Hong Kong · Asset Management
Disciplined investing across convertibles and systematic multi-asset strategies
G Capital Management (Asia) Limited is a Hong Kong based investment manager. Our Type 9 (Asset Management) licence application with the SFC is pending and we are not currently licensed.
Who we are
A focused manager, built around a small number of complementary disciplines
G Capital was established to pursue a small number of strategies where we believe rigorous research and structural discipline produce durable risk-adjusted returns: Asian convertible securities, systematic multi-asset investing in United States markets, managed futures, and global credit.
We are a small, specialised team. We do not attempt to be everything to every allocator. Each strategy is designed and intended to be managed by the people who built it, with risk limits set independently of performance incentives.
The systematic strategies run on an internal research platform whose modelling layer is built on transformer architecture. We treat that as infrastructure for research — a way to interrogate more of the data, more systematically — not as a substitute for judgement, and not as a claim about results.
What guides us
Three principles, applied consistently
Research first
Every position begins with primary research and a written thesis. We size positions on conviction and downside analysis, never on index weight.
Structural discipline
Convertible securities carry defined asymmetry; systematic strategies follow rules set in advance. In each case, structure — not short-term market views — shapes the portfolio.
Aligned interests
The investment team intends to invest alongside the firm's investors. Capacity limits and fee structures are designed to protect returns rather than to gather assets.
Our technology
A research platform built on transformer architecture
The systematic side of the firm runs on a single internal research platform. At its core are transformer-based sequence models — the attention-based architecture behind modern language models — applied here to market data rather than to text. They are used to estimate which parts of a long price, volume and fundamental history matter for the next decision, and how assets relate to one another, instead of relying on relationships assumed in advance.
Attention across assets and horizons
Self-attention lets a model weight interactions across instruments, sectors and time horizons that fixed-factor models treat as independent. We use it to learn cross-asset structure directly from data, alongside — not instead of — economically motivated features.
Built to be validated, not to be believed
Research is walk-forward: training, validation and out-of-sample periods are strictly separated, and development is segregated from validation and approval. Deployed models sit in a documented inventory with a named owner and are monitored for performance drift against pre-defined thresholds.
Models inform; the team decides
Model output is an input to a decision, never the decision itself. Signals are reviewed against their economic rationale by the investment team, which retains discretion and accountability for every position, and executes within documented risk limits. Overrides are permitted and logged.
Model risk
Systematic and machine-learning models, including transformer-based models, may fail to perform as expected or as they did in testing. Their output depends on the data on which they were trained and may reflect biases or patterns that do not persist. Simulated or research-stage results are not investment results. The firm does not use generative artificial intelligence to provide investment advice or recommendations to clients.
Our strategies
Four strategies, one risk framework
Asian Convertible Securities
A fundamental, research-driven approach to convertible securities issued by issuers across Asia. The strategy seeks positions where the embedded optionality is mispriced relative to our assessment of credit and equity volatility.
US Multi-Asset Quantitative
A systematic, rules-based strategy across liquid United States asset classes. The process is model-driven with human oversight on risk, execution and capacity.
CTA — Managed Futures
A systematic trend-following strategy across global futures markets. The strategy seeks to capture medium-term trends in equity index, interest rate, currency and commodity markets, sizing positions so that risk rather than notional determines exposure.
Global Credit
A fundamental relative-value strategy across global corporate credit, spanning investment grade and high yield. The strategy seeks pricing dislocations identified through issuer research, capital structure comparison and relative value along the curve.