InvaQuant

One portfolio. Every market. Every position explained.

Quantitative research for portfolios that span every market.

Contact

Animated diagram: signals from technical, fundamental, macroeconomic and sentiment evidence stream toward an out-of-sample test. Some are refuted and drop out. The rest converge into one portfolio.

Approach

A portfolio is one object. InvaQuant researches it as one.

Breadth

Markets are usually studied one at a time. InvaQuant’s research is designed to hold them together: across asset classes, across countries, and across kinds of evidence. Technical, fundamental, macroeconomic, sentiment.

The combination is the point. Any single analysis is a means.

Explainability

No black boxes. A position is considered only if its mechanism can be named, or if a model whose calibration has been measured can state its confidence.

Unusual positions are welcome. Unexplainable ones are not.

Evidence over conviction

A hypothesis is written down before it is measured. It is tested out of sample and corrected for the number of tests that were run.

A refutation is a result, and it is kept.

Built for the downside

A rising market is the easy case. The research is judged on how a portfolio reacts when markets fall.

The standard: what is gained on the way up should be kept when markets turn.

Breadth

Markets move together. Most of all when they fall.

Asset classes are usually studied apart. In calm markets that works well enough: each moves with some independence.

When markets fall, they tend to move together, and diversification that looked solid on paper thins out. That is why InvaQuant studies the links between the parts of a portfolio as closely as the parts.

Schematic. Line weight stands for how closely two asset classes move together. It is not market data.

Method

The test is fixed before the result is known.

Every research question follows the same sequence. The order is the discipline: nothing is measured until the way it will be judged has been written down. An idea that does not survive it is refuted, and the refutation is kept.

  1. Hypothesis

    A claim about markets, stated precisely enough that it can fail.

  2. Protocol

    Universe, period, test and refutation criterion are registered before anything is measured.

  3. Out-of-sample test

    The measurement runs as registered, on data the hypothesis was not built on.

  4. Evidence audit

    An independent check for look-ahead, survivorship, multiple testing and costs.

  5. Kept or refuted

    Both outcomes are recorded. A refuted hypothesis stays on file.

    • Kept
    • Refuted

The InvaQuant Brain

Intelligence you can audit.

AI is useful in portfolio research for one reason: it can read more than any person can. It stops being useful the moment nobody can say where an answer came from. The InvaQuant Brain is built to keep the first and rule out the second.

InvaQuant

A written memory

What InvaQuant knows about markets, asset classes and macro conditions, and every hypothesis, protocol, measurement and refutation, is condensed into linked notes. The Brain is designed so that each note can be traced to its source and carries its version history.

Nothing enters unverified

That is the rule the Brain is built around. AI agents read and distil; independent checks verify before a note is admitted. By design, sources carry reliability tiers, a weaker source never overrides a stronger one, and contradictions are shown, not hidden.

  1. Ingest
  2. Verify
  3. Knowledge

It informs. It never decides.

The Brain supplies information. Judgement sits a level above: in pre-registered research, calibrated confidence and human approval. That is what turns AI into leverage and keeps it from becoming a black box.

Every answer can be traced back

The Brain is designed so that a question runs top-down, and the path it took stays visible.

  1. Principle
  2. Research note
  3. Evidence
  4. Source

It remembers why

Every study, kept or refuted, enriches the next one. A refutation is knowledge too.

From reading to a portfolio

  1. Breadth of evidence
  2. Verified knowledge
  3. Calibrated confidence
  4. One explainable portfolio

Principles

  • Every position must be explainable.
  • A refutation is a result.
  • Judged on the downside.
  • Breadth, with discipline.
  • Confidence is measured, not asserted.

About

Founded in Switzerland.

InvaQuant is an independent quantitative research house. Its research is proprietary.

This website describes how InvaQuant thinks about portfolios. It does not offer a service.

Contact

For research collaboration, partnerships and conversations about InvaQuant.

Jakob Schäfer Founder, InvaQuant [email protected]