Hyperion: macro investing for the age of AI

Macro investing for the age of artificial intelligence

Hyperion is a systematic hedge fund built as a technology company and designed to compound through structural change while keeping drawdowns shallow. The team has co-founded or backed more than eighty companies, run pension-scale risk, built national-scale data systems and taken three funds from launch to allocator-ready.

Regulated in Jersey and the Cayman Islands, with independent administration, audit, legal counsel, prime brokerage and trading technology

Engineered from first principles

Hyperion is built in three layers, each answering one question and narrowing what the next layer may decide, so human judgement sets the boundaries first and systematic models work inside them.

Philosophy
What belongs in the portfolio?
Hyperion rests on a few deeply researched convictions about the next five to ten years: sovereign debt has outgrown repayment in real terms, populations are ageing across every major economy, artificial intelligence is restructuring how value is created, and governments will keep debasing their currencies to manage all three. Markets reprice forces this slow in bursts, and a portfolio positioned for them in advance stands on the right side of each one. Holding them fixed rules out the most expensive mistake in investing, betting against the forces that dominate a decade. They also decide how the firm is built: a world this volatile needs a regime layer to read its turns, and markets this competitive need a quantitative layer to extract returns from the space the first two layers open. The philosophy layer is the war plan, and the other two execute it.
Regime
What do current conditions allow?
The regime layer is the shield, reading where the economy sits in its cycle from growth, inflation, liquidity, monetary and fiscal policy, positioning and a long tail of smaller indicators, each tracked by its rate of change. Its output is a gate: in favourable states the portfolio carries full risk through the long upswings, and in hostile ones exposure to the affected assets is cut for as long as the state lasts, which makes the gate the architecture's main defence against catastrophic loss. Price action has to confirm each transition before exposure moves, trading a short lag at turning points for protection against whipsaw. The calibration accepts type 1 errors, de-risking into a hostile state that proves brief, to avoid the type 2 error of staying long into one that lasts. The same reading sets which signals the quantitative layer runs at full weight, trend-following in benign states and mean reversion under tighter volatility control in hostile ones, so its models compete for capital only in the conditions they were built for.
Quantitative
How should each position be sized?
The quantitative layer is the sword, acting on the shortest horizon inside whatever the regime layer has opened and turning it into three decisions: whether to hold a position, how large it should be and how fast it may change. Because the layers above have settled what the portfolio is for and when it may take risk, this layer lets Hyperion build toward a firm that puts the newest tools in quantitative finance to work: machine learning ensembles and factor models, bright-spot models that find where AI is improving company economics cohort by cohort, market-neutral books built from the same models, AI trading pods that research and score opportunities across the investable universe and, furthest out, portfolio construction on quantum optimisation. Some of this is running, some is close to production and some is still research, and every piece earns its place the same way, competing for capital as pods do inside a multi-strategy fund, through a single deterministic process that weighs each signal by its realised record and its correlation with the rest. No model or agent holds authority over capital, which is what lets the firm use AI this aggressively. Models will come and go, and the process that weighs them is the part built to last.

People, governance and structure

A team drawn from markets, pension-fund risk, data science, engineering and institutional fund operations, with each person owning a different part of the firm and independent oversight at every level.

Team

The people who built the systems before the first allocation and keep their own capital in the fund.

Thorbjørn Rønje
Co-founder and chief investment officer
- 10+ years in macro and markets
- Profitable, auditable trading record
- Serial tech founder and investor
- All his liquid wealth in the fund
Thorbjørn has spent more than a decade as a macro strategist, trader and investor across public and private markets, and peers know him for being early to the shifts that define each era. He called the subprime collapse in a 2007 thesis, backed Ethereum at pre-seed and has backed more than eighty companies at an early stage, Coinbase and Groq among them. Much of that investing was funded by his own trading, a strong record that is documented and auditable on request. Being part of so many of those builds has given him deep company-building experience, which he has put into Hyperion’s team, brand and institutional setup. With Kristian running the firm as chief executive, Thorbjørn leads the investment side: he owns Hyperion’s investment philosophy and regime framework, holds final authority over its investment architecture and keeps his own capital in the fund.
Kristian Buchardt
Co-founder, CEO and chief quant
- PhD and affiliated professor at the University of Copenhagen
- 10+ years of pension liability modelling ($26–85bn)
- Architect of Hyperion's risk and quantitative core
Actuarial scientist and affiliated professor at the University of Copenhagen, holding an industrial PhD and peer-reviewed research in financial mathematics and biometric risk. Spent more than a decade responsible for valuation and risk modelling at two of Denmark's largest pension funds ($26–85bn liabilities). One of Berlingske's Talent 100: "the master of PFA's intellectual machine room." His work centers on pricing, hedging, and managing risk in incomplete markets: the problem Hyperion exists to solve. Leads the firm and architects its quantitative core.
Chris Valleau
Co-founder and chief technology officer
- 10+ years in predictive modeling and intelligence systems
- Built national forecasting capabilities at the US CDC
- Architect of Hyperion’s autonomous AI research engine
Chris is a data scientist and systems builder with a decade of experience across energy, public health, and open-source intelligence. His work spans predictive modeling, anomaly detection, and building analytical systems to anticipate emerging risks and support decisions under uncertainty. At the US CDC, he helped establish national capabilities for infectious disease forecasting and outbreak analytics. At Hyperion, he designs and builds the firm’s autonomous AI research engine and oversees the development of its signal generation and execution infrastructure.
Ashley Sumner-Pike
Co-founder and chief operating officer
- Goldman Sachs, Credit Suisse, Pictet
- Took three funds to allocator-ready
- Owns regulation and operations
Has taken funds from zero to allocator-ready. Close to two decades building institutional hedge fund operations across Jersey, London, and internationally, with senior roles at Ovata Capital, Schroders, BNP Paribas Securities Services, Pictet Asset Management, Royal London Asset Management, and Credit Suisse, most recently as chief operating officer at Corbiere Capital Management. Depth in over-the-counter derivatives, margin optimisation, cross-jurisdiction compliance, fund formation, and service provider negotiation. Owns Hyperion's regulatory backbone and fund operations.
Tsvetomir Batinov
Lead developer
- Former technical founder
- Built the agent infrastructure
- Owns research through to execution
Full-stack systems engineer and former technical founder. Lead architect of the agent infrastructure, the orchestration layer that coordinates research units, and the pipeline that carries validated output through to execution.
Bozhidar Valkov
Quantitative engineer
- Low-latency systems at one of UK's largest brokerages
- End-to-end ownership from research to live trading
- Data integrations, backtesting, & validation infrastructure
Previously part of the quantitative team at one of the UK's largest brokerage firms, developing low-latency execution systems capable of processing over 100,000 orders per second. Owns the end-to-end technical lifecycle of quantitative strategies, from data-provider integrations and strategy implementation to backtesting, validation, and live trading integration. Alongside his core engineering responsibilities, he also contributes to quantitative research.

Credentials are documented and can be verified on request.

Compounding without the drawdowns that interrupt it

Drawdown control is the binding constraint on every design decision in the strategy. The objective is strong long-run returns, and the architecture is built to keep losses from becoming the kind a portfolio spends years recovering from: a 35% loss needs a 54% gain to recover.

Drawdown control

Regime-conditional exposure
The macro regime decides how much risk the portfolio is permitted to carry. When the environment turns hostile, exposure to the affected asset classes is reduced or removed and capital rotates into the positions that hold value in that state. When the environment is favourable, the portfolio runs at full risk. The system accepts the occasional cost of de-risking into a transient signal to avoid staying long into a regime shift.
Fast shocks and slow declines
Fast market events arrive before any macro classifier can read them, so a hedging overlay holds convex positions that pay in volatility expansions, sized to a small annual budget. Slow declines are the opposite problem, with prices bleeding for weeks while volatility stays calm and option hedges sit idle, and for those a governor scales exposure down on sustained downside variance.
Beyond the backtest
A backtest is a single realised path, the one sequence of returns history happened to produce. The strategy is therefore also run across tens of thousands of simulated environments, with heavy-tailed shocks, correlations that tighten under stress, and realistic regime durations, so that the depth, duration, and frequency of drawdowns are understood before they are experienced.

Oversight and structure

Advisors

Three practitioners we have known for years, whose judgement we rely on and whom we consult often, each on the part of the architecture closest to their own work.

Andreas Steno Larsen
Macro strategy and regime
The youngest chief strategist in Nordea's history and former chief investment officer of a Nordic macro fund. Founder of Steno Research, one of Europe's leading independent macro research houses, and chief macro analyst at Real Vision. Co-founder of Nowcast.IQ, an AI-native macroeconomic analytics platform whose clients include central banks. Nowcast.IQ is owned by Bifrost Studios, Hyperion's parent company, and Hyperion discloses the relationship as a related party. His work on nowcasting and real-time economic measurement sits close to the problem the regime layer solves, and he advises on its structure.
Peter Garnry
Quantitative strategy
Fifteen years at Saxo Bank, latterly as chief investment strategist leading global strategy across macro, equities, bonds, foreign exchange, and commodities, and before that as head of quantitative strategies. Built Saxo's primary equity factor model, covering more than two thousand global stocks across quality, value, momentum, and beta, and created the Stronghold dynamic asset allocation strategy around tail-risk management. Chief executive and co-founder of Gesda Capital. Advises on the quantitative layer.
Daljit Dhaliwal
Trading strategy
Macro and event-driven trader, featured in Jack Schwager's Unknown Market Wizards, with a verified track record running more than nine years. He combines quantitative leading indicators with qualitative narrative analysis, and his depth sits in the questions the regime layer exists to answer: recognising when an environment has changed, sizing into it, and standing down when it has not. He has stress-tested Hyperion's architecture and advises on trading methodology and the risk framework.

Directors

Independent directors sit on the boards of both the manager and the fund, appointed to hold the structure to its mandate.

Firm directors

Fund directors

The fund directors are independent of the investment manager and act on behalf of investors.

Service providers

: Prime broker
: Prime broker
: Fund administrator
: Auditor
: Fund legal counsel
: Order and portfolio management

Structure

Investment manager: Hyperion Quantitative Strategies Limited, Jersey
Master fund: Cayman Islands
Feeder, offshore: Cayman, for non-United States investors and United States tax-exempt investors
Feeder, onshore: Delaware, for United States onshore taxable investors
Regulation: Jersey Financial Services Commission (manager), Cayman Islands Monetary Authority (fund)

Institutional foundation

Hyperion was built to pass operational due diligence on day one: the controls, liquidity and documentation an allocator checks are in place, live and open to inspection.

Manager alignment

Every member of the team invests their own capital in the fund. In first-loss mandates, the manager's capital absorbs losses before any investor's does.

Independent controls

The administrator calculates net asset value independently of the manager, and Deloitte audits the fund. Execution is electronic, every trade is reconciled daily against broker records and risk limits are enforced systematically.

Liquid implementation

The strategy trades liquid listed instruments only, and positions are sized so the whole portfolio could be exited within two trading days without exceeding a tenth of average daily volume.

Governance and documentation

A full policy set covers compliance, risk, valuation, cybersecurity and business continuity. The due diligence questionnaire and operational due diligence pack are complete and sit in the private portal, so an approved allocator can move from decision to funded in three to four weeks.

Access

Hyperion is onboarding a limited number of institutional allocators, strategic limited partners and family offices. Strategy research, performance reporting, due diligence materials and service provider documentation are available to professional investors through the private portal, with access reviewed before it is granted.

Investor login

Engineered from first principles

Philosophy

Regime

Quantitative

People, governance and structure

Team

Thorbjørn Rønje

Kristian Buchardt

Chris Valleau

Ashley Sumner-Pike

Tsvetomir Batinov

Bozhidar Valkov