Skip to content
Eon-Labs KB

Nautilus Project Discussion and Next Steps

Source: Notion | Last edited: 2026-04-01 | ID: 4abd84d4-8d9...

Hi Chris,

Thank you for your prompt response and clarifications. We’re happy to see that you’re excited to work with us, and your availability starting May 7th seems reasonable. To address your questions:

  1. For the ML model, we understand that it will be for demo purposes and will serve as a stub to show where one can be inserted into the strategy.
  2. As for whether we’d like this project to be open source or not, we’d like to ensure we’re on the same page regarding the deliverables first. We can discuss this aspect further once we have a clearer understanding of the project.
  3. Concerning the delivery formats, we’d like it to be python modules / scripts other than notebooks. Before we proceed, we’d like to set up a conference meeting to discuss the project further and ensure we’re on the same page regarding the deliverables. Charlie and I are available between 9:00 AM to 11:59 AM (China time). Please let us know your availability, and we’ll set up the call accordingly.

For our first meeting, we have two primary objectives:

  1. We would like to verify that Nautilus suits our use cases. We’d like you to demonstrate Nautilus, focusing on three performance aspects: a. The trading API, with particular attention to order latency, execution latency, and market data latency, to ensure the framework’s trading performance. b. The backtest, where we want to ensure its ability to handle large data that can’t fit in memory (reading data chunk by chunk), the completeness of the simulation exchange (including latency model, Queue model, and market impact simulation etc.), and the backtest’s performance, including speed and API completeness. c. Performance analysis tools
  2. Below is the requirements list. We are happy to answer any questions you may have about it. You can direct your questions to Charlie or me during the meeting. Project Name: End-to-end example from order book data ingesting to backtest analysis to performance analysis

Requirements:

  1. Starting with a single market (we will supply Binance BTCUSDT M Futures L2 Orderbook & trades data).
  2. Reading & writing data in chunks.
  3. OpenAI Gym-based ML model (no ML coding needed; we will supply a toy example and answer any questions).
  4. Loading data in chunks.
  5. Market-making strategy (OpenAI Gym-based, with collaboration to define the reward function): a simple pure market-making demo to demonstrate Nautilus API.
  6. Backtest with market price impact, partially filled orders, order latency model, etc., functions included.
  7. Performance Analysis (Visualization, Attribution etc.) For your reference, here’s a sample Gym Env:

SwitcherEnv/SwitcherEnv.py at main · spacegoing/SwitcherEnv

We will provide the necessary data for the project. Here are the links for DEPTH and SNAPSHOT data:

https://eonlabs-orderbook-data.s3.ap-northeast-1.amazonaws.com/BTCUSDT_T_DEPTH_2022_11/BTCUSDT_T_DEPTH_2022-11-01.tar.gz

For Trades data, you can find it on the Binance website:

https://data.binance.vision/?prefix=data/futures/um/daily/trades/BTCUSDT/

Additionally, here’s the Binance Perpetual Contract Trade Data (Tick-by-Tick) Download Python script for your reference:

https://eonlabs.notion.site/Binance-Perpetual-Contract-Trade-Data-Tick-by-Tick-Download-6e1779adca0c404aae0d56fc19c0261f

Thank you, and we look forward to our upcoming discussions and potential collaboration.

Regards,

Terry