QuantforteLabs

Quantitative research lab

We treat trading as an experimental science.

We build the infrastructure first: the engine, the measurement, and the tests that can prove us wrong. The strategy comes after — and only if it survives.

AI-assisted research Deterministic execution Out-of-sample validated
Building · pre-launch

Who we are

We are a small, technical lab. No gurus, no signal-selling, no promises of returns. We write software, design experiments, and learn continuously — from the literature, from the data, and above all from what breaks — and the market is where we find out whether the experiments were worth anything.

We come from discretionary trading. That is where you learn something no simulation teaches: a trade can work out for the wrong reasons, and your own judgement is the first thing that needs auditing. When that craft meets engineering, statistics and the scientific method, the question stops being “which system do I trade?” and becomes “how do I prove this system works before risking anything?”

Hence our uncomfortable conviction: most of what looks like an edge is a measurement error. So we spend more effort on being able to falsify our own ideas than on having them. It is slower. It is the only way that what survives is real.

How we think

01

Technology first

A deterministic, reproducible, auditable engine is not an implementation detail. It is the precondition for any serious claim about a strategy.

02

Scientific method, not intuition

Hypothesis, rejection criteria written before looking at the data, negative controls, and correction for the number of trials. What any laboratory would demand.

03

Diversification over brilliance

The goal is not the perfect strategy but a set of sound ones, decorrelated enough that the book survives what none of them would survive alone.

04

Survival is the first objective

Controlled growth and risk-adjusted return. A spectacular result gets more scrutiny, not a celebration — it is usually the signature of an error, not of skill.

Where we are going

We are aiming at a real portfolio — consistent, and verifiable by third parties. Not a promise: a track record anyone can audit.

  1. Research and design Months spent studying the method before the first line of code: what can be measured, what cannot, and why almost everything that looks like it works does not.
  2. Infrastructure The engine, the measurement, and the tests that can prove us wrong.
  3. First book A set of decorrelated strategies, validated out-of-sample, executing in demo.
  4. Real capital Live execution at controlled size. The priority is execution correctness, not profit.
  5. Auditable track record Results verifiable by an independent third party, published with their methodology — including what did not work.

AI sits at the centre of our research: it proposes and explores hypotheses at a scale a small team could not reach, which is exactly why we can afford to discard the vast majority. Where it does not sit is in the signal or in execution, because there determinism and auditability are non-negotiable. That boundary is an engineering decision, and it will move when the evidence justifies it: learning has an obvious future here, and it will have to earn it through the same tests as everything else.

Current status