# A Practical and Economical Bayesian Approach to Gas Price Prediction

arXiv: https://arxiv.org/abs/2305.00337 · HTML: https://mlsanity.com/paper/2305.00337/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 Apr 2023.

## Built on by (most cited first, 1 of 1)

- [Blockchain Transaction Fee Forecasting: A Comparison of Machine Learning Methods](https://mlsanity.com/paper/2305.08105/) (2023)

## Abstract (arXiv)

On the Ethereum network, it is challenging to determine a gas price that ensures a transaction will be included in a block within a user's required timeline without overpaying. One way of addressing this problem is through the use of gas price oracles that utilize historical block data to recommend gas prices. However, when transaction volumes increase rapidly, these oracles often underestimate or overestimate the price. In this paper, we demonstrate how Gaussian process models can predict the distribution of the minimum price in an upcoming block when transaction volumes are increasing. This is effective because these processes account for time correlations between blocks. We performed an empirical analysis using the Gaussian process model on historical block data and compared the performance with GasStation-Express and Geth gas price oracles. The results suggest that when transactions volumes fluctuate greatly, the Gaussian process model offers a better estimation. Further, we demonstrated that GasStation-Express and Geth can be improved upon by using a smaller training sample size which is properly pre-processed. Based on the results of empirical analysis, we recommended a gas price oracle made up of a hybrid model consisting of both the Gaussian process and GasStation-Express. This oracle provides efficiency, accuracy, and better cost.