PROTO:Grid Trading

Systematic Grid-Based Order Placement

Place buy and sell orders at fixed price intervals to capture profits from price oscillations

Abstract

In this project, we implement a Grid Trading strategy on VN30 Index Futures (VN30F1M). The strategy places a series of buy orders below the current price and sell orders above the current price at fixed intervals. When the price oscillates within a range, the strategy systematically buys low and sells high, capturing small profits from each price movement. This approach works best in sideways or range-bound markets.

Introduction

Grid trading is a structured trading method that creates a "grid" of orders at predetermined price levels. Unlike trend-following strategies, grid trading does not attempt to predict market direction. Instead, it profits from the natural oscillation of prices within a trading range.

The core concept is simple:

• When price drops, buy orders are triggered, accumulating positions
• When price rises, sell orders are triggered, taking profits
• The process repeats as price moves up and down through the grid levels

This strategy is particularly suitable for:

• Range-bound or sideways markets
• Assets with high volatility but no clear trend
• Automated trading systems that can manage multiple orders

Hypothesis

We place buy and sell orders at fixed price intervals (grid levels) around a reference price:

Grid Level Formulas:

• Buy levels: reference_price - i × grid_step for i = 1, 2, ..., num_grids
• Sell levels: reference_price + i × grid_step for i = 1, 2, ..., num_grids

Strategy Parameters:

• grid_step: The price interval between adjacent grid levels
• num_grids: Number of grid levels above and below the reference price
• position_size: Number of contracts to trade at each grid level

The reference price is reset at the start of each trading day to adapt to overnight price changes.

Data

• Data source: Algotrade database
• Data period:
  • In-sample: 2022-01-01 to 2023-01-01
  • Out-of-sample: 2024-01-02 to 2025-04-29
• Instrument: VN30F1M (VN30 Index Futures, Front Month)
• Transaction fee: 0.4 per contract per side

Data Collection

Tick-level price data

• The tick-level matched prices, bid-ask spreads, and daily close prices are collected from Algotrade database
• 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

Recommendation: We recommend using Python 3.12 for this project. While other versions may work, Python 3.12 has been tested and verified to work smoothly with all dependencies.

1. Set up python virtual environment using conda py312:

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

Data Collection

Data can be downloaded directly from the shared Google Drive folder. The data files should be placed in the data folder with the following structure:

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

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.

In-sample Backtesting Result

• The backtesting results are constructed from 2022-01-01 to 2023-01-01.
• Parameters: grid_step=2.0, num_grids=5, position_size=1

| Metric                 | Value                              |
|------------------------|------------------------------------|
| Sharpe Ratio           | 1.0777                             |
| Sortino Ratio          | 1.6530                             |
| Maximum Drawdown (MDD) | -0.7329                            |
| HPR (%)                | 83.39                              |
| Monthly return (%)     | 11.06                              |
| Annual return (%)      | 17.25                              |

• The HPR chart. The chart is located at: result/backtest/hpr.svg

• Drawdown chart. The chart is located at result/backtest/drawdown.svg

• Daily inventory. The chart is located at result/backtest/inventory.svg

Evaluation Metrics

• 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).

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

Parameters to Optimize

• grid_step: Range [1.0, 5.0] - Price interval between grid levels
• num_grids: Range [3, 10] - Number of grid levels above and below reference
• position_size: Range [1, 5] - Number of contracts per grid level

Optimization Configuration

{
    "random_seed": 2025,
    "no_trials": 20,
    "grid_step": [1.0, 5.0],
    "num_grids": [3, 10],
    "position_size": [1, 5]
}

The currently found optimized parameters are:

{
    "grid_step": 4.1,
    "num_grids": 3,
    "position_size": 3
}

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 constructed from 2024-01-02 to 2025-04-29.
• Parameters: grid_step=4.1, num_grids=3, position_size=3

| Metric                 | Value                              |
|------------------------|------------------------------------|
| Sharpe Ratio           | -1.5840                            |
| Sortino Ratio          | -2.0163                            |
| Maximum Drawdown (MDD) | -0.8990                            |
| HPR (%)                | -89.67                             |
| Monthly return (%)     | -11.08                             |
| Annual return (%)      | -80.19                             |

• The HPR chart. The chart is located at result/optimization/hpr.svg.

• Drawdown chart. The chart is located at result/optimization/drawdown.svg.

• Daily inventory. The chart is located at result/optimization/inventory.svg

Discussion

Strategy Performance Analysis

The Grid Trading strategy showed promising in-sample results with a Sharpe ratio of 1.08 and 83% HPR. However, the out-of-sample performance was significantly negative, highlighting key limitations of grid trading:

1. Trending Markets: Grid trading performs poorly in strongly trending markets. The out-of-sample period (2024-2025) experienced significant directional moves, particularly in April 2025, causing severe losses.

2. Position Accumulation: In trending markets, the strategy continuously accumulates positions against the trend, leading to large drawdowns.

3. Range-Bound Assumption: The strategy's core assumption of mean-reverting price behavior does not hold during trending periods.

Potential Improvements

• Add trend detection filters to pause trading during strong trends
• Implement dynamic grid spacing based on volatility
• Add position limits to cap maximum exposure
• Consider asymmetric grids based on market conditions

Project Structure

ProtoGrid/
├── .gitignore
├── README.md
├── requirements.txt
├── config/
│   └── config.py
├── data/
│   ├── is/
│   │   ├── VN30F1M_data.csv
│   │   └── vnindex.csv
│   └── os/
│       ├── VN30F1M_data.csv
│       └── vnindex.csv
├── metrics/
│   └── metric.py
├── parameter/
│   ├── backtesting_parameter.json
│   ├── optimization_parameter.json
│   └── optimized_parameter.json
├── result/
│   ├── backtest/
│   │   ├── hpr.svg
│   │   ├── drawdown.svg
│   │   └── inventory.svg
│   └── optimization/
│       ├── hpr.svg
│       ├── drawdown.svg
│       ├── inventory.svg
│       └── optimization.log.csv
├── backtesting.py
├── data_loader.py
├── evaluation.py
├── optimization.py
└── utils.py

Reference

[1] ALGOTRADE, Algorithmic Trading Theory and Practice - A Practical Guide with Applications on the Vietnamese Stock Market, 1st ed. DIMI BOOK, 2023, Chapter 3. Accessed: Dec. 8, 2025. [Online]. Available: Link