For most of the industry's history, an institution that wanted systematic exposure had two routes: build a quant team, or wire capital to a fund and accept the fees, the lock-up, and the loss of custody that came with it. A third route has quietly become a large market of its own: licensing the strategy itself, and running it inside your own accounts.
The scale is no longer a footnote. Bank quantitative investment strategies (QIS), rules-based strategies licensed and delivered to institutional clients, generated roughly $8.5 billion in global bank revenue in 2025, up from about $4 billion in 2019, on a market now exceeding $1 trillion, per BCG Expand and IFRE data. The commercial pattern (license the methodology, keep the assets where they are) was proven decades ago by index providers: MSCI reported a $1.7 billion run-rate as of June 2025 against more than $16.5 trillion in assets benchmarked to its methodologies. Even funds now rent outside brains: Qube Research & Technologies runs 44 external stock-picking teams through separately managed accounts, targeting 100, per Resonanz Capital (2025). Part of the same shift that made systematic execution the institutional default, the question has moved from whether to buy external trading logic to how to contract for it.
A trading strategy license grants an institution the right to use a strategy's output (its signals, its execution logic, or both) inside the institution's own accounts and technology stack. The licensor keeps ownership of the intellectual property. The licensee keeps ownership, custody, and control of the capital.
That distinction separates licensing from the two alternatives it is usually weighed against. In a fund allocation, capital leaves your custody, sits commingled with other investors' money, and comes back as an LP statement: you own units, not positions. In hiring, you own everything, but you carry fixed cost, key-person risk, and a multi-year build before the first live trade. Licensing sits between the two: you rent proven logic, deploy it on your own rails, and can turn it off. The full cost case against building is in our build-vs-buy analysis.
Four delivery models dominate, differing mainly in where the IP boundary sits.
The licensor hands over source code or compiled libraries; the institution runs, hosts, and often modifies the strategy. Maximum transparency, but also the model most licensors resist, because once code leaves the building the IP is effectively unprotectable. Where it exists, expect the highest fees, strict field-of-use limits, and audit rights running in the licensor's favor.
The most common commercial pattern. The institution receives executable signals (entries, exits, sizing) via API, sealed indicator, or managed integration, while the logic stays with the licensor. You see every trade in your own account in real time, but not the source. The trade-off is honest: you verify the strategy by its live behavior, not by reading code, which makes the agreement's verification rights load-bearing.
The bank model: the strategy is expressed as a rules-based index and delivered through a swap or structured note. Operationally simple and scalable (the format behind that $1 trillion market), but it reintroduces bank counterparty exposure and buries costs inside the swap spread rather than a visible license fee.
Trade ideas or full external teams compensated as a share of the P&L they generate: the structure behind alpha-capture platforms and Qube's external SMA pods. Alignment is strong, but so is the licensor's incentive toward risk; profit-share deals demand the tightest risk-limit and measurement clauses of any model.
New buyers negotiate price and skim the rest. Practitioners do the reverse: in a licensing deal, the agreement is the risk architecture.
License scope and exclusivity. What exactly is licensed (one strategy, a family, future versions), on which instruments, for how much capital, and whether the grant is exclusive, exclusive-by-segment, or non-exclusive. Capacity language matters: a non-exclusive license to a capacity-constrained strategy is a dilution risk that never appears on a term sheet.
Fee structure. Three patterns dominate: a flat annual license fee; an AUM-linked fee in basis points on capital deployed (the direct descendant of index licensing, where ETF fees typically run 2-4bps of linked assets); and profit-share. Flat fees favor large deployments, bps favor small ones, and profit-share shifts risk to the licensor. Hybrids (a floor fee plus bps or a performance kicker) are common.
IP protection and sealed-code delivery. How the logic is protected (sealed delivery, no-reverse-engineering covenants, confidentiality surviving termination) and, symmetrically, warranties that the licensor owns the IP and no third party can enjoin your use of it mid-deployment.
Kill-switch and termination. The clause institutions most often under-negotiate. Who can halt the strategy, how fast, and on what triggers (drawdown limits, connectivity failure, regulatory event)? On termination, what happens to open positions: orderly wind-down under the strategy's own exit logic, immediate flattening, or handover to your discretion? Because the strategy runs in your account, you always retain the physical ability to flatten; the agreement should make that right explicit and penalty-free.
Track-record verification rights. The right to see live account statements distinguished clearly from backtests, to have records confirmed by an administrator or auditor, and to receive ongoing live-vs-model performance reporting. If a licensor resists delineating live from simulated results, the negotiation has already told you what you need to know.
Updates and versioning. Strategies decay and get recalibrated. The agreement should state whether updates are included, whether you can pin a version, and how material logic changes are disclosed. You licensed a specific methodology and are entitled to know when it changes.
Whether external alpha works as a commercial model is settled: QIS and index licensing settled it. The question is whether the agreement in front of you protects your capital, your data, and your exit.
Against building: a credible in-house effort of three to five senior quantitative researchers runs roughly $2-5 million per year in compensation alone at 2025-26 market rates, before data and infrastructure. Bloomberg terminals list at $31,980 per seat per year, and Morgan Stanley's rule of thumb puts data spend near $1 million per $1 billion of AUM in year one (vendor-sourced, but directionally consistent). That is the annual burn before the first validated signal, with no certainty one arrives.
Against allocating: even with average management fees compressed from 2.0% toward 1.3-1.5% (Eurekahedge; HedgeCo, Nov 2025), a fund still layers fees on the full allocation, and the capital physically leaves your custody. A license fee (flat or bps) buys systematic exposure while the assets never move.
The most underpriced feature of licensing is structural: the strategy runs in the institution's own account. Every position is visible in real time. Nothing is commingled. No gate, no side pocket, no waiting on an administrator's statement. If the relationship ends, nothing has to be redeemed, because nothing ever left.
Allocators are already voting for this structure through its nearest cousin, the separately managed account. Hedge fund SMA assets grew roughly six-fold, from $66 billion (3.4% of industry AUM) in 2010 to $315 billion (7.1%) in 2024, with investor requests for SMA access up about 50% in 2024 alone, per Dechert (Oct 2025). Hedgeweek's "Separate Ways II" (2026) found 60% of allocators reporting SMA demand up over the past twelve months, and zero reporting a decline. A licensed strategy is the logical end point of that demand curve: SMA-grade custody and transparency, without delegating discretion. We compare the three structures in SMA vs. fund vs. licensed strategy.
Licensing shifts the diligence target from a fund manager to a strategy vendor, but the discipline transfers almost intact: verified live performance clearly delineated from backtests, operational depth behind the signal, clean regulatory posture. Our strategy due-diligence framework covers the full checklist; two compliance signals deserve emphasis here.
First, the SEC's Marketing Rule (206(4)-1) restricts advertising hypothetical performance (backtests, models, targets) to general audiences; a September 2023 sweep fined nine advisers a combined $850,000, with five more charged in April 2024. Second, CFTC Rule 4.41 requires simulated results to carry a prescribed cautionary statement, and the CFTC has charged signal sellers for passing off simulated track records as real. The practical read: a licensor who volunteers compliant disclosure (labeling every simulated figure, offering verification of the live record unprompted) is showing you operational maturity. One who leads with an unlabeled backtest is showing you the opposite.
Licensing fits institutions that want systematic exposure with control: family offices adding a systematic sleeve without standing up a quant desk; RIAs who need client assets to stay at the existing custodian; fund PMs renting a strategy into their book the way Qube rents external teams. It fits poorly where the institution cannot support execution operationally, or genuinely wants discretion delegated. That is what funds and managed accounts are for. For the cohort that wants the strategy without surrendering the assets, licensing is the third door, and the market behind it is already measured in trillions.
Algo Alpha licenses institutional systematic strategies that run in your own accounts. Capital never leaves your custody. Learn about the institutional program.
Structures vary more than headline prices: flat annual license fees, AUM-linked basis-point fees (index licensing's precedent runs 2-4bps of linked assets), profit-share, or hybrids with a floor plus a performance component. The right comparison is against the alternatives: a 3-5 person senior quant team runs roughly $2-5M per year before data and infrastructure, and a fund allocation layers management and incentive fees on the entire allocation while capital leaves your custody.
Usually not. Most institutional licenses use sealed delivery: you receive executable signals or a sealed strategy integration while the logic remains the licensor's IP. Fully delivered code exists but commands the highest fees and the strictest terms. What you should always get instead: real-time visibility of every position in your own account, documented methodology, and contractual verification rights over the live track record.
In a fund, your capital moves to the manager's vehicle, is commingled with other investors, and you own LP units subject to the fund's liquidity terms. Under a license, the strategy runs inside your own account at your own custodian: you see every position in real time, nothing is commingled, and terminating the license requires no redemption because the assets never left.
The agreement should give the licensee explicit verification rights: live account records confirmed by an independent administrator, broker, or auditor, with live performance clearly delineated from backtests. Regulators reinforce the standard: the SEC's Marketing Rule restricts advertising hypothetical performance, and the CFTC has charged signal sellers for presenting simulated results as real, so a licensor who volunteers independent verification is displaying operational maturity, not doing you a favor.
That is defined by the termination and kill-switch clauses, which should be negotiated up front: notice periods, halt triggers, and the treatment of open positions (orderly wind-down under the strategy's exit logic, immediate flattening, or handover to your discretion). Because the strategy runs in your own account, you always retain the physical ability to flatten positions. The agreement should make that right explicit and penalty-free.