Qraft Technologies: Initiation of Coverage

GAII Intelligence defines the Quant, Portfolio & Analytics market as software and model services that apply machine learning to the quantitative core of investment management: generating return and risk forecasts, constructing and rebalancing portfolios, measuring and attributing performance, and estimating and minimising the cost of trading. Buyers are portfolio managers, quantitative researchers, heads of trading and the chief investment officers and technology teams who sign off on systematic processes.

Capabilities we treat as mandatory

• Point-in-time and backtest integrity: model inputs and simulations free of look-ahead and survivorship bias, with a documented method a quant team can reproduce.
• Out-of-sample and live evidence: performance or cost-reduction claims supported by live or held-out results, not only in-sample backtests.
• Institutional data handling: a stated position on where client holdings and order data go, how long they are held, and whether they are used for model training.

Capabilities we treat as differentiating

• Explainability of signals: forecasts, weights and cost estimates traceable to their drivers in a form a portfolio manager or risk committee can challenge.
• Independent, verifiable track record: live results published through regulated funds, third-party ratings or client-run A/B tests over several years.
• OMS/EMS and data-platform integration: native delivery into order and execution management systems, FIX, and the firm's data platform and Python stack.
• AI-assistant access: models and analytics reachable from the assistants portfolio managers and traders already use, through APIs or MCP, without loss of rigour.

Qraft competes with quantitative model and signal providers selling to asset managers, with AI portfolio platforms such as Arabesque's AutoCIO, and in Korean execution with brokers' in-house and vendor algorithms.

The following are the capabilities and characteristics that, in our assessment, most distinguish Qraft Technologies within the Quant, Portfolio & Analytics market. Each is drawn from publicly documented product capability, regulatory filings, published interviews with management, or independent review.

• Live public record: AI-selected US ETFs since 2019, with performance disclosed in SEC filings.
• Execution franchise: AXE adopted by half of Korea's securities firms (company), covering equities and futures, and Taiwan since 2024.
• Model depth: 200+ prediction models and a research-heavy team (company).
• Institutional partners: LG AI Research, KB Securities, Korea Investment Management, Hana Life, FSITC and RHB.
• Strategic backing: SoftBank's $146m investment in 2022.
• Revenue growth: Revenue up about 58% in 2025, from a small base.

The following are factors that prospective clients should weigh, and areas where our public review did not find sufficient evidence to reach a settled view. Several are common across vendors at this stage of the market and are not specific to Qraft. We encourage prospective clients to raise each of them directly with the company; we have done the same, and will update this note as answers arrive.

• Track record: AMOM trailed the S&P 500 over one year (7.94% vs 17.88%), five years and since inception to December 2025.
• Product shrinkage: QRFT liquidated in July 2026; LQAI holds about $2m; AMOM now sponsored as a Defiance fund.
• Economics: Operating loss of KRW 12.2bn on revenue of KRW 4.65bn in 2025.
• Assurance: No SOC 2, ISO 27001 or data-handling terms located.
• Methodology: Point-in-time handling and model methods not documented publicly.
• Capital: No priced round disclosed since January 2022.
• Scale: About 50 staff across three offices.

Qraft Technologies was founded in Seoul in 2016 by Kim Hyung-sik (Marcus Kim), a Seoul National University electrical engineering graduate who built a quantitative trading program during graduate study. It has offices in Seoul, Hong Kong and Tokyo, and runs an AI hedge fund through Qraft Capital (BVI) Limited.

SoftBank Group invested about $146m (KRW 170bn) in January 2022, as its sole investor in that round; the valuation was not disclosed. 2025 revenue was KRW 4.65bn and the operating loss KRW 12.2bn. The founder holds 62% of common shares.

Company-reported figures are shown as such and are not independently verified. Verification is carried into the vendor questionnaire.

Qraft sells models rather than a platform: selection, allocation and execution engines delivered through partners' products. Figure 2 sets out our assessment of how it maps onto the quant and portfolio workflow.

AI quantitative investment solutions

Security-selection and asset-allocation engines behind ETFs, model portfolios, insurance and wrap products.

Market intelligence and execution

AXE execution engine and the MONA platform: VWAP, POV and other AI algorithms, monitoring and TCA reporting.

LLM modules

Automated reports, content feeds, translation and partner chatbots such as Korea Investment Management's ACE AI Chatbot (2025).

AI hedge fund

Qraft Soliton Master Fund with OPIM (September 2025).

Scope of the product today

We did not locate a client-facing portfolio platform, holdings ingestion, risk or attribution tools.

Qraft builds its own deep-learning and reinforcement-learning models for stock selection, allocation and execution, and LLM modules for reports. Management has described more than 1,000 models, with short-horizon models run in-house because of limited capacity.

The live record is public through US ETFs; methodology is not.

Qraft uses market and fundamental data, a data partnership with Benzinga, and execution data from AXE deployments in Korean equities and futures since 2018 and Taiwan equities since 2024.

Point-in-time handling, factor coverage and client-data ingestion are not documented publicly.

Qraft fits asset managers, insurers and brokers that want to license AI selection, allocation or execution models into their own products.

Portfolio managers looking for a tool to use directly will not find one; risk and attribution sit elsewhere.

Qraft delivers models through partners' regulated products: US ETFs run by a registered adviser that licenses its process, and partner funds in Asia. Figure 3 sets out what our public review located.

We would encourage buyers to request security attestations, data-handling terms, model governance documentation and current financial position.

The offsetting factor is regulated distribution: partners' compliance processes have accepted Qraft's models into public funds.

Licence fees are not published. Qraft-powered US ETFs charge 0.75%, which is the funds' fee, not Qraft's.

Buyers should compare licence terms with live, out-of-sample results.

Qraft sits in the Emerging quadrant of the segment matrix, the lowest of the three companies in the segment so far.

GAII Quant, Portfolio & Analytics coverage, 21 October 2026, scored on the segment variant of Data & Content and Workflow Fit; not directly comparable with Research & Intelligence or Risk, Compliance & Surveillance scores.

11.1 Where Qraft is strongest relative to peers

• Against signal vendors, on a multi-year live public record.
• Against execution vendors in Korea, on adoption.
• Against platforms, on willingness to license models into partners' products.

11.2 Areas of narrower functional focus

Qraft has built models and partnerships rather than a platform or enterprise documentation.

• Results: trails its benchmark.
• Scale: small revenue.
• Assurance: not located.
• Methodology: not published.

Scores are on a 0–10 scale. Each dimension score is the mean of its sub-criteria, and the GAII Overall Score is the weighted mean of the six dimensions. The detailed sub-criteria scores that build each dimension are set out in Appendix A.

These are the areas where our public review did not find sufficient evidence to form a settled view. We have put them to the company and will update this note when they are answered. Prospective clients may find them a useful agenda for their own evaluation call.

• Licence terms for AIED and LQAI after the Defiance rebrand.
• Reasons for QRFT liquidation.
• Point-in-time and backtest methodology.
• Live results of the Qraft Soliton fund.
• SOC 2 or ISO 27001.
• Data-handling and no-training terms.
• AXE client count and slippage evidence.
• Pricing.
• Runway and funding plans.
• Headcount trend.
• Explainability method for reports.
• US and Japan expansion status.

This note is an assessment of a vendor against a defined framework, not a recommendation to buy or to decline. The following are the considerations we would put at the centre of an evaluation of Qraft specifically.

If you run portfolios, quant research or trading

• Read the SEC-filed ETF record before any model trial.
• Test AXE where Korean execution matters.
• Ask for out-of-sample results on your universe.

If you are responsible for technology or vendor risk

• Open with Figure 3; request attestations.
• Confirm data residency for licensed models.
• Review model governance.

If you are assessing the category rather than the company

• AI ETFs have a public, mixed record.
• Model licensing is a hard business to scale.
• Execution AI is a more durable niche.

Each dimension score in Section 12 is the mean of the sub-criteria below. Data & Content and Investment Workflow Fit use the Quant, Portfolio & Analytics segment variant.