There Are Only Four Things You Can Do With a Capital Gain

In August 2026, Bloomberg promoted a Big Take feature with this line: “The wealthiest 1% have found a way to eliminate their taxes. Here’s how Cliff Asness’s AQR, the world’s biggest hedge fund, helps make it happen.” The card image was blunter: “World’s biggest hedge fund teaches the wealthy how to slash taxes to zero.”¹

Asness replied within hours, and his objection was narrower than the volume suggested:

“You can’t repeatedly say ‘eliminate their taxes’ for a deferral not elimination strategy. Do you say that about an S&P 500 buy and hold strategy?”

@CliffordAsness, August 4, 2026. He also noted that the article itself hedged the claim in small print, and the social post did not.

He is right on the point he is making, and it is the same point a smaller account had made ten months earlier in a thread that is worth more than the argument it settled.

Four places a gain can go

In October 2025, Brent Sullivan of Tax Alpha Insider asked publicly whether tax-aware long/short was a temporary solution or a long-term hold. The reply that landed the argument came from an account with 384 views on the post:

“it’s a tax deferral strategy…anything other than a longer term hold either requires recognition of gains, acceptance of a long only appreciated stock portfolio, or some sort of estate strategy.

the only way to eliminate a gain without tax is by gifting or a permanent loss of capital… you don’t have to pay a manager to lose money.”

chip (@yooo_chip), October 16, 2025. Sullivan’s reply in the same thread is the other half of the answer: “Deferral is quite valuable. And step-up is, of course, possible in the long slice.”

That exchange is the whole subject in two posts. One person names the mechanism, the other names its value, and neither is wrong. Here is the full set of destinations, with the guide that covers each in depth.

Buy and hold belongs in the deferral row, which is exactly why Asness’s S&P 500 question lands. If postponing a gain counts as eliminating taxes, then every index investor in America has been running a tax dodge since 1976.

What deferral is worth, measured

The strongest evidence here is not from a manager. Ivković, Poterba and Weisbenner studied individual trades at a discount broker and compared the same investors’ taxable and tax-deferred accounts, which separates the tax effect from the general reluctance to sell.² They found real lock-in: a stock that does not appreciate has a median holding period of 42 months in a taxable account, rising to 58 months at 1.5% monthly appreciation, with no such pattern in tax-deferred accounts.

Their Table 9 is the number that matters. It reports the effective accrual capital gains tax rate against a 28% statutory long-term rate, gain-weighted:

Two decades of deferral cuts the rate by more than half, which is why “it is only deferral” understates the case. Getting from 13% to 0.6% is a different mechanism entirely. That last step is § 1014, and it requires dying with the position.

The algebra behind the first half is unremarkable. Postponing a tax payment T for n years while earning k on the money you kept is worth T[1 − (1+k)⁻ⁿ]. Our guide to what tax alpha actually means prices a single harvested loss across six different exits and finds a spread of roughly fifty times, so we will not repeat that here.

The exclusions that are real

Not every tax benefit is a postponement. These are genuine statutory exclusions under current law, and each carries a condition that gets left out of the pitch.

QSBS deserves its own paragraph because the 2025 tax act rewrote it and the rewrite has a trap. Stock acquired after July 4, 2025 now gets a tiered exclusion: 50% at three years, 75% at four, 100% at five or more. The gross-asset ceiling rose from $50M to $75M, the per-issuer cap from $10M to $15M, and the alternative minimum tax preference was repealed retroactively. Stock acquired on or before that date keeps the old regime on every dimension.³

The trap: the act did not amend § 1(h), which still routes § 1202 gain into 28-percent-rate gain. So at the three-year and four-year tiers, the portion that is not excluded is taxed at up to 28% plus the 3.8% net investment income tax, not at 20%. Selling at year three is worse than “half tax-free” makes it sound, and the tiers read as cliffs rather than proration.

There is also one case where the permanence runs the wrong way. Sell a security at a loss in a taxable account and buy the replacement inside your IRA, and Revenue Ruling 2008-5 disallows the loss under § 1091 without the § 1091(d) basis increase that normally preserves it.⁴ In every other wash sale the loss is deferred. In that one it is gone.

Two ways to move a portfolio without selling it

Both of these get described as tax-free diversification. Both are deferral with carryover basis, and they solve opposite problems, which is the part most coverage blurs.

Section 351 conversions into a new ETF

You contribute securities to a newly seeded ETF and receive shares without recognizing gain. It works, and Bloomberg’s analysis of SEC filings counted 105 ETFs created this way holding $22.1 billion at launch and deferring at least $6.5 billion of embedded gains, more than half of them listed in 2025.⁵ Cambria and ETF Architect opened the category with the Cambria Tax Aware ETF (TAX) in December 2024, seeded with $27 million.⁶

The requirement that surprises people: your portfolio has to be already diversified before you start. Treasury Regulation § 1.351-1(c)(6)(i) borrows the test in § 368(a)(2)(F)(ii), so no single issuer may exceed 25% of value and the top five may not exceed 50%. Transferors also need 80% control of the new fund afterward under § 368(c).⁷

Which means a § 351 conversion cannot fix a concentrated founder position, the exact problem most readers assume it was invented for. Basis is substituted under § 358(a)(1) and carries over to the fund under § 362(a), so the gain is preserved on both sides of the transaction. What the ETF can do afterward is reshape the portfolio using § 852(b)(6), the in-kind redemption exemption that makes ETFs tax-efficient generally. The “you need at least 11 holdings” rule you will see in advisor decks is a practitioner shorthand for the five-issuer arithmetic, not something any statute says.

Section 721 exchange funds

These take the concentrated position that § 351 will not touch. You contribute stock to a partnership and, after roughly seven years, withdraw a diversified slice of the pool without recognizing gain along the way.⁸ The price is structural: the fund must hold about 20% illiquid assets, usually leveraged real estate, to stay under the investment-company threshold in § 721(b), and the lockup exists because § 704(c)(1)(B) triggers gain if contributed property is distributed to a different partner within seven years. Your basis follows you into whatever you withdraw.

Two vehicles, opposite entry requirements, identical tax character. The gain is still there in both.

Opportunity Zones split the two ideas on a date certain

Qualified Opportunity Zones are the cleanest illustration that deferral and exclusion are separate things, because in December they visibly separate. A QOZ investment does both: it defers the original gain you rolled in, and it can exclude the fund’s own appreciation after ten years.

Under the original program, deferred gain must be included in income in the tax year containing December 31, 2026, whether or not anything has been sold. IRS Notice 2026-40 confirms that this “deemed included gain” cannot be rolled into the new program: “no amount of deemed included gain can be eligible gain with respect to which an election under either prior or current § 1400Z-2(a) may be made.”⁹

The exclusion half survives untouched. The same notice confirms the investor “remains potentially eligible to make an election under § 1400Z-2(c) on the later sale,” so ten-year appreciation in the fund still escapes tax. A genuinely new gain invested on or after January 1, 2027 gets the rebuilt program, where inclusion moves from a fixed date to five years after the investment.

One investor, one vehicle, and the two features come apart on a calendar date. The deferral comes due. The exclusion does not.

What 2026 tested

The Bloomberg feature was the loudest moment in a year of pressure on tax-aware strategies. Reported straight, the participants agree on the mechanism more than the headlines suggest.

AQR concedes the mechanism in its own research. Krasner and Sosner found that the large net losses these strategies produce arise “not from an increased realization of capital losses but rather from the deferral of capital gains, especially short-term gains on long positions.”¹⁰ A later AQR paper models the liquidation tax directly: $100M over 30 years at 150/50 leverage, 0.7% annual excess return, a 0.45% fee, and death in year 25, producing $1,294M after liquidation against $808M for direct indexing and $780M for an index fund.¹¹ That is a coherent argument. It is also not the same claim as “zero.”

The best counterweight comes from BlackRock researchers. Goldberg, Cai and Schneider found a 130/30 portfolio generates 2.7 times the capital losses of long-only and 4.41% annual pre-liquidation tax alpha, then reported the two haircuts that matter: liquidation cuts the benefit roughly in half, and lacking outside short-term gains to offset cuts it by about 40%. Together, they concluded, those investor characteristics “may render a long-only loss harvesting portfolio or a low-cost ETF more desirable.”¹² Elm Wealth reached a similar place from a different direction, which we cover in Do You Actually Need Direct Indexing?

A short seller pushed harder. Nate Koppikar of Orso Partners, who is short Affiliated Managers Group and profits if the thesis lands, argued the category is a regulatory accident waiting to unwind. His two hardest numbers come from AQR’s own marketing material as reviewed independently by Institutional Investor: gross exposure of up to 600% on one separately managed account strategy, and a projection of 723% cumulative net capital losses over ten years.¹³ His comparison to Renaissance Technologies’ roughly $7 billion basket-options settlement in 2021 is his analogy, not a regulator’s.

Three developments are better documented than the short thesis and did not originate with anyone holding a position. Custodians pulled back: Fidelity restricted new long-short SMA openings from late 2025 and extended the pause indefinitely, later raising fees on some accounts, while Schwab capped adviser long-short allocations at 30% of custodied assets in April 2026. Fidelity has declined to explain publicly, so the risk-management rationale reported around it comes from advisers and analysts rather than the custodian.

Treasury spoke, narrowly. At a Wall Street Tax Association meeting on July 21, 2026, IRS and Treasury officials described certain transactions as “too good to be true” and named § 351 in-kind contributions to newly seeded ETFs, tiered ETF structures, ETFs holding crypto, box spreads distributed in kind, and swaps producing ordinary losses. The same officials affirmed “the viability of section 351 contributions to ETFs in which the contributed assets fit the investment profile of the ETF,” and “an absence of concern about many traditional tax-planning strategies.”¹⁴ Tax-aware long/short equity accounts were not on the list.

As of this writing, no formal guidance, notice, or transaction-of-interest designation has been issued on any of it. For scale, the only formal designations in this neighborhood are Notices 2015-73 and 2015-74 on basket options and contracts, and they are eleven years old. What that means for a reader is narrow: scrutiny is not proof that a strategy is invalid, and current legality is not a promise about future treatment.

Price your own embedded gain

Enter your taxable portfolio and how you expect the gain to leave it. The waterfall separates the tax you will never pay from the tax you will pay later, because those two numbers get added together in most marketing and they are not the same thing.

What we recommend

For nearly every self-directed investor, the durable version of this is unglamorous and does most of the work available:

• Fill tax-advantaged space first. A 401(k), IRA, and HSA solve the location problem without any structuring.
• Hold broad, low-turnover funds in taxable accounts, which keeps realization on your schedule rather than a manager’s.
• Use specific lot identification when you sell. It often matters more than harvesting does.
• Harvest losses when it is nearly free and does not change what you own.
• If you were going to give anyway, give appreciated shares rather than cash. That converts deferral into exclusion, which is the one clean upgrade available to most people.
• Realize gains deliberately in low-income years. Selling inside the 0% bracket raises your basis at no tax cost, which quietly shrinks every future bill.

For anything more elaborate, one question does most of the screening: would you want this strategy, manager, leverage, and fee if the tax benefit were zero? A no is a warning. Tax treatment can improve an investment you already wanted. It should not be the reason you want it.

Then ask for the reporting that answers the actual question: after-tax return assuming full liquidation, the current embedded gain by character, results without a basis adjustment at death, and results if the manager’s alpha is zero. Gross harvested losses answer none of these.

Who this changes the answer for

Advanced deferral structures can be rational for someone with large recurring taxable gains from a business sale or concentrated stock, a multi-decade horizon, high federal and state rates, liquidity elsewhere, genuine comfort with leverage and shorting, a charitable or multigenerational plan, and independent tax counsel who has read the actual documents.

They make considerably less sense for someone with no gains to offset, most of their wealth in retirement accounts, income low enough to sit in the 0% capital gains bracket, a need for liquidity in the next few years, a portfolio that is already three index funds, or an expectation of abandoning the strategy if it underperforms for two years. Unwinding is where the embedded gain shows up, and unwinding early is the most expensive version of this.

One more group deserves a warning: anyone holding a dangerous concentration purely because their heirs will get a step-up. That plan requires you to survive the position’s risk, never need the money, and have the rule survive too. Our guide on what company stock has to beat prices that tradeoff directly.

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 reports unrealized gains and losses per lot, separates short-term from long-term, 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 and it places no trades.

The part that matters for this guide is that it reports the current-year tax reduction and the deferred tax liability as two separate numbers on the same screen. Those are the two quantities this entire argument is about, and most tools show you only the first one.

Frequently asked questions

Does tax-loss harvesting eliminate taxes or defer them?

Defer them, by default. Harvesting a loss cuts your tax bill today and cuts the replacement position’s cost basis by the same amount, so the bill returns when you sell. It becomes elimination only if the replacement is donated or held to a basis adjustment at death. The deferral is still worth having, especially over long horizons and when a short-term gain is being offset at ordinary rates.

Can a 351 exchange diversify my concentrated stock tax-free?

No. A § 351 conversion requires the contributed portfolio to already pass a diversification test, with no single issuer above 25% of value and the top five below 50%. A concentrated position fails on entry. The vehicle built for that problem is a § 721 exchange fund, which accepts concentration but locks your capital up for about seven years and holds leveraged real estate to qualify. Neither one steps up your basis.

Will step-up in basis still exist when I die?

Nobody can tell you that. § 1014 is current law and the 2025 act left it alone while setting the federal estate exclusion at $15,000,000 for 2026 deaths, with inflation indexing starting in 2027. Proposals to replace it with carryover basis or to tax gains at death have been introduced repeatedly and have not passed. Treat it as a real benefit you should plan around and a poor reason to hold a position you would otherwise sell.

What happens to my Opportunity Zone gain on December 31, 2026?

If you still hold a qualifying investment from the original program, the remaining deferred gain is included in income in the tax year containing that date, whether or not you sell anything, with payment due in 2027 depending on your overall position. That amount cannot be re-deferred into the rebuilt program. Your ten-year exclusion on the fund’s own appreciation is unaffected. This is worth raising with your tax preparer before year end rather than in April.

Is deferral still worth it if tax rates go up?

Often, though less so. Deferral pays you the return on money you have not handed over yet, and that benefit is independent of the rate. What a rate increase does is raise the eventual bill, which can swamp the timing benefit over short horizons and rarely does over long ones. The asymmetric case is the reverse: deducting a loss at a high ordinary rate today and repaying at a capital gains rate later is where the arithmetic is genuinely favorable.

Do I need a tax-aware long/short manager?

Almost certainly not, unless you have large recurring realized gains from somewhere else, since the strategy produces losses that need something to offset. Without that supply, published estimates of its benefit fall by roughly 40%, and full liquidation removes roughly half of what remains. Run the pre-tax test first: if you would not buy the leverage, the factor exposure, and the fee on their own merits, the tax treatment is not a reason to start.

Key takeaways

• A gain has four destinations. You pay it, give it away, meet a statutory exclusion, or postpone it. Most products sold as elimination are postponement.
• Deferring twenty years takes the effective rate from about 28% to about 13%. Reaching 0.6% takes a basis adjustment at death, which is a different mechanism with a different requirement.
• A § 351 ETF conversion cannot diversify a concentrated position: the 25/50 test has to pass before you contribute. Exchange funds handle concentration instead, at the cost of a seven-year lockup.
• QSBS now excludes 50% at three years and 100% at five, but the non-excluded portion still runs through the 28% rate bucket plus the 3.8% net investment income tax.
• Original-program Opportunity Zone gains are recognized on December 31, 2026 with no re-deferral available, while the ten-year exclusion on fund appreciation survives.
• Judge any tax strategy on after-tax wealth after full liquidation and fees. If you would not own it with the tax benefit set to zero, the tax benefit is not the reason to own it.

Related guides

• What “Tax Alpha” Actually Means one harvested loss priced across every exit, and why 1% a year is the wrong number.
• Tax-Aware Long-Short the leveraged version, the constructive-sale limit, and a hurdle calculator.
• Do You Actually Need Direct Indexing? when a paid harvesting service clears its fee, and the decay problem.
• How to Sell Tax-Efficiently: SpecID and Tax Lots the destination in the first row, done well.
• Donate Appreciated Stock or Use a DAF? the exit that turns deferral into exclusion, and the 2026 rule changes.
• Your Company Stock Has to Beat the Market by 3 Points a Year what deferral is worth against what concentration costs.
• Tax-Aware Decumulation how to pay the deferred bill at the lowest rate you can arrange.

Sources

1. Bloomberg, “A Tax Strategy for the Rich Built the World’s Largest Hedge Fund” by Loukia Gyftopoulou, Katherine Burton, Sridhar Natarajan and Justina Lee (Finance, The Big Take, August 3, 2026). Deck: “Cliff Asness’s AQR supercharged tax-loss harvesting, erasing IRS bills for the wealthiest.” The “eliminate their taxes” and “slash taxes to zero” language quoted above is from Bloomberg’s own social promotion of the piece, which is what Asness was responding to.
2. Ivković, Z., Poterba, J., & Weisbenner, S. (2005). Tax-Motivated Trading by Individual Investors. American Economic Review 95(5), 1605–1630. Effective accrual rates from Table 9, computed against assumed 40% short-term and 28% long-term statutory rates.
3. 26 U.S.C. §§ 1014, 1015, 121, 1202, 1211(b), 408A(d), 223(f), 351, 358, 362, 368(c), 721, 704(c)(1)(B), 852(b)(6), current through laws in effect August 4, 2026. QSBS amendments by Pub. L. 119-21 § 70431. 2026 dollar amounts from Rev. Proc. 2025-32.
4. Rev. Rul. 2008-5, 2008-3 I.R.B. 271: the loss is disallowed under § 1091 and basis in the IRA is not increased under § 1091(d).
5. Bloomberg, “The Latest Tax Dodge for the Ultra-Rich Is a Customized ETF”, analysis of SEC filings.
6. Cambria Investment Management and ETF Architect (December 18, 2024). Cambria and ETF Architect Launch Cambria Tax Aware ETF (TAX) with $27 Million, press release via Nasdaq.
7. Henry-Moreland, B., & Sullivan, B. (March 12, 2025). Using Section 351 Exchanges To Tax-Efficiently Reallocate Portfolios, Kitces.com.
8. Henry-Moreland, B. (April 29, 2026). When To Use Exchange Funds To Diversify Concentrated Holdings, Kitces.com.
9. Notice 2026-40, Transitional Guidance on Qualified Opportunity Zones under §§ 1400Z-1 and 1400Z-2, sections 4.01 and 4.02.
10. Krasner, S., & Sosner, N. (2024). Loss Harvesting or Gain Deferral? A Surprising Source of Tax Benefits of Tax-Aware Long-Short Strategies, The Journal of Wealth Management, Summer 2024.
11. AQR Capital Management (July 31, 2025). The Impact of Liquidation Taxes on the Lifecycle Benefits of Tax-Aware Long-Short Strategies. Published without an individual byline.
12. Goldberg, L. R., Cai, T., & Schneider, B. (2024). A guide to 130/30 loss harvesting. Journal of Asset Management 25, 445–459.
13. Celarier, M. (June 3, 2026). Will the Booming ‘Tax Aware’ Hedge Fund Business End in Tears? Institutional Investor. Leverage and loss-projection figures are from AQR marketing materials reviewed by the publication; the characterizations are Koppikar’s, who is short AMG.
14. K&L Gates (July 22, 2026). IRS and Treasury Discuss Current Issues With ETFs and Tax Aware Strategies. Officials were not named in the account. See also Notices 2015-73 and 2015-74 for what a formal designation looks like.

Editor’s note

Educational content, not tax advice. Statutory citations were verified against the U.S. Code as current through August 4, 2026, and 2026 dollar amounts against Rev. Proc. 2025-32; tax law changes and several of the provisions described here have pending legislative proposals. The calculator applies one rate throughout and prices only the gain that exists today, so bracket effects, short-term character, state moves, and future appreciation are outside it. Claims attributed to a short seller are identified as such, and no formal IRS or Treasury guidance had been issued on Section 351 conversions or tax-aware long-short strategies as of publication. Talk to a tax professional before acting.