FlowLab

Welcome to agent-based modelling

FlowLab specialises in agent-based modelling of financial markets.

Inside

Inside the simulator

FlowLab builds a market the way a market is actually built: from the participants upward. Four steps, from the traders to the effect of your own order.

01

A market is its participants

A market is not a curve. It is a population of traders — some providing liquidity, some following price, some acting on a view of value — each with its own objective, horizon and tolerance for loss. FlowLab begins there, with the participants rather than with the price.

02

They meet in an order book

Every participant sends orders to the same place. Orders rest in the book until something crosses them, and matching runs by price and then by time. Nothing about the price is agreed in advance. It is the residue of who was willing to trade, at what level, and when.

03

Price is an outcome, not an input

Because participants react to what the book shows them, the price path emerges from their interaction rather than being drawn beforehand.

04

Your order changes the market

Introduce a real order and the simulated participants respond to it. Depth thins, the price moves against the trade, and other traders adjust to what they observe. This is the part recorded history cannot provide, and the reason a decision can be examined before it is taken.

Foundational principles of FlowLab

01

Calibrated against observed markets

FlowLab is calibrated against order-level exchange data. Historical and simulated markets pass through the same feature extraction process, allowing the simulator to be evaluated using the same measurements as the market it represents.

02

Reproduce every run

Every simulation is generated from a recorded configuration and seed. A result can be reproduced exactly, investigated in detail and generated again under different assumptions.

03

Examine every assumption

Parameters, assumptions and limitations are documented alongside the results. Clients can see what drives an output, which conclusions it supports and where caution is required.

Pre-trade and execution analysis

Products · 01

Execution plans tested in a market that reacts to them

Relevant for pension funds and asset managers reallocating capital or gradually building or unwinding large positions. FlowLab can compare execution plans and show how market impact and liquidity may evolve before a plan is adopted. Because the simulated participants react as the execution unfolds, the plan is tested in a market that changes in response to the trades themselves.

Example output

What a study looks like

In preparation

An example report for this product is being prepared and will appear here shortly, showing the form the output takes, the assumptions recorded alongside it and the conclusions it supports.

In the meantime we are glad to walk through the method and show what a study of this kind produces.

contact@flowlabtechnologies.com

Market scenario generation

Products · 02

A narrative about the future, turned into a market experiment

Relevant for investors, asset managers and risk teams with a view of how future events may unfold, but facing the question of what those events could mean for markets. In FlowLab, that scenario can be expressed as a news flow, whether historical, synthetic or a combination of both, which agents interpret and react to. Their interactions turn a narrative about the future into a market experiment, providing an informed assessment of how prices could develop.

Example output

What a study looks like

In preparation

An example report for this product is being prepared and will appear here shortly, showing the form the output takes, the assumptions recorded alongside it and the conclusions it supports.

In the meantime we are glad to walk through the method and show what a study of this kind produces.

contact@flowlabtechnologies.com

Trading strategy development and testing

Products · 03

A strategy tested against participants that adapt to it

Relevant for trading desks, asset managers and market makers developing or evaluating a strategy. The strategy can operate inside a responsive simulated market across repeated runs, revealing both its performance and how other participants adapt to it. Reinforcement learning models can be trained natively inside FlowLab, while adversarial agents can be used to find where a strategy fails.

Example output

What a study looks like

In preparation

An example report for this product is being prepared and will appear here shortly, showing the form the output takes, the assumptions recorded alongside it and the conclusions it supports.

In the meantime we are glad to walk through the method and show what a study of this kind produces.

contact@flowlabtechnologies.com

Hypothesis testing

Products · 04

Which ideas hold, and under which assumptions

Relevant for quantitative investors and trading teams turning observations about financial markets into new trading ideas. FlowLab makes those ideas testable: whether a proposed relationship appears, which assumptions sustain it and where it breaks down. This helps distinguish hypotheses worth developing from those that only work under narrow conditions.

Example output

What a study looks like

In preparation

An example report for this product is being prepared and will appear here shortly, showing the form the output takes, the assumptions recorded alongside it and the conclusions it supports.

In the meantime we are glad to walk through the method and show what a study of this kind produces.

contact@flowlabtechnologies.com

Overview

Markets built from the behaviour of their participants

01

Pre-trade and execution analysis

Compare execution plans before one is adopted, and watch market impact and liquidity evolve as the trade works, in a market that reacts to it.

Read more
02

Market scenario generation

Express a view of the future as a news flow. Agents interpret it and react, turning a narrative into a market experiment.

Read more
03

Trading strategy development

Run a strategy inside a responsive market across repeated runs. Train models natively, and use adversarial agents to find where it fails.

Read more
04

Hypothesis testing

Test whether a proposed relationship appears, which assumptions sustain it, and where it stops holding.

Read more

Live demos

Coming soon

Interactive demonstrations of the simulator, arriving here shortly.

Descriptions of a market simulator only go so far. This page will let you operate one: set the conditions, place an order into a book that responds to it, and watch what the decision costs as it fills.

The demonstrations run in the browser, with no installation and nothing to request in advance. Anyone assessing whether this work is relevant to their institution can see the mechanics for themselves before speaking to us.

If you would like to be told when these are available, or would prefer to be shown them directly, write to us.

contact@flowlabtechnologies.com

About us

The people you work with

Rasmus S. Mørkøre
Co-Founder and Chief Executive Officer

Rasmus S. Mørkøre

rasmusmorkore@flowlabtechnologies.com
Benjamin Baadsager
Co-Founder

Benjamin Baadsager

benjaminbaadsager@flowlabtechnologies.com
Christian Clasen
Co-Founder

Christian Clasen

christianclasen@flowlabtechnologies.com
Høgni Eldevig
Co-Founder

Høgni Eldevig

hognieldevig@flowlabtechnologies.com

contact@flowlabtechnologies.com