Beancount.io LogoBeancount.io

Direct Indexing for Tax-Loss Harvesting in 2026: The Loss Machine ETFs Cannot Build

14 min readMike ThriftMike Thrift
Direct Indexing for Tax-Loss Harvesting in 2026: The Loss Machine ETFs Cannot Build

In 2025, the S&P 500 climbed roughly 25%. Almost every diversified fund investor saw their account balance go up — and almost no one harvested a tax loss. Yet a single direct-indexing manager, BlackRock's Aperio, quietly captured an estimated $3 billion in tax-deductible losses across its book during that same up year. How is that possible? And what does it mean for the rest of us watching ETF investors miss those losses every single year?

The short answer: when you own an index fund, you own one security. When you own a direct index, you own five hundred. Those five hundred lots dance independently. Even when the index roars, dozens of constituents are underwater, and in a properly designed direct-indexing account each of them is a harvestable loss waiting to be turned into a tax refund.

Direct indexing is no longer a Wall Street curiosity. Cerulli projects direct-indexing assets will surpass $800 billion by the end of 2026, on pace to reach roughly a third of the retail SMA market. Minimums have collapsed from $1 million to as little as $5,000. Fees that used to run 50 basis points now start at 9. The strategy has become the most credible "third wave" of indexing since Vanguard launched the first index mutual fund in 1976 and SPDR launched the first ETF in 1993.

This guide walks through what direct indexing is in 2026, the tax math that makes it work, the statutes that constrain it, the platforms competing for your assets, and the very specific situations where it earns its keep — and a few where it doesn't.

The Structural Asymmetry: Why ETFs Cannot Harvest Losses For You

ETFs are famously tax-efficient. Most never distribute capital gains to shareholders. The mechanism is the in-kind redemption: when an authorized participant redeems a basket of ETF shares, the fund hands over a slice of its lowest-basis stock positions, tax-free, under Subchapter M. That flushes unrealized gains out of the fund without triggering a taxable event for remaining shareholders.

This same mechanism, however, is also a one-way valve. ETFs can shed gains through in-kind redemption, but they cannot pass realized losses to you, the individual shareholder. When a stock inside SPY drops below cost, the fund either holds the position (loss stays unrealized) or sells it (loss stays at the fund level and is used to offset future fund-level gains, benefiting all shareholders diffusely). You get nothing on your Form 1099-B.

Your only opportunity to harvest a loss from an ETF is when the entire fund drops below your basis. In most years, the S&P 500 finishes higher, and your loss harvesting opportunities round to zero. The structure prevents what direct indexing makes inevitable: per-security loss capture even in up markets.

What Direct Indexing Actually Is

A direct-indexing account is a separately managed account (SMA) in which you personally own anywhere from 150 to 500 individual stocks chosen and weighted to track a target index — most commonly the S&P 500, but also the Russell 1000, MSCI USA, or Russell 3000. There is no fund wrapper. There is no fund ticker. There is just your name on the title of every share, with a unique cost basis for every lot.

Fractional shares are what made the modern version of this possible. A decade ago, replicating the S&P 500 required buying 500 whole shares — putting the strategy out of reach for anyone without seven figures. Today every major broker fractionalizes, so a $5,000 account can replicate index weights within basis points.

What does that buy you? Five hundred independent tax events per year, instead of one. Five hundred opportunities to sell a loser, harvest the loss, and immediately buy a correlated replacement — without disturbing your index exposure. That is the whole game.

The Tax Alpha: How Much Is It Really?

Industry estimates of "tax alpha" — the after-tax return improvement direct indexing produces versus a plain ETF — converge on a range of 0.5% to 2.0% annualized in early years, decaying with time. The decay curve matters as much as the headline:

  • Years 1–3: 1.5% to 2.5% tax alpha. Fresh basis means most lots are still close enough to harvest opportunities.
  • Years 3–5: 0.8% to 1.5%. The lowest-hanging losses are gone.
  • Years 5–10: 0.3% to 0.8%. Most lots have appreciated past their original basis.
  • Year 10+: Often less than 0.3% unless you keep adding cash or the market drops sharply.

A 2025 study by wealth platform Range, drawing on real client accounts, found that direct-indexed portfolios harvested $18,281 of losses per account on average (average balance $413,759), while ETF-only accounts harvested just $4,808 — a 3.8x advantage. At a top federal-plus-state marginal rate, that's roughly $6,700 of tax savings per account, per year.

For a high earner in California paying 37% federal, 13.3% state, and 3.8% net investment income tax — an effective marginal rate near 54% on short-term gains — every dollar of harvested loss is worth more than half a dollar in cash.

The Rules You Cannot Forget

Direct indexing operates under four statutes and one revenue ruling that have not changed for 2026 but bite anyone who ignores them.

IRC §1091 — The Wash-Sale Rule

If you sell a security at a loss and buy a "substantially identical" security within 30 days before or after the sale (a 61-day window including the day of sale), the loss is disallowed for the current year. The disallowed amount is added to the basis of the replacement security under §1091(d), so it isn't destroyed — just deferred until you sell the replacement.

This is why every direct-indexing platform uses a "correlated replacement" strategy. Sell Coca-Cola at a loss, buy PepsiCo at the same exposure for 31+ days, then swap back. The index tracking error stays within basis points, and the loss is locked in.

IRC §1211 — The $3,000 Ordinary-Income Offset

Net capital losses offset capital gains dollar-for-dollar. Excess net losses can offset up to $3,000 of ordinary income annually ($1,500 if married filing separately). This cap has been the same since 1978 and is not indexed for inflation — a recurring criticism in tax policy circles.

IRC §1212 — Indefinite Carryover

Capital-loss carryovers do not expire. A 30-year-old who harvests $200,000 of losses in a 2026 drawdown can carry the unused balance forward indefinitely — applying $3,000 against ordinary income each year, plus using the rest against future capital gains. The losses extinguish only at death.

Rev. Rul. 2008-5 — The IRA Trap

If you sell a stock at a loss in your taxable account and the same stock (or a substantially identical one) is purchased within 30 days inside your IRA or Roth IRA, the loss is permanently disallowed. Unlike a regular wash sale, there is no basis adjustment to the IRA shares — the loss simply vanishes.

This is the single most common direct-indexing mistake. Your spouse's 401(k) auto-investing into an S&P 500 fund every two weeks is buying every single stock in your direct-indexing SMA. If you're harvesting Apple at a loss in your SMA on the same week the target-date fund inside your spouse's 401(k) buys Apple at its scheduled contribution date, you've just deleted that loss forever. Most brokers do not flag this — they report wash sales only within a single account. The cross-account work falls on you, your tax preparer, or the AI overlay tools (Mezzi, Frec, and Wealthfront now coordinate across linked accounts).

2026 Capital Gains Brackets

The federal long-term capital gains 0% bracket runs to $49,450 for single filers ($98,900 MFJ). The 20% top bracket starts above $545,500 single ($613,700 MFJ). The 3.8% Net Investment Income Tax under §1411 applies on top of that for households with modified AGI above $200,000 single or $250,000 MFJ.

Tracking the Index Within Basis Points

A direct-indexing manager doesn't typically buy all 500 names — they buy 150 to 250 chosen by an optimizer to minimize ex-ante tracking error subject to sector and factor constraints. Liquid mega-caps (Apple, Microsoft, Nvidia, Alphabet, Amazon) are nearly always held at full weight. Smaller names get sampled.

Typical tracking error targets sit at 20 to 100 basis points annualized, which is to say, in any given year your account return will deviate from the index by less than a percentage point. Custom screens — ESG exclusions, concentrated-stock offsets, sector tilts — can push that to 150–250 bps, but most investors stay close to neutral.

Who Should (and Shouldn't) Use Direct Indexing in 2026

Direct indexing is not for everyone. The math works best for:

  • Households in the 32% federal marginal bracket or higher (roughly $200K+ single, $400K+ MFJ).
  • Residents of high-tax states: California, New York, New Jersey, Hawaii, Oregon.
  • Investors with a recurring pipeline of capital gains: RSU-heavy tech employees, founders post-liquidity event, hedge fund and PE partners with carry distributions, active traders.
  • Accounts of $500,000 or larger in taxable. Below that, the fee premium relative to a 3-bp ETF often eats the tax alpha. (Lower minimums exist at Frec and Wealthfront, but the absolute dollar benefit shrinks quickly.)

Direct indexing makes less sense for:

  • IRAs and Roth IRAs (no taxes, nothing to harvest).
  • Investors in the 0% or 15% LTCG bracket.
  • Accounts where the investor expects to never sell (e.g., low-basis "buy and hold until death" portfolios that will get a stepped-up basis under §1014).
  • Small accounts in low-tax states.

The Loss-Exhaustion Problem

The single most underdiscussed weakness of direct indexing is that it has a shelf life. Every loss you harvest lowers a lot's basis — or replaces it with a different security that may itself appreciate. Over time, the population of lots sitting below water shrinks. By year 5, in a generally rising market, most accounts have harvested 70–80% of the lifetime tax alpha they will ever produce. By year 10, the account often looks like a single highly-appreciated position with very little remaining loss capacity.

That leaves you holding a portfolio with massive embedded gains. Switching providers may trigger a forced liquidation. Even rebalancing or rotating into a different index produces a tax event. The strategy can paint you into a corner.

There are three mitigation strategies. First, keep adding new cash; fresh contributions create new lots with fresh basis. Second, ride out market drawdowns — a 20% correction resets the entire loss-harvesting opportunity set. Third, layer on a long/short "tax-aware extension" (130/30 or 250/250 portfolios from managers like AQR, Aperio, or Quantinno) that engineers loss capacity through short legs. These are sophisticated and typically reserved for $5 million-plus accounts.

The 2026 Platform Landscape

Direct indexing has democratized faster than almost any wealth-management product in memory. Current options:

PlatformMinimumFee (bps)
Frec S&P 500$20,0009
Wealthfront S&P 500 Direct$5,0009
Wealthfront US Direct Indexing$100,00025
Fidelity Managed FidFolios$5,00020–70
Schwab Personalized Indexing$100,00040
Vanguard Personalized Indexing$250,000 (advisor only)~20
Parametric (Morgan Stanley)$250,000–$1M25–35
BlackRock Aperio$1M+20–35

What cost 50 basis points and required $1 million in 2018 now costs 9 basis points at a $5,000 minimum. The forces driving this: zero-commission trading, fractional shares, cloud-based optimization, and an arms race between Fidelity, Schwab, Vanguard, BlackRock, and a wave of fintech entrants.

The Concentrated-Stock Unwind Playbook

The single highest-leverage use of direct indexing in 2026 is paired with a concentrated low-basis position — founder stock, RSU vests, an executive's company holdings.

A founder selling $5 million of vested stock at a 10% basis faces $4.5 million of long-term capital gains. At 23.8% federal plus state, that's well over $1 million in tax. A direct-indexing SMA funded alongside the gradual sell-down can generate $30,000 to $100,000+ of harvested losses per year, offsetting a sliver of the gain — and more importantly, soaking up future RSU vests as they continue.

Right after a company sale, IPO unlock, or hedge-fund liquidity event is the highest-value moment to open a direct-indexing account: fresh cash, fresh basis, and several years of high tax alpha before depletion.

Five Mistakes That Destroy Your Tax Alpha

  1. Cross-account wash sales. Your IRA, your spouse's 401(k), your robo-advisor — the IRS aggregates across the household. Brokers do not. Use a coordinator tool, or at minimum, do not hold the same index in both taxable and retirement accounts.
  2. The Rev. Rul. 2008-5 IRA trap. Loss in taxable + same stock buy in IRA within 30 days = loss permanently disallowed. No basis step-up. Disable automatic IRA contributions during loss-harvesting windows or route them to different securities.
  3. Charitable misalignment. When donating from your SMA to a donor-advised fund, specify high-appreciated, low-basis lots — those give you a fair-market-value deduction and avoid the embedded gain. Most platforms allow lot-by-lot specification but only on request.
  4. Embedded-gain blindness. By year 7 or 8 your SMA may carry a 30–40% unrealized gain. Plan the eventual exit: gradual gifting, charitable bequest, or holding until death for stepped-up basis under §1014.
  5. Direct indexing in accounts that are too small. A $50,000 SMA at 9 basis points may save you a few hundred dollars in losses while complicating your tax return by 30 pages of Form 8949. The juice is not always worth the squeeze.

The Bookkeeping Reality

Here is the part the platforms don't put in their marketing decks. A buy-and-hold VOO investor has maybe four transactions a year — quarterly dividend reinvestments. A direct-indexing account generates hundreds to thousands of lot-level trades per year across hundreds of tickers.

Your Form 1099-B is no longer a one-pager. It becomes a 20- to 50-page document. Form 8949 — where each disposition gets its own line — explodes accordingly. Wash-sale tracking, cost-basis reconciliation, and cross-account coordination become real work.

This is where plain-text accounting earns its keep. Accurate per-lot bookkeeping isn't optional when you're harvesting losses at the constituent level — every cost basis, every lot date, every wash-sale adjustment has to be tracked precisely or you'll miss losses (leaving money on the table) or double-count them (inviting an IRS notice).

Keep Your Lot-Level Records Bulletproof

A direct-indexing SMA can generate over a thousand individual tax lots in a year — and every wash sale, every replacement security, every harvested loss needs to be tracked at the lot level. Beancount.io provides plain-text accounting that natively models per-lot cost basis with {cost, date} annotations, importing broker CSVs without losing lot identity. The result: complete transparency, no vendor lock-in, and tax reports you can derive programmatically rather than reconcile manually. Get started for free and see why investors with complex portfolios are switching to plain-text accounting that can keep up with the trade volume direct indexing creates.

2026 Outlook

Direct indexing is in the middle of its breakout moment. Cerulli projects $800 billion of assets by year-end, growing at a 12.3% CAGR — faster than ETFs, mutual funds, or any other SMA category. Only about 14% of advisors currently use it, which means the runway is long.

The headline trends to watch:

  • AI-driven optimization rebalancing daily across linked accounts, with automatic wash-sale checks across IRA, 401(k), and spousal accounts. Frec, Wealthfront, and Mezzi are leaders.
  • Fee compression continuing. Frec's 9 basis points may not be the floor.
  • Long/short tax-aware overlays moving downmarket. What used to be a $5M-minimum AQR strategy is now offered at $1M at multiple shops.
  • Tax-aware ESG and concentrated-stock customization as the table-stakes differentiators.

If you have the income, the gains, the marginal rate, and the bookkeeping discipline to handle hundreds of lots a year, direct indexing in 2026 is the closest thing to free money in personal finance — limited only by the half-life of your own basis and the wash-sale rule. ETFs cannot give you this. Mutual funds cannot give you this. Only owning the stocks themselves can.