How Losing Money Became Wall Street's Hottest Product

How Losing Money Became Wall Street's Hottest Product

Here is a pitch that is apparently working on billionaires: give the manager $100 million, and over the next ten years the money will roughly triple — and, along the way, the manager will produce $580 million in harvestable losses, nearly six times the amount that went in. Losses you get to use against the tax bill on the things you already own. The investment does well, and the real product is the losing. That is the pitch AQR, the quant shop founded by Cliff Asness, has spent years refining, and it has made "realize a loss" the most attractive trade in institutional finance.

This looks like one of those "Wait, how is that legal?" stories, and the honest version of the answer is interesting: the mechanism is legal, the losses are real for tax purposes, and the thing everyone is fighting about is what the strategy is. The most useful question is not whether AQR is doing something illegal; it is whether the machine is selling tax-loss harvesting or tax-loss manufacturing. The basic point is that tax-loss harvesting — selling the losers in your portfolio to offset the taxes on the gains elsewhere in your portfolio — is an old, ordinary, perfectly legal thing. AQR's version is old tax-loss harvesting in a new costume: add short positions and leverage, realize losses at a scale that has nothing to do with any loss the client actually suffered, and sell the result to the people with the largest tax bills.

Start with the ordinary version, because it is the reference point for how strange the upgrade is. For decades the sharpest investment professionals have argued that taxes are the biggest cost investors pay — Cliff Asness himself has cited a 1993 article by Rob Arnott arguing that investors lose more to taxes than they gain from beating the market. Direct indexing turned that observation into a product: instead of owning the S&P 500 through an ETF, you own five hundred individual stocks, and whenever one of them dips, you sell it, take the loss to offset a gain somewhere else, and buy a similar-but-not-identical stock to keep your exposure. The wash-sale rule — the IRS rule that disallows a loss if you buy back a "substantially identical" security within thirty days — is dodged by replacing the loser with a cousin, not a twin. That version of harvesting is mainstream, respectable, and low-drama. It is also small, because ordinary investors can only write off $3,000 of realized losses against ordinary income each year; the losses only matter at scale when they are offsetting large capital gains.

What the new version adds is the machine part. In a long/short tax-aware account, the manager holds a basket of long positions and a basket of short positions side by side, plus leverage. When something in the long basket falls, the manager sells it and takes the tax loss; the short leg keeps the economic exposure alive. Because the two legs offset each other, the client is not actually losing the money — the client is losing it for tax purposes. The realized losses can then absorb the gain from, say, finally selling the concentrated stock position that has been in the family since the pre-IPO days.

And then it stopped being a niche. By 2025 AQR had grown its tax-aware strategies from about $3 billion to roughly $70 billion; assets in the specific tax-aware long/short category stand at about $150 billion; the broader "tax alpha" universe has crossed the trillion-dollar line; and wealthy investors were still pouring about $1 billion a week into variations while the investigation was being written. By the end of 2025, industry rankings put AQR at No. 1 among the world's largest hedge fund managers, and AQR's total assets were above $140 billion by early 2026, with tax-aware products close to half the firm.

That ranking is the part of the story worth sitting with for a second. AQR did not get to the top of the hedge fund world on its old quant factors. The reporting on its rise is explicit that growth was driven substantially by surging demand for tax-aware strategies among wealthy individuals. The clients are billionaires, founders, venture capitalists, and early employees of public companies, many of them in high-tax California and New York or freshly moved to low-tax Texas and Florida. AQR even markets a version, Delphi Plus, pitched as a steady stream of losses to shelter high annual income. The whole business exists because of a gap in the rate schedule — the top federal rate on ordinary income is 37 percent while the top rate on long-term capital gains is 23.8 percent — and because wealthy people would rather pay a manager than pay that gap.

Now the fee math, because it is the part that should give an ordinary investor a moment of pause. The strategy does not work purely because the tax saving is huge; it has to work net of what the manager charges. An independent analysis this spring looked at the leveraged long/short version on a $10 million concentrated position and found that over twenty years the investor would pay about $2.7 million in fees while avoiding roughly $3.7 million in taxes — more than half as much in fees as the tax being avoided. After fees and costs, the strategy actually underperformed the boring alternative of selling and reinvesting in an index fund, and even assuming the manager added a half a percentage point of real stock-picking alpha, the strategy only won by about a coin flip. The polite translation of the label "diversification without a tax bill" is: a fee, paid every year, to have someone manufacture losses on your behalf.

Investor: "I want to diversify without paying tax." Manager: "Perfect — we'll realize losses for you, and you'll pay us more than half the amount of tax you didn't pay." The best case in that analysis was not a case where the client outsmarted the tax code; it was the case where the client died within ten years and the assets got a step-up in basis. The "double your money and get six times your money in losses" pitch is real, and the reason it can coexist with the fee math is that the intended client is not deciding between the strategy and an index fund. The intended client has already decided not to sell.

Which is where the boundary lives. Nobody disputes that the machine produces losses. The live argument is about what the machine is, and the US Treasury has started answering. In July, Kevin Salinger, deputy assistant secretary for tax policy, told an industry seminar that the department would not ignore aggressive tax planning, that some of the new strategies were "potentially abusive" and delivered outcomes Congress did not intend, and he urged investors to be cautious about offers that appear unusually attractive. (The concern apparently also brushed the ETF "heartbeat trade", the creation-redemption maneuver that keeps funds from realizing gains.) The Treasury offered no new rules — AQR called it an information-gathering exercise — but the warning did real things on the ground: Charles Schwab and Fidelity have been limiting new accounts for the aggressive strategies, and AQR has stopped publishing asset figures for the tax products while its own disclosures now acknowledge that the IRS could disallow the benefits, retroactively, and that penalties may apply. There is no public indication the IRS is investigating AQR. The bets on both sides are explicit — some clients told the investigation they are, in effect, confident the tax treatment holds.

The market's answer so far is: business as usual, mostly. Affiliated Managers Group, the publicly traded firm whose fund stables include some heavily exposed to AQR's tax business, fell about 7 percent when the Treasury comments landed, and shorts have been circling the name. But AMG has recovered: the stock traded near $356 this week, up close to 24 percent year to date, even though the most recent flow shows large block orders leaving the stock while retail money arrives. That is a market quietly pricing the classification argument as a slow problem rather than a fast one — the machine survives the warning, the sophisticated money trims anyway.

Strip the labels off, and the world's largest hedge fund is a machine for manufacturing losses, and its clients are paying fees to rent it. That is the joke hiding inside the respectable name: the "hedge" in the world's biggest hedge fund is a hedge against its own clients' tax bills. The outcome for all parties depends on one classification call — whether the tax code treats "realize your losses" as a right everyone has, or treats a factory that sells losses as a tax product. The Treasury just moved one step down that road, the brokers have quietly closed a few doors, and the managers, the ones being paid either way, have already structured the business to collect no matter how it lands. One sentence of IRS guidance either direction will move more wealth than any stock factor AQR ever traded, and the strange thing is that nobody on either side of the bet seems particularly surprised.