Order Book Features for High-Frequency Price Prediction

Summary

The document suggests three input features for training a binary classifier to predict bid or ask price movement from high-frequency order book and trade data: bid-ask spread, bid-ask volume imbalance, and signed transaction volume. Signed volume is positive for buyer-initiated market orders and negative for seller-initiated market orders.

These features summarize quoted trading cost, the relative displayed depth on each side, and the direction of executed order flow. They provide a starting point for an SVM classification task, but the document does not specify prediction horizons, label construction, sampling frequency, normalization, or validation methods. It offers no empirical performance results, so the features should be treated as candidates to evaluate rather than a proven predictive set.

Key ideas

• Bid-ask spread is a candidate feature for predicting short-term quote movement.
• Bid-ask volume imbalance captures differences in displayed buying and selling depth.
• Signed transaction volume records whether market orders were buyer- or seller-initiated.
• The listed features are suggestions and the document provides no measured prediction results.
• Feature usefulness depends on label design, time horizon, and validation.

Tags

# Feature for Maching Learning(SVM) in High Frequecy Order Book?


# Feature for Maching Learning(SVM) in High Frequecy Order Book?












I am trying to implement machine learning to predict the movement of bid and ask price but is unable to find the proper feature for training set. I am using Support Vector Machine for binary classification.

## Answer by jaamor (score 3, accepted)

https://quant.stackexchange.com/a/16230

Features could include:

- Bid-ask spread

- Bid-ask volume imbalance

- Signed transaction volume

The sign in the Signed transaction volume is positive if the buyer has issued a market order and negative if the seller issued a market order.

A great introductory plain English paper on high frequency trading machine learning applications can be found here.

A good blog post with information relevant to the subject can be found here.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.