Application of Tullock Contest Success Function to transaction fee optimization in high-throughput networks

Objectives. This paper investigates the feasibility of applying a Contest Success Function (CSF) to optimize priority fee expenditures in blockchain networks utilizing priority fee auctions. We analyze the model's ability to describe the relationship between "effort" (bid amount) and the probability of successful transaction inclusion.

Methods. We conducted an experiment to compare the efficiency of the canonical CSF model (Tullock contest strategy) against a baseline strategy – a simple average of the target percentile derived from historical data. Both strategies utilized context gathered from historical datasets to generate bid proposals for the subsequent block. Performance was evaluated based on two primary metrics: the average effort (mean bid size) and the success rate (the percentage of bids successfully landing within the target percentile). The experiment comprised a total of 632 rounds of bid generation.

Results. The experimental trials yielded performance metrics indicating that the canonical CSF model is well-suited for cost optimization in priority fee auctions. Specifically, strategies with decisiveness parameters  and  demonstrated the most favorable results. Furthermore, a positive correlation was observed between the decisiveness parameter  and the strategy's performance; a decrease in the value of  led to a corresponding decline in the strategy's efficiency for cost optimization.

Conclusion. The experimental results demonstrate the overall effectiveness of the canonical CSF for optimizing priority fee expenditures. This approach is particularly relevant for sectors with rapidly advancing blockchain integration, most notably financial services, investment management, and trading. However, it is worth noting that the proposed methodologies are agnostic to specific implementations or projects; they are applicable to any network that implements a transaction prioritization mechanism via additional fees. There remains significant scope for further research, including the introduction of novel performance metrics, the use of more specialized datasets, and the investigation of different historical context window sizes for the model.

1. Tullock G. Efficient Rent Seeking. New York, Springer, 2001, 408 p.

2. Corchon L. C., Dahm M. Foundations for contest success functions. Economic Theory, 2010, no. 43, pp. 81–98. https://doi.org/10.2139/ssrn.1144105.

3. Hwang S.-H. Contest success functions: Theory and evidence. Working paper. Amherst, MA, University of Massachusetts, 2009, 26 p. https://doi.org/10.7275/1066820.

4. Shumov V. V. A study of contest success function for battles (combats, operations). Control Sciences, 2020, iss. 6, pp. 19–30 (In Russ.). https://doi.org/10.25728/pu.2020.6.3.

5. Zheng X., Wan Z., Lo D., Xie D., Yang X. Why does my transaction fail? A first look at failed transactions on the Solana blockchain. Proceedings of the ACM on Software Engineering, 2025, no. 2, pp. 1489– 1512. https://doi.org/10.1145/3728943.

6. Scaperdas S. Contest success functions. Economic Theory, 1996, no. 7, pp. 283–290. https://doi.org/ 10.1007/BF01213906.