PROTO:Market Maker

Place bid-ask base on inventory

Place limit order in two sides base on current inventory quantity

Abstract

In this project, we utilize inventory quantities to simultaneously place both bid and ask orders. The prices of these orders are adjusted dynamically in response to changes in the matched market price. Forced sale scenarios and asset expiration dates are accounted for by incorporating additional fees into the asset's valuation.

Introductin

In market making, one common approach to liquidity provision involves simultaneously placing bid and ask orders based on the current inventory levels held by the market maker. This strategy dynamically adjusts order prices in response to changes in the matched market price, allowing the market maker to maintain balanced exposure while capturing the bid-ask spread. The positions are held overnight.

Hypothesis

We place bid and ask price with our formula:

• $$bid = price - step * (max(inventory, 0) * 0.02 + 1)$$
• $$ask = price - step * (min(inventory, 0) * 0.02 - 1)$$

The step size should exceed the sum of the transaction fee and slippage. Bid and ask prices are updated either every 15 seconds or upon the execution of a position.

Data

• Data source: Algotrade database
• Data period: from 2022-01-01 to 2025-04-29
• Each sell or buy side will be charge 0.4 / 2 fee.

Data collection

Daily closing price data

• The daily close price, bid, ask and tick price are collected from Algotrade database using SQL queries.
• The data is collected using the script data_loader.py
• The data is stored in the data/is/ and data/os/ folders.

Implementation

Environment Setup

1. Set up python virtual environment

python -m venv venv
source venv/bin/activate # for Linux/MacOS
.\venv\Scripts\activate.bat # for Windows command line
.\venv\Scripts\Activate.ps1 # for Windows PowerShell

2. Install the required packages

pip install -r requirements.txt

3. (OPTIONAL) Create .env file in the root directory of the project and fill in the required information. The .env file is used to store environment variables that are used in the project. The following is an example of a .env file:

DB_NAME=<database name>
DB_USER=<database user name>
DB_PASSWORD=<database password>
DB_HOST=<host name or IP address>
DB_PORT=<database port>

Data Collection

Option 1. Download from Google Drive

Data can be download directly from Google Drive. The data files are stored in the data folder with the following folder structure:

data
├── is
│   ├── VN30F1M_data.csv
│   └── VN30F2M_data.csv
└── os
    ├── VN30F1M_data.csv
    └── VN30F2M_data.csv

You should place this folder to the current PYTHONPATH for the following steps.

Option 2. Run codes to collect data

To collect data from database, run this command below in the root directory:

python data_loader.py

The result will be stored in the data/is/ and data/os/

In-sample Backtesting

Specify period and parameters in parameter/backtesting_parameter.json file.

python backtesting.py

The results are stored in the result/backtest/ folder.

Optimization

To run the optimization, execute the command in the root folder:

python optimization.py

The optimization parameter are store in parameter/optimization_parameter.json. After optimizing, the optimized parameters are stored in parameter/optimized_parameter.json.

Out-of-sample Backtesting

To run the out-of-sample backtesting results, execute this command

python evaluation.py

The script will get value from parameter/optimized_parameter.json to execute. The results are stored in the result/optimization folder.

In-sample Backtesting

Running the in-sample backtesting by execute the command:

python backtesting.py

Evaluation Metrics

• Backtesting results are stored in the result/backtest/ folder.
• Used metrics:
  • Sharpe ratio (SR)
  • Sortino ratio (SoR)
  • Maximum drawdown (MDD)
• We use a risk-free rate of 6% per annum, equivalent to approximately 0.023% per day, as a benchmark for evaluating the Sharpe Ratio (SR) and Sortino Ratio (SoR).

Parameters

In-sample Backtesting Result

• The backtesting results are constructuted from 2022-01-01 to 2023-01-01.

| Metric                 | Value                              |
|------------------------|------------------------------------|
| Sharpe Ratio           | 1.5619                             |
| Sortino Ratio          | 2.3335                             |
| Maximum Drawdown (MDD) | -0.1891                            |

• The NAV chart. The chart is located at: result/backtest/nav.png
• Drawdown chart. The chart is located at result/backtest/drawdown.png
• Daily inventory. The chart is located at result/backtest/inventory.png

Optimization

The configuration of optimization is stored in parameter/optimization_parameter.json you can adjust the range of parameters. Random seed is used for reconstructing the optimization process. The optimized parameter is stored in parameter/optimized_parameter.json The optimization process can be reproduced by executing the command:

python optimization.py

The currently found optimized parameters with the seed 2025 are:

{
    "step": 2.940955612440289
}

Out-of-sample Backtesting

• Specify the out-sample period and parameters in parameter/backtesting_parameter.json file.
• The out-sample data is loaded on the previous step. Refer to section Data for more information.
• To evaluate the out-sample data run the command below

python evaluation.py

Out-of-sample Backtesting Result

• The out-sample backtesting results are constructuted from 2024-01-02 to 2025-04-29.

| Metric                 | Value                              |
|------------------------|------------------------------------|
| Sharpe Ratio           | -0.0536                            |
| Sortino Ratio          | -0.0673                            |
| Maximum Drawdown (MDD) | -0.2137                            |

• The NAV chart. The chart is located at result/optimization/nav.png.
• Drawdown chart. The chart is located at result/optimization/drawdown.png.
• Daily inventory. The chart is located at result/optimization/inventory.png

Reference

[1] ALGOTRADE, Algorithmic Trading Theory and Practice - A Practical Guide with Applications on the Vietnamese Stock Market, 1st ed. DIMI BOOK, 2023, pp. 52–53. Accessed: May 12, 2025. [Online]. Available: Link