Mid Price Prediction

The mid_price_prediction package provides tools for analyzing and predicting mid prices in financial trading using order book data. It includes classes and methods for processing trading data, constructing order books, and predicting price movements.

Installation

To install the mid_price_prediction package, run the following command:

bashCopy code
pip install mid_price_prediction

Modules

OrderBook Class

The OrderBook class in OrderBookClass.py represents and manipulates an order book for trading data. It provides functionalities to update the order book based on transaction data, retrieve top price levels, and process trading data for a given day.

Features

• Update the order book with new transaction data.
• Retrieve top k price levels from the order book.
• Process and output data for trading days.

Usage

pythonCopy code
from mid_price_prediction.OrderBookClass import OrderBook

# Create an instance of OrderBook
order_book = OrderBook()

# Example: Processing a day's trading data
day_files = ['res_20190610.csv', 'res_20190611.csv', 'res_20190612.csv']
for day_file in day_files:
    order_book.create_output_for_dayfile(day_file, 5)

PricePredictionModel Class

The PricePredictionModel class in PredictionModel.py is used for predicting future price movements based on historical order book data.

Features

• Reads and transforms trading data.
• Normalizes features for model training.
• Performs feature engineering.
• Trains a logistic regression model for price prediction.
• Evaluates the model and plots the coefficients.
• Performs feature selection and quantile analysis.

Usage

pythonCopy code
from mid_price_prediction.PredictionModel import PricePredictionModel

# Initialize the model with data files
data_files = ['day1.csv', 'day2.csv', 'day3.csv', 'day4.csv', 'day5.csv']
model = PricePredictionModel(data_files)

# Prepare data and train the model
model.prepare_data()
model.train_model()

# Evaluate the model
model.evaluate_model()  # Pass test data as needed
model.plot_coefficients()

# Perform feature selection
model.perform_feature_selection(3)

# Quantile analysis on timestamp differences
quantile_list = [0.1, 0.2, 0.3, 0.4, 0.5]
model.plot_timestamp_difference_distribution(quantile_list)

Requirements

The mid_price_prediction package requires the following libraries:

• numpy
• pandas
• scikit-learn
• matplotlib
• mlxtend

---