Evaluating Orderbook Imbalance as a Midpoint Forecaster
Open Access
Author:
Gabbireddy, Vineeth
Area of Honors:
Statistics
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
Christoph Hinkelmann, Thesis Supervisor Matthew D Beckman, Thesis Honors Advisor
Keywords:
Orderbook Imbalance HFT Market Making
Abstract:
The proliferation of electronic trading has helped liquidity takers gain better access to liquidity while reducing transaction costs, masking order sizes, and hiding intents. However, market makers still face the challenge of incorporating such masked information (adverse selection effects) into their mid-prices. Dealers could stand to benefit from including such information in making fairer mid-prices from both an economic and liquidity-providing perspective. Current standards, such as the micro-price and volume-adjusted mid-price (VAMP), attempt to incorporate trade and volume data to adjust the mid-price forecast accordingly. Sourcing order book data from Coinbase, we aim to create a new metric that provides a more accurate mid-price estimation based on order book imbalance. This method differs from previous standards by dynamically accounting for the linearity between order book imbalance and forward returns based on the observed data. The model performs well in predictions but requires more computational resources than other existing standards.
