Direct Indexing vs ETF: Your Bracket Decides at 32%

Two investors each put $500,000 into a broad US large-cap strategy. Maya is married filing jointly, lives in Palo Alto, California, and lands in the 32% federal bracket. Dale is single, lives in Austin, Texas, and sits in the 12% bracket. The robo-advisor pitch is identical for both: pay 0.40% for a separately managed account that holds 200–500 individual stocks tracking the S&P 500, so software can harvest losses on the laggards while the index rides up.

Maya should take the deal. Dale should buy a 0.03% S&P 500 ETF and ignore the pitch entirely. Same balance, same strategy, opposite answer — because the value of a harvested loss is a function of your marginal rate, and Maya’s rate is roughly four times Dale’s. The decision turns on one number: your combined capital-gains rate. Above about 25%, direct indexing wins. Below it, the fee wins.

What a harvested loss is actually worth

Direct indexing’s entire edge over an ETF is tax-loss harvesting at the individual-security level. When a single holding dips below cost basis, the software sells it, books the loss, and buys a not-substantially-identical replacement to keep the index exposure. The booked loss is the product. Its value equals the loss amount × the tax rate it offsets. So the rate is the whole game.

Harvested losses get used in a fixed order under IRC §1211 and §1212:

1. Short-term losses net against short-term gains — the most valuable match, because short-term gains are taxed at your full ordinary rate.
2. Long-term losses net against long-term gains — taxed at the preferential 0%/15%/20% LTCG rates.
3. Net capital loss offsets up to $3,000 of ordinary income per year (IRC §1211(b)); single filers and MFJ share the same $3,000 ceiling.
4. Anything left carries forward indefinitely to future years (IRC §1212(b)).

A harvested loss is not free money — it lowers your cost basis in the replacement security, so it’s really a deferral plus a possible rate-arbitrage. The deferral compounds; the rate arbitrage (harvest at a high rate today, recognize gain at a lower rate or stepped-up basis later) is the durable win. Both scale with your marginal rate. At a high rate the deferral and arbitrage are large. At the 0% LTCG rate they collapse toward zero.

The combined-rate math, filer by filer

The relevant rate is your combined capital-gains rate: federal LTCG (or ordinary rate for short-term) + the 3.8% Net Investment Income Tax under IRC §1411 (lesser of net investment income or MAGI over $200K single / $250K MFJ) + your state rate. Here is the spread across the two filers and a middle case.

Maya’s long-term harvest is worth 37.1 cents on the dollar, and her short-term harvest — offsetting a short-term gain taxed at her 32% ordinary rate + 3.8% NIIT + 13.3% California (which taxes capital gains as ordinary income, no preferential rate) — is worth nearly 49 cents. Dale’s harvest is worth essentially zero: in the 12% ordinary bracket his taxable income is under the 2026 LTCG 0% ceiling ($48,350 single), so his long-term gains are already untaxed and there is nothing for the loss to shield. The mid-case New Yorker sits right on the 25% line — the genuine toss-up zone.

The crossover: where the fee catches the harvest

Direct indexing costs roughly 0.30%–0.40% of assets per year. A broad S&P 500 ETF costs about 0.03%. So you pay an extra ~0.32% — about $1,600/year on a $500K account — for the privilege of harvesting. The question is whether the after-tax value of the losses you harvest exceeds that drag.

Industry and academic estimates put the “tax alpha” from harvesting at roughly 1%–1.5% of portfolio value per year in early years (it decays as unrealized losses get used up). That 1%–1.5% is the loss volume, not the cash benefit. Multiply it by your combined rate to get the actual benefit:

That is the crossover. Around a 25% combined rate, the harvest benefit and the fee drag cancel. Maya clears it; the New Yorker is a wash; Dale is underwater by the full fee every year. This is also why “same balance” does not settle the question — two people with identical $500K accounts can land on opposite sides of the line purely on bracket and state.

Worked decision: Maya (CA, 32%) vs Dale (TX, 12%)

Maya: buy the direct-indexing SMA

Maya holds RSUs from a tech employer and routinely realizes short-term gains when she diversifies out of her concentrated position. Those gains are taxed at 32% + 3.8% NIIT + 13.3% CA = 49.1%. Direct indexing lets her bank short-term losses inside the SMA to absorb those gains. If the SMA throws off $6,250 of harvestable loss in year one (1.25% of $500K) and she uses it against short-term RSU gains, that loss is worth $3,069 in tax saved. The extra fee cost her ~$1,600. Net win: about $1,469 — before counting the multi-year carryforward and the basis arbitrage when she eventually donates appreciated lots or gets a step-up.

Dale: buy the ETF

Dale’s taxable income keeps him in the 0% long-term capital-gains bracket (under $48,350 single for 2026). Any long-term gain he realizes is untaxed federally, and Texas has no income tax. A harvested loss offsets a 0% gain — worth nothing. He could still use up to $3,000/year against ordinary income at 12% (a $360 benefit), but that’s less than a quarter of the $1,600 extra fee. Dale buys VOO or a comparable S&P 500 ETF at 0.03%, saves the 0.32% every single year, and lets the lower fee compound for decades. Over 30 years on a growing balance, that fee gap alone is worth tens of thousands.

What most people miss

The dominant myth is that account size decides direct indexing — that once you cross some $100K or $250K minimum, it’s automatically worth it. It isn’t. The platform minimum tells you whether you’re allowed to use direct indexing; your marginal rate tells you whether you should. A $2M Florida retiree in the 0% LTCG bracket gets less harvesting value than a $250K California engineer in the 32% bracket. Three more things people overlook:

• Harvesting decays. The 1%–1.5% loss volume is a front-loaded estimate. After a few years of a rising market, most lots are above basis and there is little left to harvest. The fee, however, is forever. The crossover gets harder to clear as the account ages — model the later years, not just year one.
• The wash-sale rule (IRC §1091) can void the harvest. Buying a substantially identical security within 30 days before or after the sale disallows the loss. Holding the same S&P 500 ETF in your 401(k) or buying it back in another account can trip it. This is the failure mode that turns paper tax alpha into nothing.
• Low basis becomes a future liability. Every harvest lowers your basis, so you’re storing up a larger gain for later. The strategy only pays if you exit at a lower rate, donate the appreciated lots to charity, or die holding them and get a §1014 step-up. Plan the exit when you plan the entry.

State tax is the swing factor most pitches ignore

Federal LTCG + NIIT tops out at 23.8%. That alone barely clears the 25% crossover — meaning in a no-income-tax state, direct indexing is a close call even for high earners. The state rate is what tips it decisively. California adds up to 13.3% (taxing gains as ordinary income), New York 10.9%, New Jersey 10.75%, Oregon 9.9%, Hawaii 11% — pushing combined long-term rates to 33%–37% and short-term rates near 50%. By contrast, Texas, Florida, Nevada, Washington, South Dakota, Tennessee, Wyoming, Alaska and New Hampshire impose no income tax on this income (Washington’s 7% gains tax only applies to long-term gains over $250K). Same federal facts, and the answer flips on which state line you live behind.

This is why the high-tax-state high-earner is direct indexing’s home turf, and the no-tax-state low-bracket investor is exactly who should not buy it — even though both may be shown the identical sales deck.

The decision lever

Run one calculation before you sign anything: your combined capital-gains rate = federal LTCG (or your ordinary rate, if you harvest against short-term gains) + 3.8% NIIT if your MAGI is over $200K single / $250K MFJ + your state’s rate on gains. If that number is comfortably above 25% — and you actually have gains to offset — direct indexing’s harvesting clears the fee and the SMA is worth it. If it’s at or below 25%, or you sit in the 0% LTCG bracket, the extra 0.30%–0.40% fee is a permanent drag the harvesting can’t repay. Buy the low-fee S&P 500 ETF and let the fee gap compound in your favor.