The separately managed account is the hedge fund industry's loudest structural story. Hedge fund SMA assets grew from 3.4% of industry assets under management (roughly $66 billion) in 2010 to 7.1%, or $315 billion, by 2024, a near-sixfold increase; allocator requests for SMA access rose about 50% in 2024 alone (Dechert, October 2025). Hedgeweek's Separate Ways II survey (2026) found 60% of allocators reporting SMA demand up over the trailing twelve months, with SMA capital reaching $42 billion in 2025 against $26 billion in 2023, a 61% jump in two years.
The driver is not fee arbitrage. It is structural: allocators want transparency into positions, control over mandates, and, above all, custody of their own capital. The commingled fund asks for delegation on all three. The SMA hands two of them back. But the debate as usually framed misses a third structure entirely: licensing the strategy itself and running it in your own brokerage account. It is the least-covered option and, for a specific kind of allocator, the most direct. This piece lays out all three (mechanics, trade-offs, and which allocator each one actually fits) as part of the broader question of why institutions use algorithmic trading at all.
The default structure. The allocator subscribes for limited-partnership interests, wires capital to the fund's account, and owns a pro-rata claim on a pooled portfolio the manager runs at full discretion. Everything operational (execution, financing, administration, audit) is the manager's problem.
That is the honest case for the fund: full delegation. It also offers access to capacity-constrained strategies that will not open an SMA for anyone below nine figures, and operational simplicity: one subscription document, one K-1, one line on the portfolio report. For allocators without the staff to monitor positions daily, delegation is a feature, not a bug.
The costs are equally structural. Opacity: the LP sees monthly estimates and a quarterly letter, not positions. Liquidity terms written by the manager: lock-ups, gates, and notice periods that surface precisely when the allocator wants out. The fee stack: average management fees have fallen from the old 2% toward roughly 1.3-1.5%, with performance fees near 15% and about 44% of funds evaluating further cuts (HedgeCo, November 2025), but pressure on fees is not the same as absence of fees, and fund expenses sit on top. And the item most allocators now put first: the capital leaves your custody. Whatever happens at the fund (operational failure, fraud, a side-pocket) happens to your money inside someone else's structure.
In an SMA, the allocator opens an account in its own name (typically at its own prime broker or custodian) and grants the manager trading authority over it. The manager runs the same strategy; the assets never leave the allocator's custodial perimeter.
The appeal is exactly what the fund withholds. Position-level transparency, daily if the allocator wants it. Custom mandates: leverage caps, excluded instruments, risk limits written into the trading agreement. Custody stays home, which is why SMA demand has become a one-way street in allocator surveys.
In Hedgeweek's 2026 Separate Ways II survey, 60% of allocators said SMA demand rose over the past twelve months, and precisely zero reported a decline.
Tellingly, the structure's biggest endorsement comes from inside the industry: multi-manager platforms now use external SMAs to access other managers' strategies. Qube Research runs 44 external stock-picking teams via SMAs, with positions feeding its central risk book and a stated target of 100 pods (Resonanz, 2025), and per Hedgeweek, 9% of managers already allocate to peers this way, with 18% more planning to from 2026. When hedge funds themselves prefer the SMA wrapper, the structural argument is settled.
The trade-offs: minimums are high, because a manager must justify running a dedicated book. The operational burden shifts to you: the account is yours, so financing, reconciliation, and monitoring infrastructure are yours too. And the subtler point allocators sometimes miss: the manager still trades it. You can see everything and terminate quickly, but discretion (and the discretionary fee) remains theirs.
The third structure is the one the fund-versus-SMA literature skips: the allocator licenses the rights to run a systematic strategy (signals or executable logic) in its own brokerage account. No LP interests, no delegated trading authority. The allocator holds the account, receives the strategy's output, and the capital never touches the provider.
What that returns to the allocator is the maximum of every control dimension: full custody, since nothing is wired anywhere; real-time position-level transparency, because the positions are literally in your account; no commingling with any other investor; and termination at will: stop running the strategy and you are out, with no notice period, no gate, no redemption calendar. Economics are set by contract (typically a flat license or a fee structure negotiated directly) rather than a fund's pooled expense stack. (For how these agreements are actually constructed, from flat license versus profit share to IP protections and kill-switch terms, see our piece on how strategy licensing works.)
The honest trade-offs run the other way. You own execution. The infrastructure that turns signals into fills (connectivity, monitoring, fail-safes) is your responsibility or your broker's, not a fund COO's. The diligence burden is entirely yours: there is no administrator or auditor standing between you and the provider's track record, so verification (backtest-versus-live delineation, independent confirmation, operational review) has to be done properly before a dollar moves (our due-diligence framework for external strategies covers the full checklist). And there is no fiduciary manager between you and the market. The license buys you the strategy; the judgment about running it stays in-house.
The fund takes custody and pools it; the SMA and the license both leave assets in your name. The license goes one step further: no third party holds trading authority at all.
Fund: periodic estimates and letters. SMA: full position access, on the reporting lag you negotiate. License: real time, because the book is your book.
Fund: management plus performance plus fund expenses, even in a compressed 1.3-1.5% and ~15% world. SMA: negotiated management/performance terms, often better at size. License: contractual, a defined cost against a defined sleeve, with no pooled expense pass-through.
Funds set minimums in the single-digit millions; SMAs typically demand far more, because dedicated books are expensive to run. Licenses are governed by commercial agreement rather than a manager's capacity math: the constraint is whether the mandate is worth the provider's while, not a fund's operational floor.
Fund: redemption windows, notice periods, gates. SMA: terminate the trading agreement and the positions are already yours. License: switch it off. The right to stop is immediate and unilateral in only one of the three.
This is the mirror image of control. The fund carries everything; the SMA hands you custody-side operations; the license hands you the whole stack: execution, monitoring, and diligence included. Control is not free.
Fund and SMA: the manager trades. License: you (or your broker's infrastructure) do. That single line explains most of each structure's other properties.
Delegation-first allocators: the fund. If the institution wants exposure without building any monitoring or execution capability, and accepts opacity and liquidity terms as the price, the commingled fund remains the rational choice.
Control at scale: the SMA. Allocators writing large tickets, with operational staff and a custodian relationship, get the manager's discretion plus transparency and custody. This is where the industry's marginal dollar is demonstrably going.
Technology-forward allocators with existing brokerage infrastructure: the license. For institutions and family offices that already run brokerage accounts, are comfortable owning execution, and are sizing systematic exposure as a 2-4% sleeve alongside a traditional portfolio, licensing delivers the SMA's custody and transparency benefits without ceding discretion to an external manager, a pattern we examine in how family offices use algorithmic trading. The candid requirement: the diligence and monitoring discipline has to exist in-house, because the structure provides no one to do it for you.
Most sophisticated allocators will end up using more than one of these. The mistake is not choosing the "wrong" structure. It is evaluating only two of the three.
Algo Alpha's institutional model is the third structure: licensed systematic strategies run in the client's own account, with zero custody taken. Details at algoalpha.co/institutional.
In a commingled fund, investors pool capital into a vehicle the manager owns and controls; each investor holds LP interests and sees periodic reporting. In a separately managed account, the account is opened in the investor's own name at its own custodian, and the manager is granted trading authority over it, so the investor keeps custody and position-level transparency while the manager still executes the strategy.
Substantially more than fund minimums, because the manager must run and reconcile a dedicated book. Where funds commonly accept single-digit-million subscriptions, SMA mandates are typically reserved for large tickets, which is why the structure has historically been dominated by institutions and platforms rather than smaller allocators.
A contractual arrangement in which an allocator licenses the rights to a systematic strategy (its signals or executable logic) and runs it in its own brokerage account. No capital is transferred to the provider and no trading authority is delegated: the allocator keeps custody, sees every position in real time, and can terminate at any time. The trade-off is that execution infrastructure, monitoring, and due diligence become the allocator's responsibility.
The SMA and the licensed strategy both keep assets in the allocator's own custodial account. The commingled fund does not: capital is wired to the fund vehicle. The licensed strategy is the strictest form: not only do assets stay home, but no external party holds discretionary trading authority over them.
Yes, and sophisticated allocators frequently do: for example, fund positions for capacity-constrained strategies, SMAs for large core systematic mandates, and a licensed strategy for a small self-directed sleeve. The structures are access mechanisms, not ideologies; the portfolio question is which control-versus-delegation trade-off each allocation warrants.