Bitcoin Ordinals: Bitcoin Price and Transaction Fee Rate Predictions

Abstract

This study is the first to investigate the significance of Bitcoin Ordinals-related data (i.e., inscriptions on satoshis) in predicting both Bitcoin's transaction fee rates and its price. Key contributions include:

Dataset Construction: A comprehensive dataset combining Bitcoin chain data, Ordinals index data, and Ordinals market data, as well as a counterpart dataset excluding Ordinals-related data.

Findings: Active Ordinals markets correlate with elevated proportions of Ordinals-related fees and higher average Bitcoin transaction fee rates. It’s posited that Bitcoin protocol upgrades—namely SegWit and Taproot—created opportunities for Ordinals, and growing user interest impacted both blockchain activity and Bitcoin price.

Predictive Modeling: Using MAE, RMSE, and MAPE metrics, the study analyzes prediction models. They employ the TemporalFusionTransformer (TFT) as a baseline and introduce a fine-tuned Chronos model. Results show:

Ordinals-related data significantly enhances prediction accuracy for both fee rate and price.

The Chronos model performs comparably or better than TFT for short-term forecasts, especially in low-noise scenarios, with distinct strengths like fast execution and suitability for cloud deployment.