Investing in Pre-IPO Shares

On November 6, 2025, Charles Schwab announced an agreement to buy Forge Global for about $660 million in cash, $45.00 a share; the deal closed on March 2, 2026. Forge was the largest and best-known marketplace for private-company stock in the United States, and on Schwab’s own account of the business more than $17 billion of private-company shares had been bought and sold on it. Less than four years earlier the same business had completed a SPAC merger with Motive Capital Corp and listed on the New York Stock Exchange in March 2022 at a valuation of roughly $2 billion. The venue that sold outsiders exposure to the private-market boom lost about two-thirds of its own value in the boom’s own currency, and the buyer was a discount broker.

Nothing about that transaction was a scandal. It is simply the clearest available price for the thing this guide is about. A market with a genuine reason to exist — employees and early investors who need cash years before an exit, and outside capital that wants in before the listing — was valued by a strategic acquirer at a third of what public markets paid for it in 2022, in the same months that the assets it traded were being marked up faster than ever.

That contradiction is the guide. Private-company marks and private-company realisations are two different numbers, produced by two different processes, and the distance between them is where an outside buyer’s return lives or dies. The distance does not always run against you: SpaceX listed on Nasdaq in June 2026 above the insider mark set six months earlier, and a Stripe tender in February 2026 cleared the valuation Stripe set at the 2021 peak. Section 7 puts both cases next to the three that went the other way. What does not change with the direction is the cost of the chain between your money and the shares, and that is what sections 6, 8 and 13 measure.

The wider private-credit market, its managers and its redemption mechanics are the ground of our hub flagship, Investing in Private Credit, with the listed and semi-liquid wrappers covered in Investing in Listed BDCs and Investing in Interval Funds and Non-Traded BDCs, and the same mark-versus-realisation problem worked on a court claim in Investing in Litigation Finance and on a song catalogue in Investing in Music Royalties; this guide takes one narrower question, which is what happens to a person who buys equity in a company that has not listed.

Three legal structures carry almost every dollar an outsider puts into a pre-IPO company, and they differ in what you own, who can take it away from you, and when. The fee stack, the tax treatment and the failure modes all follow from the structure rather than from the company.

The direct transfer

A direct transfer is the real thing: the seller’s shares move on the company’s stock ledger and your name appears on it. You own stock, with whatever rights that class carries. It is the cleanest outcome and the rarest one for a small buyer, because it requires the company’s consent, a stock transfer agreement, the waiver or expiry of every transfer restriction, and a minimum size that makes the paperwork worth the company’s while. On Hiive, whose live order book is the closest thing the market has to a screen, the standard minimum transaction is $25,000, rising to $100,000–$250,000 for some direct transfers in high-demand names.

The special purpose vehicle

An SPV is a single-purpose fund, almost always a Delaware LLC or limited partnership, that buys the shares and sells you a membership interest in itself. The SPV appears on the company’s ledger; you do not. You own a claim on a fund whose only asset is stock. It is the default for retail-sized tickets, because a company will approve one transfer to one entity far more readily than a hundred transfers to a hundred strangers, and because a fund interest can be sold in $5,000 pieces where a share transfer cannot.

The SPV is where the economics get complicated. It has a manager who charges for managing it, it may charge carried interest on gains, and it carries administrative and legal costs. Critically, an SPV can own an interest in another SPV that owns the shares — a structure the trade literature calls layering — with a fee at each level. When you cannot name every entity between your money and the company’s stock ledger, you cannot compute what you are paying.

The forward contract

A forward is a bilateral contract in which you pay now for shares delivered on a future triggering event, usually an IPO or acquisition, while the seller keeps legal title in the meantime. As a Buzko Legal note on forward purchase agreements sets out, the attraction is that no company consent is needed at signing, because no transfer happens at signing; the disadvantages are that the seller may not be able to deliver, that the company may oppose the arrangement, and that you may wait years for settlement. A registered fund’s own risk disclosure puts it more bluntly: the market for forward contracts is substantially unregulated, can experience lengthy periods of illiquidity, and exposes the holder to counterparty, leverage, liquidity, pricing and volatility risk (Investment Managers Series Trust II, Form 497K, FY2026).

A forward is not stock. If the counterparty fails you are an unsecured creditor of a company you have never met, holding a contract that references a security you do not own.

Three channels move private-company stock directly, and only two of them are meaningfully open to an individual. A fourth route moves no stock at all: the registered fund wrapper of section 9 buys the exposure itself and sells you a share of the fund. Knowing which of the three an offer came from tells you most of what you need to know about the price you are being quoted.

Marketplaces

Forge Global is the largest, with more than $17 billion transacted across thousands of companies on Schwab’s account of it, and since March 2, 2026 it has been a Charles Schwab subsidiary. EquityZen was acquired by Morgan Stanley Wealth Management in January 2026; the following month Morgan Stanley cut buy- and sell-side fees to 2.5% from 5% for most transactions, to 2.0% above $1 million, and kept the industry’s lowest minimums: $5,000 for individual share purchases and $20,000 for curated funds. Hiive runs a live order book rather than a deal-by-deal negotiation, which means you can see resting bids and asks rather than a broker’s indication, and publishes its maximum commissions on its Form CRS.

The consolidation matters more than any single fee schedule. Between January and March 2026 the two largest independent marketplaces for private stock became subsidiaries of a wirehouse and a discount broker. That plausibly makes the market safer; it does not change the thing being sold, which is a fund interest in an illiquid asset priced off a model.

Company-sponsored tender offers

The cleanest liquidity here is the tender offer: the company itself, usually with an institutional buyer it has chosen, buys a fixed quantity of employee shares at a fixed price on a fixed date. Nasdaq Private Market, founded inside Nasdaq in 2013 and spun out in July 2021 as a standalone company with SVB, Citi, Goldman Sachs and Morgan Stanley as investors alongside Nasdaq, is built around running these programmes. Tenders solve every problem this guide describes — no markup, no SPV, no ROFR risk, a blessed price, settlement in weeks.

They are also almost entirely closed to you. The buy side of a tender is negotiated with existing investors and invited institutions, and while a company will occasionally admit an approved outsider on terms it sets itself, that is the rarest way into this market rather than a channel you can plan around. The best-priced, best-documented route into a late-stage private company is reserved for people already on the cap table. The routes open to an outsider are the ones the company tolerates rather than the ones it runs.

Sponsor-organised SPVs and the grey channel

The third channel is the individual sponsor who assembles an SPV and markets it through a network, a newsletter, a broker or a cold call. Some are legitimate and fairly priced; the SEC’s enforcement record in section 12 shows what the rest look like. The distinguishing feature is not the structure, which is the same LLC everyone uses, but whether the offering document discloses the sponsor’s acquisition cost.

A private company can carry four different prices on the same day, and they are not versions of one another. Understanding which one you are being quoted is the single most valuable piece of technique in this market.

The four prices

The last primary round is what a new investor paid the company for newly issued preferred stock. It is the headline number and systematically the highest of the four, because preferred stock carries liquidation preferences, ratchets, board seats and information rights that common stock does not. A “$50 billion company” is usually one whose most recent preferred share, multiplied by every share as if it were that share, comes to $50 billion.

The 409A valuation is the board’s determination, supported by an independent appraisal, of the fair market value of the common stock, used to set employee option strike prices. Because it prices common rather than preferred, and because a low number is good for employees and defensible to the IRS, it sits well below the headline. The gap between the preferred round price and the 409A common price is the market’s cleanest available measure of how much of a private valuation is structure rather than value.

The secondary market price is what somebody actually paid another shareholder for existing stock, usually common. It is the only one of the four produced by two parties with money at risk and no interest in the number being high.

The model price is what an index or a platform publishes. Forge Price, which drives the Forge Private Market Index, comes from a proprietary model incorporating publicly available primary round information, secondary transactions and indications of interest on Forge and other platforms. An indication of interest is not a trade, and a model blending trades with quotes will be smoother, more current and less realisable than a print.

What the spread tells you

When a secondary trade prints well below the last primary round, that is rarely a bearish signal about the company. It is usually a statement about what class of stock is changing hands and what the seller needs. An employee selling common stock to pay a tax bill on exercised options is not expressing a view; they are meeting a deadline. That is why the secondary discount to the last round persists across the market and across cycles, and why a buyer who reads it as a bargain has misread the mechanism.

The reverse case is the dangerous one. When an SPV offers you shares above the platform’s own last observed trade, the excess is the sponsor’s markup: a certain, immediate, permanent loss on day one, quantified in section 13.

Every late-stage private company controls who owns its stock, and the instrument that does the controlling is the right of first refusal. It is the reason a pre-IPO purchase can be agreed, priced, funded and then simply not happen.

How a ROFR works

A right of first refusal is a provision in the company’s charter, bylaws or shareholder agreement that requires a selling shareholder to present any third-party offer to the company first. The company then has a defined window — commonly 30 days — to buy the shares itself on the same price and terms, or to assign that right to an existing investor. If it does, the seller sells, but not to you. Your capital comes back, your months of diligence and paperwork are gone, and the price you agreed has just been used to buy stock for somebody the company preferred.

A co-sale right often sits alongside it, letting other holders join the sale pro rata and shrinking the block you were buying. Companies also impose outright transfer restrictions: consent requirements, bans on transfers to competitors, caps on holders of record, and blanket prohibitions on transfers to SPVs.

Why companies use it

The purpose is control of the cap table without renegotiating every sale, and it is rational. A company approaching an IPO wants a clean register, wants to avoid the reporting obligations that come with exceeding holder-of-record thresholds, and does not want its price discovered in public by a broker’s order book while it raises primary capital. Some of the most sought-after private companies go further, refusing to recognise SPV interests at all.

The consequence never appears in the marketing. Your trade is not done when you agree a price; it is done when the transfer is recorded, and between those two moments sit the ROFR window, the company’s consent, and the possibility of a refusal. The structures that route around the problem — forwards and SPVs — do so by not transferring anything, which is to say they solve a legal problem by creating a counterparty problem.

Invest Alternative collects the Forge Private Market Index daily as pre-ipo.fpmi, from Yahoo Finance, with observations from August 30, 2021 and a live file holding the most recent 500 sessions, from 223.90 on September 5, 2024 to 683.60 on September 3, 2026. Our Pre-IPO sub-index, built on that series and rebased to 100 on September 2, 2025, stood at 187.949 on September 8, 2026, up 87.79% over one year, at a 2.2% weight in our composite. Our whole-composite level moved 0.29% over the same year: the pre-IPO sleeve was, on our own numbers, the strongest thing in the alternative-asset universe we measure.

A 205% gain with a 5% drawdown

Across those 500 sessions the index compounded 205.3%, a gain of 3.05 times. Its deepest peak-to-trough fall was 5.44%, from 706.97 on July 20, 2026 to 668.53 on August 25, 2026. Its annualised daily volatility was 20.3%, which is roughly equity-like; its median absolute daily move across all 499 changes was 0.15%, which is not. Those three facts do not describe an asset. They describe a measurement process.

The resolution is in five sessions. Five of the 499 daily changes exceeded 4% in absolute terms: February 24, 2025 (+20.2%), July 31, 2025 (+4.2%), May 14, 2026 (+9.0%), July 9, 2026 (+12.1%) and July 22, 2026 (−5.0%). Multiplied together those five produce a factor of 1.45; the other 494 sessions produce 2.10. Almost half the two-year gain, in compounding terms, arrived on five days when a constituent was re-marked or the index reconstituted — the January 2026 rebalance added thirteen companies and removed ten, including Neuralink, Thinking Machines Lab and Polymarket — rather than on any day a buyer could have transacted at a moving price.

Why this matters more than the level

Strip the jumps out and the index’s annualised volatility falls to 8.7%. An asset that moves 0.15% on a median day and 20% on five days is being valued, not traded. That is not a criticism of Forge’s methodology, which is transparent about blending primary round information, secondary transactions and indications of interest, and about valuing daily with an annual rebalance on equal weights. It is a statement about what any index of private marks can be.

The practical consequence is that you cannot buy the index’s return. No vehicle replicates an equal-weighted, annually reconstituted basket of private companies at the model price, without a markup and with the ability to sell at the printed level; every route to the asset class charges you the difference between a mark and a trade. Our tape is a measure of sentiment about late-stage private valuations, not a return series an investor earned. The same distance between a carrying value and a realisation is the whole argument of our guide to Litigation Finance, where the mark is a fair-value estimate of a lawsuit that has not settled.

Four separate charges sit between an outside buyer and a private share, and only the first of them is normally quoted in the offer. Added together they are the largest single determinant of what a pre-IPO position returns, larger in most cases than the difference between a good company and a mediocre one.

The commission

The platform takes a transaction fee from each side, and the two published schedules sit a long way apart. Hiive’s Form CRS of June 1, 2026 lists maximum commissions of 4.85% for buyers and 5.75% for sellers, the buyer rate stepping down above $250,000 of transaction value and the seller rate above $500,000. EquityZen, after Morgan Stanley Wealth Management cut its schedule in February 2026, charges 2.5% to buyers up to $1 million, 2.0% above that, and 2.5% to sellers. A small-ticket round trip on the more expensive of the two costs about a tenth of the position before the asset does anything.

Forge has published no single schedule, which is itself worth knowing before you ask for a quote. Its own fee note describes success-based commissions charged only on a completed transaction, with seller fees of roughly 2% to 4% on most direct transactions and as low as zero; the Form CRS of its broker-dealer says the commission is typically 5% and may be higher below its $100,000 minimum transaction size. No revised schedule has been published since the Schwab acquisition closed on March 2, 2026, so the number that applies to you is the one you are quoted in writing.

The markup

The markup is the difference between what the sponsor paid for the shares and what it charges you for an interest in the vehicle holding them. It is not a fee in any document: it is embedded in the entry price, invisible unless the offering discloses the acquisition cost, and the largest number in this section.

The range is now on the record. A Forbes investigation of the market published on May 26, 2026 puts upfront access fees across the sponsor-organised channel anywhere from under 5% to as high as 18%, charged before any management fee or carry. It then works a three-layer structure on conventional 2-and-20 terms in which a $2 million investment growing to $10 million by the listing hands nearly $5 million to intermediaries before tax — close to five of the eight million dollars of gain, decided by how many vehicles the money passed through rather than by anything the company did.

The enforcement record puts the top of that range higher still. An SEC complaint filed on August 14, 2026 alleges investors were charged prices averaging about 46% above what the sponsor paid, while being told upfront fees were at most 12.5%; prosecutors put the median markup at Linqto, a platform that raised more than $450 million before its 2025 bankruptcy, at around 60%. Both cases are in section 12, and both were invisible to the buyer at the point of sale.

A markup does something worse than cost you money: it resets the base from which carried interest is computed. If a sponsor buys at $7.50, offers you an interest at $8.25 and takes 20% of gains above $8.25, it has booked the first $0.75 of your money and then charges you for earning the rest.

The management fee and the carry

Single-asset SPVs typically charge an annual management fee on committed capital and carried interest on gains; 1–2% and 10–20% are conventional. The fee runs for the life of the vehicle, which ends when the company lists or is acquired and which nobody can predict. A 1.5% annual fee on a position held seven years is 10.5% of capital, charged on an asset that produced no cash in the interim.

Layering

Each of those charges appears once per layer. When an SPV buys into another SPV, or a feeder sells interests in a fund holding a forward that references the shares, the management fees add, the carries compound and the markups stack. The trade literature identifies the same four risks each time: undisclosed manager incentives, fee compounding across layers, loss of control over timing and distribution, and pricing opacity. The combination that punishes hardest is many layers and a long hold, which is precisely what this asset class produces.

The best evidence about what pre-IPO buying returns is the cohort that bought at the last peak and has now had four to five years to resolve. The 2021 vintage is the case study because the marks of that year were the highest ever recorded, the exits that followed were public and dated, and the gap between the two is measurable. Read it as a cohort and not as a law: three of the four names below reached a public market below their 2021 mark, the fourth has since been marked above its own peak, and the largest listing of 2026 priced far above the mark that preceded it.

What the marks said and the listings paid

Four of the most heavily traded private names of 2021 make the point. Three of them have reached a public market, and in each of those three the first public price came in below the private mark that preceded it. Instacart was valued at about $39 billion in a March 2021 primary round and listed on September 19, 2023 at $30 a share, a valuation near $9.9 billion. On the way it cut its own internal common-stock valuation four times in 2022 alone — to about $24 billion in March, $15 billion in July, $13 billion in October and roughly $10 billion by the year’s end — each cut a 409A determination by an independent appraiser, and each one public knowledge long before the listing.

The other two ran the same way. Klarna was marked at about $45.6 billion in June 2021, raised again in July 2022 at about $6.7 billion, and listed on the New York Stock Exchange on September 10, 2025 at $40 a share, a valuation near $15.1 billion. Reddit was marked at $10 billion in a Fidelity-led August 2021 round and priced its IPO at $34 a share on March 20, 2024, a valuation around $6.5 billion, before rising sharply on its first day and afterwards. In all three the direction of travel between the last private number and the first public one was down, and in all three the private number had been public for years.

What that cost a secondary buyer

Consider a buyer of Instacart exposure through an SPV in mid-2021 at the round price, with a 10% markup and 2.5% in fees: their entry cost about $42.9 billion on the company’s own scale, the 2023 listing valued it near $9.9 billion, and a 180-day lockup stood between the listing and any sale. That is a loss of roughly three-quarters of capital on a company that was not a fraud, did not fail, and remains a functioning business with real revenue. The company was fine; the entry price was not. Between the mark and the resolution sat two to four years of management fees on a position that paid nothing, and every month of that delay was fee drag on a mark that was already too high.

The cases that cut the other way

Stripe is the fourth name, and it belongs in the record because it did the opposite. Marked at $95 billion in March 2021, it raised at about $50 billion in March 2023 — and then worked back past its own peak, with employee tender offers struck at $91.5 billion in February 2025, $106.7 billion in September 2025 and $159 billion in February 2026. A 2021 buyer of Stripe at the round price is well ahead. A 2021 buyer of the other three, at the same moment and through the same channel, is not.

The largest listing of 2026 went the same way. SpaceX set an insider share deal at about $800 billion in December 2025 and went public on Nasdaq on June 12, 2026 at $135 a share, raising roughly $75 billion at a valuation near $1.77 trillion, the largest IPO on record; it closed its first day at $161. A holder who bought at the December mark was well ahead within six months, and nothing in this guide argues otherwise.

What those two cases do not do is refund the cost of getting there. An outsider’s return in the good case still runs through the markup, the layers, the annual fee and the lockup set out in sections 6, 8 and 13; a SpaceX buyer three vehicles deep waited past the June listing to find out how many shares they owned, and some of them had bought through a special purpose vehicle the company was pruning off its own cap table before the offering.

The honest version of this section is therefore narrow and it is the one the rest of the guide is built on: the marks of a single cohort ran ahead of the prints, and the cost stack runs against you in every cohort. Direction is the part you cannot underwrite. Structure is the part you can.

The venue’s own record

The sharpest single data point is the one in the cold open. Forge Global was valued near $2 billion at its March 2022 SPAC listing and sold to Schwab for $660 million in a deal that closed on March 2, 2026, a fall of about two-thirds — a marked-to-market verdict on the economics of intermediating this market, delivered during the strongest two years private marks have recorded. If the toll booth lost two-thirds of its value while traffic was at a record, the toll was not the problem.

A listing does not make a pre-IPO position liquid. It starts a clock, and the clock runs while the price moves without you.

The underwriter lockup

The standard arrangement is a 180-day lockup: a contractual undertaking by pre-IPO holders, agreed with the underwriters rather than imposed by any rule, not to sell into the market after the offering. Terms vary deal by deal. Some agreements stagger early releases against elapsed time and against price thresholds, so that a tranche comes free once the stock has held a set premium to the offer price for a set number of days; direct listings often carry shorter restrictions or none at all. Because the terms are contractual, the only reliable source for them is the prospectus and the vehicle’s own agreement, not the coverage.

The SpaceX lockup sequence

The 2026 listing is the live case, and a holder still inside a vehicle needs the order of events rather than the headline. The disclosed sequence runs like this.

• June 12, 2026, the listing. Shares priced at $135 on Nasdaq and closed the first day at $161. Holders bound by the underwriters’ agreement could not sell at either price.
• August 6, 2026, the first release. More than 900 million shares came out of restriction early under the staggered terms, and the stock rose about 6% afterwards rather than falling — the reminder that the direction of a release is not knowable in advance either.
• December 8, 2026, the 180-day expiry. The main lockup ends. This is the date on which a direct holder becomes a seller if they choose to be, and the first date on which the market sees what the pre-IPO register wants to do.
• Into June 2027, the insider tail. A longer 366-day restriction binds Elon Musk and certain insiders, so supply keeps arriving for roughly six months after the general expiry.

For a direct holder that is a calendar. For an SPV holder each of those dates is a starting gun rather than a finish line: at expiry some vehicles distribute shares in kind and leave the decision to you, some sell and distribute cash on the manager’s view of timing, and some impose restrictions beyond the underwriter’s.

Layering stretches the wait in a way the marketing never mentions. A first-layer SPV typically has about thirty days to distribute once it receives shares, and every layer below it waits its turn, so a buyer three vehicles deep in SpaceX did not know the size of the position they actually owned until well after the listing. If you hold one of those interests through an expiry, the expiry is not your date; your date is that one plus the distribution period at every layer between you and the company, and the answer is in the operating agreement. Read it before the expiry rather than after.

The first public price is the honest one

Everything before a listing is an estimate; the opening print is the first number produced by people who can be wrong in public. The 2021 cohort’s record is that the estimate ran ahead of the print, and there is a structural reason it should: private marks are set by the party with the most to gain from a high number, refreshed slowly, and blended from primary rounds whose price includes preferences that common stock does not carry. Public prices are set by strangers who may short. The rule that follows is unglamorous — underwrite a pre-IPO position to the first public price, not to the next private round — because an entry price that only works if the company lists above its last private mark is a bet on the marks rather than on the business.

Several registered vehicles offer pre-IPO exposure without accreditation, and the way they are priced is a lesson in what happens when a fixed pool of illiquid assets meets unconstrained retail demand.

The listed fund and its premium

A listed closed-end fund can only respond to demand through its price: the share count changes slowly and the private portfolio is repriced quarterly. Destiny Tech100 (DXYZ), a listed closed-end fund holding stakes in private technology companies, is the clearest demonstration of what that does at the extreme. It listed on the New York Stock Exchange on March 26, 2024 and hit $105.00 on April 8, 2024 against a reported net asset value of $5.07 a share for the quarter ended March 31, 2024 — a price roughly twenty times the assets behind it, the highest premium the market has produced for a fund of this kind. NAV then rose through 2025, to $6.31, $6.92, $11.37 and $19.97 at December 31, and the premium narrowed as it did.

The figures that follow are a dated observation and not a standing description of this fund. In May 2026 the shares changed hands at $61.66 against the last published NAV of $24.56 — the March 31, 2026 figure, since this fund reports quarterly — a premium of 151%, on a total expense ratio reported at 6.28% for 2024. It did not last. Net asset value reached $34.30 at June 30, 2026 on a portfolio valued near $1.64 billion while the share price fell through the summer: quote pages carried the shares between about $31.50 and $33 in the first days of September 2026, which is at or slightly below that last published NAV. The premium is gone. A fund that traded at roughly twenty times its assets in April 2024, and at 151% above them in May 2026, now trades about where they are marked. Premiums on this fund move violently in both directions, so read every price and NAV in this section with its date attached and look up the current pair before acting on any of them.

The mechanism recurs wherever a listed vehicle holds private assets, because the demand curve can change in an afternoon and neither the share count nor the portfolio can. When retail investors want private exposure and can only get it through one ticker, the ticker reprices rather than the assets. A buyer at the 151% premium of May 2026 needed the portfolio to roughly 2.5× before breaking even on net asset value, while the annual expense eroded that base. That is a wrapper risk, entirely separate from whether the underlying holdings are good.

Interval funds and crossover funds

A second family of wrappers holds private positions inside registered funds with low or no accreditation requirement and periodic rather than daily liquidity. These are the interval and tender-offer funds, which repurchase a set share of the fund at set intervals — Rule 23c-3 requires an interval fund to offer between 5% and 25% of its shares every three, six or twelve months — and the crossover funds, which hold a public portfolio with a private sleeve inside it.

Two published schedules set the range, and the spread between them is the whole decision. The ARK Venture Fund (ARKVX) is a closed-end interval fund investing in public and private innovation companies, with quarterly repurchase offers of at least 5% of shares at net asset value and a gross expense ratio of 4.71% against a net 2.90% after waivers. The Fundrise Innovation Fund charges a flat 1.85% annual management fee with no carried interest and a $10 minimum. One of those is priced like a private fund and one like an index product, for exposure that is qualitatively the same.

What these wrappers solve is access; what they inherit is the redemption problem, which is the subject of our guide to Interval Funds and Non-Traded BDCs and works the same way in every asset class. A fund promising periodic liquidity on assets that have none pays for the promise with a queue. Four things decide whether one is worth owning: the private sleeve’s share of assets, the total expense ratio including acquired fund fees, the repurchase policy and its history of proration, and whether the private positions are held directly or through SPVs whose fees sit below the fund’s own. A fund holding SPV interests charges you its fee on top of theirs, which is the layering of section 6 arriving inside a registered wrapper.

Synthetic exposure

A third route runs through offshore derivative venues: perpetual futures referencing private companies, cash-settled against a synthetic index and offered with leverage. Kraken listed such a contract on SpaceX before that company’s June 2026 listing and added OpenAI and Anthropic pre-IPO perpetuals on September 6, 2026, at up to 5× leverage, priced against a proprietary Kraken PreMarket Synthetic index that is exponentially smoothed and clamped within 0.25% of the mark specifically to suppress liquidations in a thin book.

That is not ownership of anything: no shares, no votes, no claim on the company, and no path to a distribution when one lists. A price deliberately clamped to a quarter of a percent of its own mark is also a price that will not track the asset at the moment tracking matters, which is the moment the asset gaps. Kraken excludes customers in the United States, the EEA, Canada, Australia and New Zealand from these products entirely, and that exclusion is worth reading as information about the product rather than as an obstacle to be routed around.

Four rules decide who may buy private stock, in what quantity and through what vehicle, and each of them shapes the offers you will see.

Accredited investor status

Almost every pre-IPO offering is sold under Regulation D, and to participate you must generally be an accredited investor: individual income above $200,000 (or $300,000 jointly) in each of the two most recent years with a reasonable expectation of the same, or net worth above $1 million excluding the primary residence. The SEC’s amendments of August 26, 2020 added qualification-based routes that do not depend on wealth at all, designating holders in good standing of the Series 7, 65 and 82 licences. The dollar thresholds have not been indexed since 1982, so the accredited population has grown through inflation alone.

Under Rule 506(b) an issuer may not advertise and may include up to 35 non-accredited purchasers, and may rely on a purchaser’s own representation of accredited status absent information to the contrary; under Rule 506(c) it may advertise but must take reasonable steps to verify accreditation rather than accept a self-certification. SEC staff guidance in March 2025 eased that burden where the minimum investment is large enough, but it did not remove it. An offer that arrived through a public advertisement and asked you only to tick a box is wrong.

Resale restrictions

Shares bought privately are restricted securities. Rule 144 conditions their public resale on a holding period — six months for a reporting company, one year where the issuer does not file reports — with additional conditions for affiliates. Section 4(a)(7), added by the FAST Act signed on December 4, 2015, exempts resales to accredited investors subject to conditions: no general solicitation, an operating rather than a shell issuer, and an information requirement where the issuer does not report. It carries no holding period, which is the practical difference from Rule 144. These rules are why the exit from an SPV is usually a sale of the LLC interest rather than of any stock.

Why the minimum is $25,000 and not $2,500

Investment Company Act limits explain the ticket sizes. A vehicle relying on Section 3(c)(1) may have no more than 100 beneficial owners; one relying on Section 3(c)(7) admits only qualified purchasers, generally people with $5 million of investments. An SPV assembling $10 million from at most 100 slots must average $100,000 a slot: the arithmetic of the exemption, not the sponsor’s preference, sets the minimum. Layering has a second motive here, because a feeder can multiply the number of ultimate investors behind a single slot — the outcome the holder limits exist to prevent, and the reason companies write anti-SPV language into their transfer restrictions.

The tax code contains one very large benefit for private-company stock that secondary buyers almost never qualify for, and several administrative burdens they almost always inherit. The federal rules below are current as of September 2026; the structure-specific questions belong with a professional.

Qualified small business stock

Section 1202 allows a non-corporate holder to exclude gain on qualified small business stock held more than five years, subject to a per-issuer cap. It is the most generous provision in the code for equity in young companies, and the condition that matters here is original issuance: the stock must have been acquired directly from the corporation in exchange for money, property or services. Stock bought from another shareholder — which is what every secondary purchase is — does not qualify, however long you hold it.

The provision got more generous in 2025 and none of the generosity reaches you. The One Big Beautiful Bill Act, signed July 4, 2025, improved it for stock issued after that date: a tiered exclusion of 50% at three years, 75% at four and 100% at five; a per-issuer cap raised from $10 million to $15 million, or ten times basis, whichever is greater, indexed for inflation for tax years after 2026; and a corporate gross-assets test raised from $50 million to $75 million, tested at issuance and never retested afterwards. Every one of those improvements runs through original issuance, which is the one condition a secondary buyer cannot satisfy.

The partnership route is closed too. Section 1202(g) passes QSBS benefits through a partnership only where the partnership acquired the stock at original issuance and the taxpayer held the partnership interest on the date the partnership acquired that stock and at all times thereafter until it is sold; the exclusion is further capped at the interest the partner held on that acquisition date, so buying in later or buying more does not enlarge it. A secondary SPV fails at the first test. An offering that mentions QSBS in its marketing is a reason to read the structure more carefully, not less.

What you will actually pay

Absent QSBS, the ordinary rules apply. A position held more than a year produces long-term capital gain taxed at up to 20% federally, plus the 3.8% net investment income tax above $200,000 of modified adjusted gross income for a single filer and $250,000 jointly, thresholds unindexed since they took effect, for a combined top federal rate of 23.8%. A position held a year or less, or income the structure characterises as ordinary, is taxed at up to 37%. State tax sits on top.

Structure-specific complications

Three practical points recur.

• The SPV sends a K-1, and it will be late. A single-asset fund issues a Schedule K-1 rather than a 1099, the character and holding period of the gain are set at the fund level and passed through, and K-1s routinely arrive after the filing deadline, forcing an extension every year you hold the position.
• A forward is not taxed like stock. Your holding period generally does not begin until you own the shares, and the settlement of a prepaid forward can produce a different character of income from a share sale. This is the most common place where an expected 23.8% becomes something worse.
• A loss is only worth what you can use. Capital losses offset gains without limit and ordinary income at $3,000 a year, carrying forward indefinitely; a total loss on a $25,000 interest against no other gains is worth about $1,110 a year at a 37% marginal rate, for eight years.

The largest failure this market has produced took more than $450 million from more than 13,000 customers, and no part of it required a fake company. Every asset class has a distribution of bad outcomes; three of this one’s destroy capital completely rather than merely reducing it, and each of the three is visible in the documents before the money moves.

The largest failure: Linqto

Linqto sold the promise this guide is about, in the plainest possible terms: pre-IPO access to names including SpaceX, tickets as small as $1,000, and advertising that said there were no hidden fees. It suspended redemptions in March 2025 and filed for Chapter 11 on July 8, 2025. Prosecutors allege that more than $450 million was taken from more than 13,000 customers, at a median markup of around 60% and in some cases above 200%; its founder was indicted in September 2026. The markup figures come from reporting on the charging document rather than from a filed schedule, which is the usual state of the evidence in this market.

The mechanism was the ordinary one. The shares sat inside special purpose vehicles, so what a customer bought was a membership interest whose entry price only the sponsor could see, and the difference between what the sponsor paid and what the customer paid was not a line item anywhere. Set that median against the worked example in section 13, which assumes a 10% markup and still leaves the buyer 11.4% under water on day one: a 60% median is six times that assumption, applied to a customer who had been told the fees were nil.

Nothing else about the offer needed to be false for the outcome to follow. The companies were real, the shares in some cases existed, and the marketing was accurate about the asset. What was concealed was the price of the intermediary, which is the number this guide keeps returning to because it is the number that decides the return. A platform that took in more than $450 million from retail buyers on that basis, and reached Chapter 11 inside four months of halting redemptions, is the case to keep in mind when a well-designed website offers a famous private company at a small minimum.

The undisclosed markup, and the enforcement record

On August 14, 2026 the SEC filed a complaint charging Andrew Spaventa and three entities he owned and controlled — The Spaventa Group, TSG Capital Advisors and TSG Alpha Partners — with fraud in connection with unregistered offerings of eleven private funds holding pre-IPO shares. Between December 2020 and June 2025 the defendants raised more than $74 million from more than 800 mostly retail investors, many of them retirees, solicited by more than a hundred cold-calling sales agents. The complaint alleges they told investors they would pay no upfront fees, or at most 12.5%, when the prices investors paid were on average approximately 46% higher than what Spaventa had paid for the same investments.

Four days earlier, on August 10, 2026, the SEC charged the registered-adviser end of the same market: Adit Ventures Management, its chief executive Eric Munson and three affiliated general partners, over pre-IPO holdings including SpaceX and Klarna, alleging misappropriation of client assets, millions in undisclosed fees and more than 150 principal transactions in which the adviser bought shares and then sold them on to its own client funds at a higher price without the consent those transactions require, over conduct running from April 2019 to December 2024. A Fox Rothschild client note read the two actions together as a signal of increased scrutiny of hidden markups, conflicts and unregistered activity in this market.

The SEC’s Office of Investor Education has maintained a standing alert on pre-IPO investment scams for years, listing the same signatures each time: unregistered sellers, aggressive sales practices, social media solicitation, and claims of privileged access to a company everybody has heard of. Only the company names change.

The transfer that is not honoured

A separate failure mode requires no fraud at all: if the company will not recognise the transfer, the SPV or forward you bought references shares that never move. Some of the most sought-after private companies have refused to recognise SPV interests as a matter of policy, and in the run-up to its 2026 listing SpaceX pruned vehicles off its cap table outright, so that some buyers found their “SpaceX exposure” had never been honoured by the company at all — a discovery they made at the point where the position was supposed to become worth something.

The tokenised version of the same problem was on public display in 2025. Robinhood handed European users tokens referencing OpenAI and SpaceX at the end of June, and within days OpenAI stated flatly that it had not partnered with Robinhood, was not involved and did not endorse it, and that the tokens were not OpenAI equity; the Bank of Lithuania opened a review of the product. A wrapper can reference a company without the company’s participation, and the reference is worth exactly what the issuer decides to honour. If the issuer’s approval is not in the file, you are relying on a sponsor’s relationship with a company that has every incentive to say no.

Counterparty failure

In a forward structure the seller keeps legal title until the triggering event, which can be years away. If the seller becomes insolvent, pledges the same shares twice, or simply refuses to deliver, your claim is a contract claim against a private party, enforced by litigation you fund yourself, over an asset whose value you cannot establish without discovery. The registered-fund risk language quoted earlier is not boilerplate; it is a fair description.

The slow risks

Three further risks reduce rather than destroy. Dilution: later primary rounds at lower prices, with preferences senior to whatever you hold, can leave common stock worth a fraction of the headline even when the company succeeds. Preference stacking: in a modest exit the preferred stack is paid first, and common holders — which secondary buyers usually are — can receive very little from a sale the press reports as a success. Time: a fee that runs annually against an asset paying nothing is a slow, certain loss offset only by an uncertain gain.

The arithmetic below is the whole guide in one column of numbers. Every assumption is stated, every assumption is at the friendly end of the published range, and the result is still that the investor keeps under half of what the company earns.

The assumptions

You commit $25,000 to a single-asset SPV holding common stock of a late-stage private company, through a marketplace charging the cheaper of the two published schedules in section 6. You hold four years, the company lists, the SPV distributes shares in kind after the lockup and you sell.

• Platform commission: 2.5% of the commitment, EquityZen’s schedule after the February 2026 reduction; the other published schedule runs to 4.85%.
• Sponsor markup: 10% over the platform’s last observed trade; the SEC’s August 2026 complaint describes a case at roughly 46%, and prosecutors put Linqto’s median at around 60%.
• Management fee: 1.0% a year of committed capital and carried interest of 10% over contributed capital, both at the low end of the 1–2% and 10–20% conventions.
• Underlying return: 12.5% a year, so the shares are worth 60.2% more at exit than at entry.
• Tax: 23.8%, the top long-term federal rate including the net investment income tax, with no QSBS and no state tax counted.

The arithmetic

The commission takes $625, leaving $24,375 to buy the SPV interest. At a 10% markup that $24,375 buys shares worth $22,159 at the last observed trade, so the position is 11.4% under water the moment the wire clears. Four years at 12.5% turns $22,159 of exposure into $35,495; the management fee has taken $250 a year, $1,000 in all; the carry takes 10% of the $11,120 gain over contributed capital, $1,112. Proceeds are $33,383, a gain of $8,383, and the 23.8% rate takes $1,995.

You end with $31,388: 25.6% over four years, or 5.9% a year, on a company that compounded at 12.5%. Owned directly with no markup and no fees, the same $25,000 would have become $40,045 before tax.

The two numbers to remember

The first is the break-even: on these assumptions the company must compound at 4.25% a year for four years simply to return your $25,000, and everything below that is a loss on a rising asset. The second is the hurdle: for the position to pay you 7% a year after tax, the company must compound at 14.1% a year. A pre-IPO investment is not a bet that a company will do well; it is a bet that it will do roughly twice as well as your target, for as long as you are stuck in it.

The bad case

Run the same structure against the 2021 vintage’s experience and the asymmetry is clear. If the company lists 25% below the last observed private trade, your $22,159 of exposure is worth $16,619, the fees have still taken $1,000, and you end with $15,619 — a loss of 37.5% on a company whose value fell 25%. The fee stack does not scale with success; it is charged whether or not there is any.

If, after all of that, the exposure still belongs in your portfolio, the sequence below is the order that reduces the number of ways it can go wrong. It is deliberately slow.

1. Decide the size first, in dollars, and treat it as spent. A pre-IPO position has no reliable exit, no income and no known duration. The right frame is money you have decided to live without for five to seven years.
2. Get your accreditation documentation in order before you look at offers. A verification letter from an accountant, attorney or registered adviser satisfies Rule 506(c) and takes days; doing it under deal pressure is how people accept whatever structure is quickest.
3. Choose the wrapper before the company. A registered fund with a published expense ratio, a marketplace fund at a disclosed commission and a sponsor-organised SPV are three different products with three different risk profiles. The company is the interesting question and the least consequential one.
4. Open accounts on more than one marketplace and watch the same company on each, without trading, until you know how a platform’s offer price compares with its own last print.
5. Demand four documents: the operating agreement; the full fee schedule; the sponsor’s acquisition price and date; and the company’s transfer approval or a written account of how approval will be obtained. Missing any of the four is a decline, not a negotiation.
6. Compute the day-one hole and the break-even yourself, using the arithmetic in section 13, before responding. If the break-even is above what you would underwrite for the same company as a public stock, you have your answer. Read the lockup and distribution mechanics in the same sitting, and tell your accountant about the K-1 before you invest rather than in March.
7. Start with one position, and let it run to a resolution before adding a second. The lessons of this asset class are all in the settlement, not in the entry.

The readings below would change our view, in either direction. Each is specific enough to check and dated so a reader in a later year knows what has moved.

On price and the marks

• The first public price of any large private name against the last secondary mark that preceded it. This is the only real test of whether private marks are conservative or optimistic. Across the three names of the 2021 cohort that reached a public market, the listing price came in below the private mark every time; the SpaceX listing of June 12, 2026 priced well above the December 2025 insider mark, and Stripe’s February 2026 tender at $159 billion cleared its 2021 peak. A sustained run of listings pricing at or above the last secondary print would be genuine evidence that the marks have become honest rather than merely that a single cohort was.
• Our Pre-IPO sub-index against realised exits. It read 187.949 on September 8, 2026, up 87.79% in a year against a whole-composite move of 0.29%. A sleeve that doubles while everything else we measure is flat is either a genuine repricing of the best businesses in the economy or a valuation series with nobody checking it, and the exits are the referee.

On structure and cost

• Published commission schedules after the consolidation. Both large independent marketplaces became subsidiaries of larger brokers between January and March 2026, and one cut fees by half on the way. A return toward 5% a side would say the cut was an acquisition promotion.
• Whether sponsors begin disclosing acquisition cost as standard. The SEC’s August 2026 actions attacked undisclosed markups directly. If offering documents start carrying the sponsor’s cost basis, this market improves materially; if they do not, assume the markup is at the high end of the range.
• Premiums on listed private-asset wrappers. Destiny Tech100 traded at a 151% premium to its last published net asset value in May 2026, on a 6.28% expense ratio for 2024; by early September 2026, with NAV up to $34.30 at June 30 and the shares at about $31.50 to $33, the premium had gone entirely and the fund was trading at or just below its last published marks. Sustained premiums above 100% mean retail demand for private exposure exceeds supply by a wide margin, the condition under which markups everywhere else in the chain get worse.

On the exit channel

• The rate of large listings and acquisitions, the only mechanism by which a private mark becomes cash. Our own store recorded $202 billion across 1,195 US M&A deals in the month dated July 1, 2026 and $87.2 billion of US venture deal value in the quarter dated April 1, 2026, primary capital formation continuing to dwarf the secondary channel outsiders use. A sustained fall in exit activity while marks keep rising is the configuration that produced the 2021 vintage.
• Lockup expiries at listed former private names. The price six months after a listing, against the price on the first day, is the number an SPV holder actually received, and it is rarely the number in the headline. The next test carrying a date is SpaceX, whose 180-day lockup expires on December 8, 2026, after an early release of more than 900 million shares on August 6, 2026 that the market absorbed with a 6% rise. A large expiry absorbed without a fall says the pre-IPO register is not in a hurry to sell; one that gaps down says it was, and that the marks the register was carrying were the optimistic kind.