --- but the difference is that the harvests are coordinated with a downstream plan to use the carry-forwards at a particular rate at a particular time. Harvests without that downstream coordination are deductions in search of a purpose.

In all three of these cases, the question I am asking is the same: where, specifically, is the deferred tax going to come due, and at what rate? If I cannot answer that question, the strategy is reflex, not planning.

The frame I use with clients is simple. I do not want harvesting to be a default. I want it to be a deliberate decision in a year where the answer to three questions is "yes."

The first question is whether the loss being harvested can be cleanly captured. That means the household and the spouse's household and every retirement account holding any potentially substantially-identical position has been audited for purchase activity inside the 61-day window^2,4. It means the replacement security has been chosen with a non-substantially-identical reading that would survive an IRS challenge. It means the lot-level basis record-keeping is in good order on both sides of the trade.

The second question is whether the deduction is being created against a real bracket. A household in the 0 percent long-term-capital-gain bracket should not be harvesting losses --- the deduction is being created against gains that would have been taxed at zero, and the only durable benefit is the $3,000 against ordinary income¹. A household in the 15 percent bracket should weigh the harvest against the probability that the replacement-position gain will be recognized later in the 20 percent + 3.8 percent NIIT bracket^5,6. A household in the 20 percent + NIIT bracket today has the cleanest case for harvesting, provided the recognized gain at the back end has somewhere to go --- a lower-bracket year, a charitable lump-sum, a planned step-up^3,7.

The third question is whether the household actually has a plan to use the carry-forward. A loss bank is not free --- it has been paid for in lowered basis and in replacement-position risk. The carry-forward is only valuable if it absorbs a future gain that would otherwise be taxed at a meaningful rate. Harvesting in 2026 to absorb a $3,000-per-year ordinary income deduction across the next twenty years is a poor trade¹. Harvesting in 2026 to absorb a known $2 million business-sale gain in 2029 is a structurally sound trade.

When all three answers are yes, I recommend the harvest. When any one of them is no, I usually recommend paying the tax this year and keeping the higher basis in the position. That is not a popular recommendation in an industry that tends to count harvested losses as a free win. It is the right recommendation in more cases than the industry default suggests.

Tax-loss harvesting is real planning when it is real planning. It is not real planning when it is reflex. The pitch --- a free deduction, no economic cost --- collapses under any honest look at the basis-erosion math, the wash-sale traps, the §1014 step-up, and the cases in which the household's eventual rate ends up higher than the rate they harvested against. The cost of the strategy is paid in lowered basis, in record-keeping complexity, in occasional permanent loss of deductions through unintended retirement-account trades⁴, and in tracking error on portfolios that drift further from the household's actual investment intent with each harvest.

For most clients, the right answer is somewhere in between. Harvest when the household has a known, dated, large future gain coming. Harvest when the household sits firmly in the 20 percent plus 3.8 percent NIIT band and will reasonably recognize the deferred gain at a lower rate later or at a step-up^3,5,6. Pay the tax when the household is sitting in the 0 percent bracket, when the lots being harvested would otherwise have been donated, when the household's plan is to step the portfolio up at death, or when the harvesting decision is causing the household to hold a worse portfolio. Sometimes the cleanest planning move is to look at a loss, decide it does not serve the household's long-term picture, and choose not to chase it. That is not a missed opportunity. That is a portfolio left in the right shape for the household's actual five-, ten-, and thirty-year plan.

As with every planning topic, the specifics belong in a conversation with your CPA, your estate attorney, and Perissos. The tax code's flexibility on capital-gains timing is real, and harvesting is one of several tools available to use that flexibility. It is not always the right tool. Knowing when to pay the tax is the part of the conversation that does not show up in the marketing.

All my best,

Brandon VanLandingham, CFA, CMT, CFP

Founder / CIO

¹ IRC §1211(b), Limitation on capital losses (individual taxpayers may deduct net capital losses against ordinary income up to $3,000 per year, or $1,500 if married filing separately, with the unused balance carrying forward indefinitely under §1212(b)). Limit raised from $1,000 to $3,000 by the Tax Reform Act of 1976; not indexed for inflation since. 26 U.S.C. §1211, https://uscode.house.gov/view.xhtml?req=granuleid:USC-prelim-title26-section1211; IRS Publication 550 (2025), Investment Income and Expenses, https://www.irs.gov/publications/p550.

² IRC §1091, Loss from wash sales of stock or securities (loss disallowed where substantially identical stock or securities are acquired within 30 days before or after the sale; §1091(d) adds the disallowed loss to the basis of the replacement property). 26 U.S.C. §1091, https://www.law.cornell.edu/uscode/text/26/1091; IRS Publication 550 (2025), Investment Income and Expenses, https://www.irs.gov/publications/p550.

³ IRC §1014, Basis of property acquired from a decedent (step-up to fair-market value at date of death, or alternate valuation date under §2032 if elected); IRS Publication 551, Basis of Assets, https://www.irs.gov/publications/p551.

⁴ Revenue Ruling 2008-5, 2008-3 I.R.B. 271 (loss on sale of stock or securities by an individual is disallowed under §1091 where the individual's IRA or Roth IRA purchases substantially identical stock or securities within the wash-sale period; §1091(d) does not increase the basis of the IRA or Roth IRA, with the result that the disallowed loss is permanently lost). https://www.irs.gov/pub/irs-drop/rr-08-05.pdf; IRS, Internal Revenue Bulletin 2008-03, https://www.irs.gov/irb/2008-03_IRB.

⁵ Rev. Proc. 2025-32, 2026 inflation-adjusted long-term capital-gains rate breakpoints (0%, 15%, 20% thresholds for joint, single, head of household, MFS, and trusts). IRS, https://www.irs.gov/pub/irs-drop/rp-25-32.pdf.

⁶ IRC §1411, Net Investment Income Tax (3.8 percent of the lesser of net investment income or modified adjusted gross income over $200,000 single / $250,000 joint; thresholds not indexed for inflation). IRS Topic No. 559, Net Investment Income Tax, https://www.irs.gov/taxtopics/tc559.

⁷ IRC §170, Charitable, etc., contributions and gifts (donor of long-term-appreciated property to a qualified public charity may deduct fair market value, subject to a 30 percent of AGI limit with a five-year carryforward; no capital gain is recognized on the transfer). IRS Publication 526, Charitable Contributions, https://www.irs.gov/publications/p526.

This piece is educational. It is not legal, tax, or accounting advice and is not a recommendation to take or refrain from any specific action. Tax law is fact-specific and changes regularly. Please coordinate any decisions discussed here with your attorney, your CPA, and Perissos before acting.

Perissos Private Wealth Management is a Registered Investment Adviser ("RIA"). Registration as an investment adviser does not imply a certain level of skill or training, and the content of this communication has not been approved or verified by the United States Securities and Exchange Commission or by any state securities authority. Perissos Private Wealth Management renders individualized investment advice to persons in a particular state only after complying with the state's regulatory requirements, or pursuant to an applicable state exemption or exclusion. All investments carry risk, and no investment strategy can guarantee a profit or protect from loss of capital. Past performance is not indicative of future results.

The information contained in this newsletter is intended to provide general information about market themes. It is not intended to offer investment advice. Investment advice will only be given after a client engages our services by executing the appropriate investment services agreement. Information regarding investment products and services is given solely to provide education regarding our investment philosophy and our strategies. You should not rely on any information provided in making investment decisions.

Market data, articles and other content in this material are based on generally available information and are believed to be reliable. Perissos Private Wealth Management does not guarantee the accuracy of the information contained in this material.

Perissos Private Wealth Management will provide all prospective clients with a copy of our current Form ADV, Part 2A (Disclosure Brochure), Part 2B (Supplemental Brochures), and Part 3 (Client Relationship Summary) prior to commencing an advisory relationship. You can also view these documents at any time at adviserinfo.sec.gov or by contacting us requesting a copy.

Charitable Remainder Trusts and the ILIT Wealth Replacement Strategy

The benefits of Qualified Charitable Distributions from IRA

Leveraging RMD for Qualified Charitable Distribution: A Tax-Saving Strategy