Exchange Fund vs. CRT vs. Direct Indexing: A Decision Framework for Concentrated Stock

The problem (refresher)

A concentrated single-stock position above 10–15% of investable net worth carries idiosyncratic risk that classical portfolio theory cannot diversify away without selling — and selling triggers federal long-term capital gains under IRC Section 1(h), the Section 1411 net investment income tax, and state tax in high-bracket states. A long-horizon study of the U.S. equity market by Hendrik Bessembinder (Journal of Financial Economics, 2018) found that the majority of individual common stocks underperformed Treasury bills over their full trading lives — concentration is not, on average, a reward-bearing risk. Outright sale is rarely the right move for a position with material embedded gain. For the diagnostic framing of the problem itself, see the concentrated-stock guide.

The two “exchange fund” categories

See the disambiguation block above. The remainder of this page treats the two categories as separate strategies, not as variants of a single product, and the comparison table breaks out their statutory bases (Section 721 vs Section 351) explicitly. Long-form discussion of each lands in the strategy guide pages.

Strategy 1 — Section 721 traditional exchange funds

The Section 721 traditional exchange fund is the original “exchange fund” — a pooled limited partnership that accepts contributions of appreciated stock from many contributors and issues each contributor a partnership interest in exchange. IRC Section 721 treats the contribution as a non-recognition event: no federal capital gain is realized, the contributor's basis carries over into the partnership interest, and the embedded gain is preserved inside the structure for later recognition.

The ~7-year hold is a mechanical consequence of partnership rules, not Section 721 itself. Under IRC Section 731 — the “mixing-bowl” rule on in-kind distribution of contributed property — a contributor who receives back the same (or similar) property within seven years is treated as if the partnership had sold the property. After seven years, an in-kind distribution of a diversified basket out of the partnership becomes the standard exit, with the contributor's original basis carried into the distributed shares. The fund itself must satisfy the Section 721(b) / Section 351(e)(1) investment-company test, which is why ~20% of fund assets are typically held in non-marketable qualifying assets (real estate, private equity).

Contribution minimums are typically $1M; some providers gate at $5M. Annual all-in fees run roughly 0.85–1.25% of assets plus administrative costs. Eaton Vance (now part of Morgan Stanley), Goldman Sachs, and a handful of large wirehouses are the canonical providers; access generally requires an existing wirehouse relationship and a contribution timed to a fund's open period.

Worked figure. A $1M META position with a $300K basis and a $700K embedded gain, sold today by a single California resident at the top federal bracket, faces approximately $140K federal LTCG (20%), $26.6K Section 1411 NIIT (3.8%), and ~$94K California ordinary-rate state tax (13.3%) — roughly $260K all-in, leaving ~$740K to reinvest. Contributing the same $1M position to a Section 721 fund defers the entire $260K. If the partnership compounds at ~7% net annualized over the seven-year hold and the gain is recognized at exit, the deferred-tax benefit translates to roughly $130K of additional after-tax wealth versus selling and reinvesting today, before considering diversification benefit. Numbers are illustrative; substitute your actual basis, state, and projected return before drawing a conclusion.

Long-form prose covering qualifying-fund mechanics, the IRR breakeven analysis, and the canonical provider lineup lives in the dedicated Section 721 exchange funds guide.

Strategy 2 — Section 351 ETF-conversion exchange funds

The Section 351 ETF-conversion structure is the newer answer (commercial issuance from 2024 onward) for contributors who cannot or will not lock up capital for seven years. Multiple contributors transfer appreciated stock to a newly formed regulated investment company (RIC) that elects to operate as an ETF; under IRC Section 351, the transfer is non-recognition if the contributors collectively hold at least 80% of the new entity immediately after and the entity satisfies the investment-company diversification thresholds at formation. Each contributor receives ETF shares with a basis equal to the basis of the contributed stock; subsequent sale of the ETF shares is what triggers recognition.

Kitces.com's March 2025 deep dive by Henry-Moreland and Sullivan documents the 25/50 diversification rule: at formation, no single security may exceed 25% of the RIC's portfolio, and the top five securities together may not exceed 50%. This is why Section 351 ETFs are typically formed with multiple contributors holding different concentrated names, not a single contributor.

Practical features that distinguish Section 351 from Section 721: there is no statutory 7-year lockup, daily liquidity is available once the ETF is seasoned, and contribution minimums run $250K–$1M depending on issuer. Annual expense ratios are typically 0.40–0.85%; some issuers charge an additional contribution or structuring fee on top. Cambria and Cache are the most-cited issuers; the market is young and the list is changing quickly.

Worked figure. A $500K NVDA position with a $80K basis (a typical post-2023 RSU stack for a Bay-Area engineer) contributed to a Section 351 ETF defers federal LTCG of ~$84K, NIIT of ~$16K, and California ordinary-rate state tax of ~$56K — about $156K all-in. Versus a Section 721 fund the contributor gives up roughly 2x the embedded benefit of the full 7-year deferral but in exchange gets daily-liquid access to the ETF shares; the contributor can sell a slice if the household needs cash, and the basis of any sold lot is the original basis of the contributed shares allocated pro-rata. The Section 351 wins where the contributor is below the $1M Section 721 minimum, where the seven-year lockup is a non-starter, or where the contributor wants the option to unwind on a gradual schedule.

Long-form prose covering the 80%-control test, the diversification mechanics at formation, the seasoning arithmetic, and a side-by-side issuer comparison lives in the Section 351 ETF-conversion guide.

Strategy 3 — Charitable Remainder Trust (Section 664)

A charitable remainder trust under IRC Section 664 is a split-interest, tax-exempt trust that pays a non- charitable beneficiary (typically the donor) an annuity amount or a unitrust amount for a term of years (up to 20) or for the donor's life, with whatever remains at the end going to a qualified charity. The trust itself is exempt from federal income tax under Section 664(c). The donor contributes appreciated stock; the trust sells it without recognizing capital gain; reinvests the proceeds in a diversified portfolio; and pays the donor on schedule.

Two structural choices matter. A charitable remainder unitrust (CRUT) pays a fixed percentage of the trust's fair market value each year, so the income tracks portfolio growth (or decline). A charitable remainder annuity trust (CRAT) pays a fixed dollar amount each year, set at funding and unchanged thereafter. CRUTs are the more common choice for appreciated-stock funding because they preserve the donor's exposure to long-run growth.

The donor receives a partial charitable deduction at funding equal to the present value of the projected charitable remainder, calculated using the IRS Section 7520 discount rate. Distributions to the donor are taxed under the four-tier rule of Section 664(b): ordinary income first, capital gain second, tax-exempt interest third, and return-of-principal last. AICPA's Tax Adviser September 2025 walkthrough covers the practitioner mechanics including the Section 1411 NIIT interaction and the trustee's ongoing obligations.

Worked figure. A 55-year-old donor contributes $1M of appreciated stock ($300K basis) to a 20- year CRUT with a 6% payout rate, at a Section 7520 rate near current levels. The trust sells the stock with no immediate capital-gain recognition. The upfront partial charitable deduction is roughly $150K (illustrative; depends on Section 7520, payout rate, and term). The donor receives an annual distribution starting at $60K, taxed under the Section 664(b) tiers as the trust realizes income from its diversified portfolio. Over 20 years the donor receives ~$1.2M+ of distributions on a present-value basis (depending on portfolio returns) and the residual is paid to the donor's named charity. CRT economics are most attractive when the donor genuinely wants to make a charitable gift; without that intent, the deduction does not pay for itself.

Long-form prose covering CRUT vs CRAT mechanics, the Section 7520 rate sensitivity, NIMCRUT/Flip CRUT variants, and the decision framework for when CRT beats an exchange fund lives in the charitable remainder trust guide.

Strategy 4 — Direct indexing with tax-loss harvesting

Direct indexing replicates a benchmark (most often the S&P 500 or a broad total-market index) at the security level inside a separately managed account, holding hundreds of individual stocks instead of an ETF. The structural value is that lot-level losses become available for harvest. When a single name dips, the SMA sells the lot at a loss, realizes the deductible capital loss under IRC Section 1091's wash-sale constraints, and replaces the position with a non-substantially-identical correlated stock so the index tracking remains tight.

Wealthfront's direct indexing white paper and BlackRock Aperio's institutional materials both put first-year harvested losses at roughly 1.5–2.0% of portfolio value during a year with normal market dispersion. The yield decays as the portfolio's average cost basis converges toward market — by year five typical ranges fall to 0.5–1.0% — but the compounding tax alpha while it lasts is real and applies against gains realized anywhere in the household, including the gradual unwind of the concentrated position.

Wash-sale risk is the operational constraint. Section 1091 applies across all of the donor's accounts (and a spouse's, if filing jointly), including IRAs. Most SMA providers manage internal wash-sale exposure but cannot see what the client trades elsewhere; the burden of avoiding parallel buys in the same names sits with the client. When the concentrated position is held in a different account at the same broker, providers typically exclude that ticker from the SMA universe at client instruction.

Worked figure. A $1M direct-indexing SMA in a year with normal dispersion harvests ~$18K of net realized losses. Used against $18K of long-term capital gain from the concentrated position, those losses are worth ~$5K in deferred federal tax (LTCG + NIIT) plus state. After a 0.30%/yr SMA fee ($3K), the after-fee tax alpha for the first year is roughly $2K, plus the optionality of carrying forward unused losses. Across years two through five the yield decays; the structure earns its fee when the household has an ongoing capital-gain pipeline (RSU vesting, concentrated unwind, secondary liquidity) rather than as a standalone bet.

Long-form prose covering harvest decay, factor-tilt variants, and the breakeven framework against the concentrated unwind speed lives in the direct indexing guide.

Strategy 5 — 130/30 long-short extension

A 130/30 long-short extension holds 130% long and 30% short for net 100% market exposure, with the short sleeve generating losses on names that rise (and the long sleeve harvesting on names that fall). The structural advantage over long-only direct indexing is that loss harvest now happens on both legs of the dispersion, not just the long leg. Alpha Architect's practitioner research and BlackRock Aperio's long-short white paper put the multi-year loss-harvest yield from a 130/30 at roughly 2x to 3x the equivalent long-only direct-indexing number; a peer-reviewed 2024 Journal of Asset Management paper on 130/30 loss harvesting reaches the same qualitative conclusion.

The trade-off is structure cost and operational complexity. 130/30 SMAs typically run 0.50–1.00%/yr management fee plus financing on the short sleeve; the structure requires a prime-broker relationship and is gated to accredited or qualified-purchaser clients at most issuers. Short-side tax rules are also more complex: Section 1091 wash-sale, Section 1233 short-sale rules, and Section 1259 constructive-sale rules collectively constrain how a short can be paired with a long without unintended recognition.

130/30 fits best when the household has a multi-year vest pipeline that keeps re-creating concentration. Cerity Partners' practitioner write-up on long-short equity frames the strategy specifically as a tax-aware overlay for high-net-worth tech employees with continuing RSU vesting — the loss yield earns its fee against a gain stream that is replenishing each quarter.

Worked figure. A $1M 130/30 SMA in a year of normal dispersion can produce ~$45K–$60K of net realized losses against a long-only equivalent at ~$18K. Applied to ~$50K of recurring RSU vesting gains in the same household, the loss harvest covers most or all of the year's realized concentrated-position gain at ordinary or LTCG rates. Net of a 0.75% management fee and short-side financing, the after-fee tax alpha is higher than long-only direct indexing for the same gain stream; whether it is higher than your bookkeeping tolerance is the question that determines fit.

Long-form prose with worked NVDA scenarios, the financing- cost arithmetic, and accreditation thresholds lives in the 130/30 long-short guide.

Side-by-side comparison

The table below is the canonical reference for the rest of this page. Statutory basis is the column most readers overlook; it determines the lockup, the qualifying-asset rules, and the tax mechanics on exit. Every figure traces to a statutory section or a published provider document — see the deep-dive guides for citation detail.

Fee ranges and minimums reflect issuer disclosures published through 2025 and are illustrative — quote against the current Form ADV / prospectus of any specific provider before relying on a number for planning purposes.

The decision tree

Five inputs decide the routing: position size, embedded gain, state of residence, charitable intent, and time horizon. Walk them in order — the first input that gates a branch is the one that resolves the decision.

1. Charitable intent? If the donor genuinely plans to make a multi-six-figure charitable gift in the next decade, the comparison is between a CRT and an outright donation of long-term appreciated stock to a public charity (or DAF). The CRT keeps the income; the outright gift gives the larger upfront deduction. A tax-deferral vehicle without charitable intent is the wrong tool for charitably motivated capital.
2. Position size and minimum gating. Below ~$250K, the structuring cost of every vehicle on this page eats the deferral benefit; the right answer is typically gradual sale paired with direct indexing in the rest of the household's assets. $250K–$1M routes to Section 351 ETF-conversion or direct indexing depending on the next input. $1M+ keeps Section 721 traditional exchange funds in the candidate set.
3. Time horizon and lockup tolerance. A seven-year hold on a $1M+ position is a hard constraint in real life — house purchase, business funding, college, divorce. If the household's honest answer to “can you afford to not touch this for seven years?” is no, Section 721 is out and the candidate set is Section 351 (full slug), direct indexing (gradual unwind), and CRT (if charitable intent re-enters the picture).
4. State of residence. California, New York, Hawaii, and other 9–14% high-bracket states add 9–14 points to the all-in tax cost of an outright sale and raise the deferral value of every vehicle on this page. Texas, Florida, and Washington (no state ordinary tax; Washington has a 7% capital-gains tax over a threshold) cut the deferral value roughly in half. State-of-domicile planning is its own conversation, and it changes the routing.
5. Embedded gain ratio. A position with a high basis-to-market ratio (e.g. RSUs vested in the last 18 months at near-current prices) has thin deferral value because there is little gain to defer. A position with a low basis-to-market ratio (founder stock, very early RSUs, ESPP held through a multi-bagger) is where the structures earn their fee. When the gain is <15% of market value, ordinary sale plus direct indexing usually dominates the structured options.
6. Re-concentration pressure. If the household is also vesting RSUs at a clip that re-creates concentration, a one-time diversification slug into Section 721 or Section 351 only solves the snapshot problem. Direct indexing or 130/30 needs to run in parallel against the ongoing vest stream so the concentration does not re-build the moment the structured deferral closes.

Working through the tree honestly almost always lands a household in two strategies, not one.

Combinations — when to layer two strategies

Most well-built plans for $1M+ concentrated positions use two strategies in series, not one.

1. Section 351 slug + direct indexing tail. A contributor with a $1.2M META position contributes $800K to a Section 351 ETF-conversion fund (deferring federal LTCG + NIIT + state tax on the $560K embedded gain inside the contributed slug — about $145K all-in for a California resident) and keeps the residual $400K in a direct- indexing SMA. The SMA harvests ~$6K–$8K of realized losses in year one, which the contributor uses to offset a gradual sale of the residual lots over 3–5 years. By year five the residual concentration is fully unwound, the Section 351 ETF shares are seasoned, and the household sits on a fully diversified portfolio.
2. Section 721 slug + 130/30 against ongoing vest. A senior engineer with a $2M position and continuing RSU vesting of $400K/year contributes $1.5M to a Section 721 traditional exchange fund (locks up for seven years, but defers ~$280K of all-in tax on the embedded gain) and runs a 130/30 SMA against the ongoing RSU vest. The long-short SMA harvests ~$25K–$50K of realized losses per year against the vesting stream, cutting the household's recurring concentrated-stock tax bill. By year seven the Section 721 partnership distribution becomes available and the 130/30 has substantially diversified the cumulative vest.
3. CRT + direct indexing. A charitably inclined contributor with $1M of appreciated stock contributes the full position to a 20-year CRUT and runs direct indexing in the rest of the taxable household. The CRUT distributions (~$60K/year initially) fund living expenses; the direct-indexing SMA harvests losses against any remaining concentrated-position activity (e.g. a public-company secondary, an option exercise, an ESPP disposition). The deduction at funding (~$150K) and the deferred-gain compounding inside the CRT combine with the SMA's tax alpha to produce an after-tax outcome that an outright sale would not match even at the same gross return.
4. QSBS-aware sequencing. A founder with eligible Section 1202 QSBS that has not yet reached the five-year hold should generally not contribute to a Section 721 or Section 351 fund. Contributing QSBS to a partnership under Section 721 typically breaks the Section 1202 holding period because the receiving interest is a partnership interest, not C-corp stock. Contributing to a Section 351 fund whose receiving entity is a regulated investment company is similar — RICs do not meet the Section 1202 active-business test, so the new shares do not preserve QSBS character. Either path impairs the exclusion in the typical commercial structure. The QSBS-aware path is to wait out the five-year hold, sell QSBS-eligible shares with the federal exclusion in place, and run direct indexing or 130/30 against any state ordinary-rate liability (California does not conform to Section 1202; NY partially conforms). Section 1045 rollover is available if the founder needs to exit before the five-year mark and reinvest in another QSBS within 60 days. The linked pages on Section 351 ETFs and Section 721 funds cover the QSBS-clock interaction in detail.

Every layered plan has a sequencing question: what happens first? Contribution to a Section 721 or Section 351 fund changes the basis and holding-period mechanics of the contributed shares; CRT funding changes the donor's charitable-deduction posture for the year of funding; direct-indexing and 130/30 SMAs need to be open before the realized gains they are supposed to offset hit the household's tax return. The order matters more than the choice.