Anusha Jagtap, Akash Vishwakarma, Abhishek Bidwe, Hitesh Sheregar

DOI: https://doi.org/10.5281/zenodo.24196875

Matchmaking in competitive multiplayer games is driven almost entirely by skill rating. Systems such as Elo, Glicko and TrueSkill estimate individual strength well but ignore role composition, so two teams of equal average rating can be assembled very differently and the match is decided before play begins. Fairness-aware team-formation algorithms reported in the educational and human-resource literature address compo sition, but their combinatorial search cost is unsuitable for live queues that must form teams within milliseconds. This paper proposes the Hybrid Role-Aware Team Balancing Algorithm (HRTBA), a four-stage deterministic pipeline that combines player-declared role queues, skill-sorted role buckets, balanced team assembly, and a bounded same-role skill swap. HRTBA guarantees full role coverage by construction while confining optimisation to role-preserving exchanges, giving an overall cost of O(nlogn)—the same order as the sort a rating-based system already performs. HRTBA is evaluated against Random, Skill only and Greedy baselines on simulated pools of 20 to 500 players with thirty randomised trials per configuration. Results show that HRTBA raises mean role coverage from roughly 68–76% (baselines) to 92–100%, while holding the mean cross-team skill gap within a small margin of the skill-only optimum (7.7 versus 4.9 rating points at pool size 500) and forming all teams in under 1.4 ms even for the largest pool. HRTBA therefore delivers composition-fair matchmaking inside the latency budget of a live queue, without new hardware.

matchmaking, skill rating, role composition, team balancing, fairness, real-time systems, multi-objective opti misation.

This paper presentedHRTBA, a four-stage deterministic pipeline that guarantees role coverage by construction and refines skill balancewith a bounded, role-preserving swap. EvaluatedagainstRandom,Skill-onlyandGreedybaselineson simulatedpoolsof20 to500playersover thirtyrandomised trials, HRTBA raised mean role coverage from roughly 68 76% to 92–100%, held the mean cross-team skill gap within a small margin of the skill-only optimum (7.7 versus 4.9 rating points at pool size 500), and formed all teams in under 1.4 ms for the largest pool. Its O(nlogn) cost matches the sort a rating-based system already performs, so the method is deployable in a live queue without new hardware. Future work includes weighted role preferences for flex players, coupling HRTBA with a Bayesian rating model so that the objective minimises the probability of a lopsided outcome rather than the raw rating difference, a learned role pair synergy term as a secondary objective, generalisation to five-player teams and battle-royale squads, and validation on anonymised live telemetry.

[1] A. E. Elo, The Rating of Chessplayers, Past and Present. New York, NY, USA: Arco, 1978.
[2] M. E. Glickman, “Parameter estimation in large dynamic paired com parison experiments,” Applied Statistics, vol. 48, no. 3, pp. 377–394, 1999.
[3] M. E. Glickman, “Example of the Glicko-2 system,” Boston University, Technical Report, 2012.
[4] R. Herbrich, T. Minka, and T. Graepel, “TrueSkill: A Bayesian skill rating system,” in Adv. Neural Inf. Process. Syst. (NeurIPS), 2006, pp. 569–576.
[5] T. Minka, R. Cleven, and A. Zaytsev, “TrueSkill 2: An improved Bayesian skill rating system,” Microsoft Research, Technical Report, 2018.
[6] R. Coulom, “Whole-history rating: A Bayesian rating system for players of time-varying strength,” in Computers and Games. Springer, 2008, pp. 113–124.
[7] N. Pobiedina, J. Neidhardt, M. d. C. Calatrava Moreno, L. Grad-Gyenge, and H. Werthner, “On successful team formation: Statistical analysis of a multiplayer online game,” in IEEE Conf. Business Informatics (CBI), 2013, pp. 55–62.
[8] P. Yang, B. E. Harrison, and D. L. Roberts, “Identifying patterns in combat that are predictive of success in MOBA games,” in Int. Conf. Foundations of Digital Games (FDG), 2014.
[9] A. Sapienza, A. Bessi, S. Damodaran, P. Shakarian, K. Lerman, and E. Ferrara, “Non-negative tensor factorization for analyzing online team behavior,” IEEE Trans. Comput. Social Syst., vol. 5, no. 3, pp. 719–731, 2018.
[10] Riot Games, “Role selection and matchmaking in VALORANT,” Riot Games, Industry Documentation, 2020.
[11] Blizzard Entertainment, “Overwatch role queue: Developer update,” Blizzard Entertainment, Industry Documentation, 2019.
[12] T. Lappas, K. Liu, and E. Terzi, “Finding a team of experts in social networks,” in ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining (KDD), 2009, pp. 467–476.
[13] A. Anagnostopoulos, L. Becchetti, C. Castillo, A. Gionis, and S. Leonardi, “Online team formation in social networks,” in Int. Conf. World Wide Web (WWW), 2012, pp. 839–848.
[14] M. Kargar and A. An, “Discovering top-k teams of experts with or without a leader in social networks,” in ACM Int. Conf. Information and Knowledge Management (CIKM), 2011, pp. 985–994.
[15] A. Gajewar and D. Sarma, “Multi-skill collaborative teams based on densest subgraphs,” in SIAM Int. Conf. Data Mining (SDM), 2012, pp. 165–176.
[16] I. Srba and M. Bielikova, “Dynamic group formation as an approach to collaborative learning support,” IEEE Trans. Learn. Technol., vol. 8, no. 2, pp. 173–186, 2015.
[17] X. Duan, Z. Yang, and Y. Zhang, “Multi-objective optimisation for balanced team formation,” Applied Soft Computing, vol. 101, p. 107118, 2021.
[18] E. Budish, Y.-K. Che, F. Kojima, and P. Milgrom, “Designing random allocation mechanisms: Theory and applications,” American Economic Review, vol. 103, no. 2, pp. 585–623, 2013.
[19] J. Smed, T. Kaukoranta, and H. Hakonen, “Aspects of networking in multiplayer computer games,” The Electronic Library, vol. 20, no. 2, pp. 87–97, 2002.
[20] M. Claypool and K. Claypool, “Latency and player actions in online games,” Commun. ACM, vol. 49, no. 11, pp. 40–45, 2006.
[21] O. Delalleau, E. Contal, E. Thibodeau-Laufer, R. C. Ferrari, Y. Bengio, and F. Zhang, “Beyond skill rating: Advanced matchmaking in Ghost Recon Online,” IEEE Trans. Comput. Intell. AI Games, vol. 4, no. 3, pp. 167–177, 2012

Anusha Jagtap, Akash Vishwakarma, Abhishek Bidwe, Hitesh Sheregar (2026). Proposing a Hybrid Role-Aware Team Balancing Algorithm for Skill- and Composition-Fair Matchmaking in Competitive Multiplayer Games. International Journal of Computer Techniques, 13(6), 735–741. ISSN: 2394-2231. DOI: https://doi.org/10.5281/zenodo.24196875