The Tax-Alpha Industry: The $1 Trillion Business of Manufacturing Deductions

The Million Dollar Question: A marketing deck from one of the largest quantitative asset managers asked wealth advisers to imagine a client putting $100 million into its most aggressive tax strategy and leaving it there for ten years. Roughly how much in usable tax losses did the deck project that account would throw off?
A) $12 million B) $58 million C) $300 million D) More than $580 million

Read on for the answer.

There is now a corner of the asset-management business whose actual product is not investment returns. It is deductions. This is how that industry works, who it serves, what it charges, and why the U.S. Treasury has started using the word “abusive” about parts of it.

What it is

“Alpha” is the industry’s word for return above a benchmark — the thing active managers have spent fifty years promising and mostly failing to deliver. Tax alpha is a sidestep. Instead of trying to beat the market, you try to beat the tax bill, and you count the money you did not send to the government as though it were performance. For a high-bracket investor, it often is.

The idea itself is ordinary. Tax-loss harvesting is available in any brokerage account: sell a position that has fallen, book the loss, use it to offset a gain elsewhere, buy something similar so you keep your market exposure. Robo-advisers do it automatically. The federal rules are plain — realized losses offset realized gains dollar for dollar, plus up to $3,000 of ordinary income a year, with the remainder carried forward indefinitely, provided you respect the wash-sale rule and do not buy back a substantially identical security within 30 days.

What has changed is the scale and the engineering. Bloomberg’s Great American Tax Dodge investigation reported in March 2026 that more than $1 trillion now sits in strategies devoted to delaying or shrinking payments to the government, and that roughly $150 billion of that sits in the most aggressive version, known as tax-aware long-short investing, spread across AQR Capital Management and a set of competitors. Columbia Business School’s Federico Mainardi, whose research on the subject Bloomberg drew on, studies the same question academically: who actually harvests, and what it does to investor behavior.

The distinction worth holding onto is a ladder with three rungs. Plain loss harvesting in a normal account. Direct indexing, where you own the individual stocks of an index rather than a fund so there is always something down to sell. And tax-aware long-short, where borrowed money and short positions are added specifically to manufacture more losses than the market alone would supply.

Who uses it

Not “the wealthy” as a bloc — the rungs sort by wealth level and, more precisely, by whether someone has a large taxable gain to absorb.

Below roughly $1M in investable assets, the plain version is the whole story, and it is free. Between about $1M and $5M, direct indexing starts to make arithmetic sense, though most providers set their thresholds higher; the commonly cited floor is a taxable account of $500,000 or more combined with a marginal rate above 35%.

The long-short tier is where the industry lives. Published minimums cluster between $1 million and $3 million at the advisor-gated incumbents — AQR Flex, BlackRock’s Aperio, and Quantinno all sit in that band, according to a 2026 provider comparison compiled by Frec, one of the newer direct-to-consumer entrants (worth noting that Frec sells a competing product, so its comparisons are not neutral; the fee figures below are the ones each firm publishes about itself). Newer platforms have pushed an entry tier down to $100,000. AQR Flex typically requires accredited-investor status and is available only through a registered investment adviser or wirehouse.

But the real qualifier is not net worth. It is a gain event. The clients who make this work are the ones sitting on a large realized or about-to-be-realized capital gain: a business sale, a private-equity distribution after a long holding period, an IPO or secondary sale, a stack of vesting RSUs, a real estate exit, a concentrated founder position. Bloomberg noted the strategy is drawing private-equity investors tallying decades of gains, along with venture capitalists, founders and early employees facing this cycle’s wave of stock offerings. Losses are only worth something if you have gains to put them against. Without the gain, you are paying real fees for a deduction you cannot spend.

Why they use it

The reason is a seam in the tax code, and it is easy to see once you look at the two numbers side by side.

The top federal rate on ordinary income — wages, bonuses, most interest, short-term gains — is 37%, reached above roughly $626,350 for single filers and $751,600 for joint filers in the 2025 tax year. The top rate on long-term capital gains is 23.8%, including the net investment income tax. Half a century ago the effective top rate on long-term gains ran close to 40%, and at one point it fell as low as 15%. That roughly 13-point gap between ordinary and capital treatment is the entire commercial opportunity: a strategy that converts one into the other, or that generates losses usable against the higher-taxed bucket, is manufacturing something valuable out of nothing but classification.

The second reason is time. A dollar of tax deferred is a dollar that stays invested and compounds. Cliff Asness, who co-founded AQR in 1998 after building the Global Alpha strategy at Goldman Sachs, has argued this for most of his career. In a January 2021 AQR piece he wrote that he had long been “vexed” that investing for taxable investors did not focus enough on after-tax returns, and that for private investors, tax costs can be on par with or higher than management and advisory fees. He is not wrong about that, which is part of why the argument has been so effective.

The third reason is the endgame, and it is the one that turns deferral into avoidance. Appreciated assets held until death receive a step-up in basis — heirs inherit them valued at the date of death, and the accumulated gain is never taxed as income to anyone. Deferral plus a long enough life is, functionally, forgiveness. This is the same logic that powers borrowing against wealth rather than selling: don’t realize, don’t pay, pass it on.

How it works

The plain version needs no explanation. The engineered version does, and Bloomberg’s Matt Levine has described the mechanics about as clearly as anyone: instead of spending $100 to buy stocks, you borrow $100, buy $200 of stocks, and short $100 of other stocks. Your net market exposure is still $100. But you now own twice as many positions, and you hold offsetting long and short bets where one side tends to lose whenever the other wins. Every one of those losing positions is a harvestable loss. You have not changed how much market risk you carry. You have roughly tripled the surface area on which losses can be generated.

Two tax rules force the design to be quantitative rather than crude. You cannot be long and short the same stock — that is a straddle, and the loss is not deductible. And you cannot sell a loser and immediately rebuy it — that is a wash sale. So the manager needs a factor model that can say two different stocks are economically similar enough to hedge each other, but not substantially identical in the eyes of the code. That is why this industry is staffed by finance PhDs and built on risk models like MSCI Barra. As Levine put it, apply enough mathematics to investing and what you may get out the other end is lower taxes.

Accounts are described by their leverage ratio: 130/30 means 130% long and 30% short, and the tiers run up through 140/40, 175/75, 200/100 and, at a few providers, 250/150. More leverage means more loss generation and more cost and more risk. Published academic work on the structure has found that an average 130/30 portfolio funded with cash generates roughly 2.7 times more capital losses than a long-only portfolio over its first decade.

The most aggressive products go a step further and aim at ordinary income rather than capital gains — losses that can offset a salary, not just a stock sale. AQR markets a strategy called Delphi Plus to clients who want that steady stream. That is the version regulators are looking hardest at.

What it costs

The honest answer is that nobody can tell you precisely, because half the industry does not publish prices.

The providers that do disclose sit in a readable range. Advisory fees run from about 0.50% a year at the lowest leverage tier to roughly 1.00%–1.30% at 200/100 and 250/150. On top of that sits the financing cost of the borrowed money, which the disclosing firms put at roughly 0.17%–0.86% of assets after the tax deduction for the interest, and meaningfully higher before it — the pre-tax financing cost at the most levered tier runs above 1.4%. All-in, a levered account plausibly costs somewhere between 1% and 2.5% a year of total assets.

The advisor-gated incumbents — AQR, Quantinno, Aperio — do not publish most of their pricing. Fees are negotiated through the adviser, and the adviser layers its own charge, commonly 0.50%–1.00%, on top of the manager’s fee and the financing cost. Two clients in the same strategy at different firms can pay materially different amounts and neither will know. Aperio publishes a 0.55% management fee on one specific large-cap strategy and not much else.

Set that against what the strategies claim to deliver. Parametric, now part of Morgan Stanley, estimates direct-indexing tax alpha at roughly 1.0%–1.5% a year for top-bracket investors over a decade. Simulated results for levered long-short structures run far higher — one provider’s published back-tests show hypothetical after-tax excess returns of roughly 3% at 140/40 and 6%–8% at the highest leverage — but those are simulations, not client results, and they are almost always quoted pre-liquidation, meaning before the taxes that come due when the position is finally unwound.

The gap between a 1%–2.5% cost and a 3%–8% simulated benefit is what the sales pitch lives in. Whether it survives contact with a real portfolio is the next section.

Hidden costs and tradeoffs

The most useful skepticism comes from inside the wealth-management world rather than outside it. Elm Wealth modeled a 20-year long-short direct-indexing program and found that fees consumed more than half the tax benefit. In their central case, an investor with a $10 million appreciated holding ended up with an after-tax, net-of-fees compound return of about 2.05% a year — slightly worse than the roughly 2.16% she would have earned by simply selling, paying the capital gains tax, and buying a low-cost index fund. The strategy worked exactly as advertised on the tax side. The fees ate the difference.

Then there is the exit. These structures defer tax; they do not delete it. Unwinding a levered portfolio full of unrealized gains and open short positions consumes a large share of the harvested losses in the process. Ask any provider how much of the accumulated benefit survives deleveraging and the answer is usually a paper rather than a number.

There is leverage risk, which is not theoretical. Borrowing to hold twice the stock amplifies losses as well as gains, and portfolio-margin accounts can be forced to unwind at the worst possible moment.

And there is legal risk, which the industry has begun putting in writing. AQR stopped publishing asset figures for its tax strategies and added language to its website acknowledging that the IRS could someday bar the benefits — or find the strategies illegal retroactively, in which case penalties may apply. There is no indication the IRS is investigating the firm. But a manager writing that sentence about its own product is a signal worth reading.

The distribution end is already reacting. Fidelity paused new long-short SMA accounts in late 2025 and extended the pause indefinitely. In April 2026, Schwab capped new enrollments at 200/100 leverage, imposed minimums of $1 million for Reg T margin accounts and $3 million for portfolio margin, and limited any single advisory firm to holding 30% of its Schwab assets in these strategies. When the firms earning fees on a trade start rationing it, that tells you something about how they read the risk.

What people get wrong

That it erases taxes. It defers them. The deferral only becomes permanent through death and the step-up in basis, or through charitable transfer. Everything in between is a timing benefit — a valuable one, but a timing benefit.

That the losses are free. They are the residue of real positions taking real losses with real borrowed money. The structure is designed so the portfolio’s net value can still grow while the loss pile accumulates, which feels like alchemy and is really just leverage plus diversification plus careful accounting. The risk did not disappear; it was reorganized.

That the advertised tax alpha is what you keep. Almost all of the headline figures are pre-liquidation and pre-adviser-fee, drawn from simulations rather than realized client outcomes. The after-fee, after-exit number is a different and much smaller number, and it is the one nobody leads with.

That it is a loophole in the crude sense. Nothing here contradicts the tax code. It is a very expensive, very sophisticated reading of rules Congress wrote — which is precisely the complaint. Treasury’s Kevin Salinger and Erika Nijenhuis told a Wall Street Tax Association seminar in July 2026 that the department could not ignore a market forming around transactions with results Congress did not appear to intend, calling some of the strategies “potentially abusive.” Salinger’s example was blunt: “We have seen pitch decks where they advertise that if you invest a million dollars, you may get a $300,000 ordinary loss.” Officials stopped short of announcing rules, saying they expected a serious dialogue with the market before positions harden.

That none of it applies to ordinary investors. The same code section that makes Flex work makes the plain version work. Harvest losses in a taxable account and they offset gains dollar for dollar plus $3,000 of ordinary income a year, carried forward indefinitely. Mind the 30-day wash-sale window. Hold appreciated assets to death and the basis resets for heirs regardless of the size of the estate. The difference at the top is not access to a secret rule. It is the ability to pay millions in fees to squeeze more out of the same one.

Bottom line

The answer is D. Bloomberg reported that an AQR presentation to wealth managers laid out this scenario: invest $100 million in the most aggressive Flex strategy, wait ten years while the money triples, and over that period the account may generate more than $580 million of losses usable against taxes on other investments. Nearly six dollars of deductions for every dollar put in. That is possible because leverage and short positions multiply the number of positions capable of losing money without multiplying the investor’s net market exposure — the losses are manufactured on purpose, at scale, as the product itself.

Which is the honest summary of the whole industry. More than $1 trillion is now deployed in strategies whose deliverable is a smaller tax bill rather than a larger portfolio, and the sharpest end of it has grown faster than the rules governing it. Tom Steyer put the objection to Bloomberg about as neatly as it can be put: every one of these incremental tactics is defensible, but at a societal level it is unacceptable. Morris Pearl, formerly a managing director at BlackRock, added that he would not call the people involved evil, but that a major American industry is now this sort of financial engineering. Meanwhile the firms selling it have started warning clients, in writing, that the government might someday decide they were wrong — and the brokerages holding the accounts have quietly stopped taking new ones.

---

Related reading: Taxes: How Wealth Is Structured and Preserved · Borrowing Against Wealth · Equity Compensation: RSUs, ISOs, and the Tech Wealth Engine · Carried Interest: The Most Defended Loophole in American Tax · Taxing the Billionaires: Wealth-Tax Debates From California to Warren