atelier-synth

Overview

Synthetic Data Generation for the atelier-rs engine.

Synthetic Data

Currently, the way to generate data is by starting with some basic definitions according to the templated configuration toml files. For every market, there needs to be defined the following sections:

1. [[experiments]]: information about the experiment of synthetic data generation.
2. [[models]]: One model per exchange since there will be generated one order book per exchange as well.
3. [[exchanges]]: Besides basic info like id, or, name, the sub-table [exchanges.orderbook] will contain the particular elements to generate the progressions of suc order book associated to the given exchange.

Experiments

Simple content like id and n_progressions.

Models

Since each order book is generated using one model, here are the parameters necessary for the model used.

Exchanges

Order book generation base parameters like bid_price and ask_price as the initial prices to start the generation.

Workspace

These are the other published crates members of the workspace:

• atelier-base: Core data structures and I/O tools.
• atelier-dcml: Distributed Convex Machine Learning.
• atelier-generators: Probabilistic generators and events simulation.
• atelier-synth: Synthetic Data Generation for the atelier-rs engine.

Github hosted:

• benches
• examples
• tests

atelier-synth is a member of the atelier-rs workspace