The DOI matches its Crossref record; 21 of the 21 references it cites were found in a registry or an index; 10 of the 13 sources its claims cite were checked for support.
Level the Level: Balancing Game Levels for Asymmetric Player Archetypes With Reinforcement Learning
Open accessAbstract
Balancing games, especially those with asymmetric multiplayer content, requires significant manual effort and extensive human playtesting during development.For this reason, this work focuses on generating balanced levels tailored to asymmetric player archetypes, where the disparity in abilities is balanced entirely through the level design.For instance, while one archetype may have an advantage over another, both should have an equal chance of winning.We therefore conceptualize game balancing as a procedural content generation problem and build on and extend a recently introduced method that uses reinforcement learning to balance tile-based game levels.We evaluate the method on four different player archetypes and demonstrate its ability to balance a larger proportion of levels compared to two baseline approaches.Furthermore, our results indicate that as the disparity between player archetypes increases, the required number of training steps grows, while the model's accuracy in achieving balance decreases.
Overview
CoChat Summary
TL;DR
This paper extends PCGRL to balance tile-based game levels for asymmetric player archetypes entirely through level design, using reinforcement learning to swap tiles. It reports that PCGRL balances more levels than random search and hill-climbing, but performance drops as archetype disparity grows and the model can exploit unwinnable draws.
Key findings
- PCGRL achieved higher balanced-level proportions than random search and hill-climbing across all tested setups: A vs. B 80.4% vs. 46.1% and 27.6%; A vs. C 56.5% vs. 24.6% and 16.2%; A vs. D1 72.3% vs. 46.6% and 28.8%; A vs. D2 57.9% vs. 35.2% and 26.8%; A vs. A 89.7% vs. 59.6% and 55.8% (Table 1).
- Halving the action space from `[h, w, h, w, 2]` (2592 actions for a 6x6 level) to `[h, w, h, w]` (1296 actions) sped up convergence and allowed higher rewards.
- Greater initial disparity between archetypes required more training steps and reduced the model’s accuracy in achieving balance; symmetric A vs. A initially favored one player in 50.1% of levels, while asymmetric setups strongly favored one player.
- Performance varied by about 20 percentage points across archetypes, and greater initial unfairness made it harder for the model to compensate through level modification alone.
Methods
- Balancing is formulated as a PCGRL Markov decision process on a 6x6 tile-based NMMO survival game; the RL agent swaps tiles, and balance is determined by simulated win frequencies of two heuristic agents, with reward 0.5 for equal win rates and 0 or 1 when one player wins every round.
- Four new archetypes are introduced against base archetype A: B can cross rock tiles, C acts only every second turn, D1 wins with four food resources, and D2 wins with three food resources; A cannot cross rock or water and wins with five food resources.
- Experiments use 500 generated levels, PPO with two 3-layered MLPs of sizes 64, 128, and 64, step size 512, 60 parallel environments, and 30,720 trajectories per update; results are compared against random search and hill-climbing baselines.
Limitations
- Stated: the model can exploit a strategy where neither player can access food, achieving balance because both players never win; the reward function assumes draws are always balanced, including when both players lose, and future work aims to distinguish winnable from unwinnable draws.
- Stated: as the initial disparity between archetypes increases, the required number of training steps grows and the model’s accuracy in achieving balance decreases.
- Evident from the text: evaluation uses heuristic agents and simulated win rates rather than human playtesting in this study, and the experiments are limited to a 6x6 tile-based NMMO environment with four archetypes, so broader generalizability is not established.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures


Verification
Claims the paper makes
8 Valid · 0 Not valid · 1 Not checked · 15 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“This approach is, however, limited by the fact that it only uses exactly the same archetype of agents playing against each other.”
§ Level the Level: Balancing Game Levels for Asymmetric Player Archetypes With Reinforcement Learning > 1 Introduction
What the sources say
- Supportedcites [12] 12. Florian Rupp, Manuel Eberhardinger, and Kai Eckert. 2023. Balancing of com- petitive two-player Game Levels with Reinforcement Learning. In 2023 IEEE Conference on Games (CoG). doi:10.1109/CoG57401.2023.10333248 sourceExperimental — accuracy not yet measured
The Player simulation section says the experiments used the same scripted Forage-agent for both players. The Discussion and Limitations section identifies dependence on those player types as a limitation and treats balancing for different types as future work. (LLM, conf=0.91)
- Supportedcites [13] 13. Florian Rupp, Manuel Eberhardinger, and Kai Eckert. 2024. Simulation-Driven Balancing of Competitive Game Levels With Reinforcement Learning. IEEE Trans. on Games 16, 4 (2024), 903-913. doi:10.1109/TG.2024.3399536 sourceExperimental — accuracy not yet measured
The Player Simulation section says the experiments use the scripted Forage agent for player simulations. Discussion and Limitations (Section VIII) identifies dependence on those player types as a limitation and treats balancing for different types of players as future work. (LLM, conf=0.90)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“An additional study with human play testers concludes that the balancing simulated with artificial heuristic agents actually improves the perceived balance for humans in most cases [16].”
§ Level the Level: Balancing Game Levels for Asymmetric Player Archetypes With Reinforcement Learning > 3.2 PCGRL for Game Level Balancing
What the sources say
- Supportedcites [16] 16. Florian Rupp, Alessandro Puddu, Christian Becker-Asano, and Kai Eckert. 2024. It might be balanced, but is it actually good? An Empirical Evaluation of Game Level Balancing. In 2024 IEEE Conference on Games (CoG). 1-4. doi:10.1109/ CoG60054.2024.10645642 source
The Introduction says the levels were balanced using simulations with heuristic agents. The Conclusions report human playtesting and say the approach positively influenced perceived balance in most cases, while noting that perceptions differed for some aspects and levels. (LLM, conf=0.93)
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Limited work [1,7,15], however, has been published to automatically balance asymmetric games, i.e. players with different abilities and stats.”
§ Level the Level: Balancing Game Levels for Asymmetric Player Archetypes With Reinforcement Learning > 1 Introduction
What the sources say
- Not checked — no full text of the sourcecites [1] 1. Philipp Beau and Sander Bakkes. 2016. Automated game balancing of asymmetric video games. In 2016 IEEE Conference on Computational Intelligence and Games (CIG). 1-8. doi:10.1109/CIG.2016.7860432 source
- Not checked — no full text of the sourcecites [7] 7. Pier Luca Lanzi, Daniele Loiacono, and Riccardo Stucchi. 2014. Evolving maps for match balancing in first person shooters. In 2014 IEEE Conference on Compu- tational Intelligence and Games. doi:10.1109/CIG.2014.6932901 source
- Partly supportedcites [15] 15. Florian Rupp and Kai Eckert. 2024. GEEvo: Game Economy Generation and Balancing with Evolutionary Algorithms. In 2024 IEEE Congress on Evolutionary Computation (CEC). 1-8. doi:10.1109/CEC60901.2024.10612054 sourceExperimental — accuracy not yet measured
The Introduction says little work has addressed automated generation or balancing of game economies, and the Evaluation describes automatically balancing a mage and an archer with different abilities. The paper does not make the broader claim that limited work has been published on automatically balancing asymmetric games generally. (LLM, conf=0.84)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Every source the paper cites for this claim was checked and supports it.
In the paper
“For this work, we use the swap-wide representation to define the action space of the MDP, as it gives the best results in [13].”
§ Level the Level: Balancing Game Levels for Asymmetric Player Archetypes With Reinforcement Learning > 4.1 Improving the Action Space
What the sources say
- Supportedcites [13] 13. Florian Rupp, Manuel Eberhardinger, and Kai Eckert. 2024. Simulation-Driven Balancing of Competitive Game Levels With Reinforcement Learning. IEEE Trans. on Games 16, 4 (2024), 903-913. doi:10.1109/TG.2024.3399536 source
Section VII-B says the authors chose swap-wide because it performed best in their previous results. Table I in Section VI shows it had the highest improvement rate (64.1%), though it tied swap-narrow on the percentage of levels balanced (48.1%). (LLM, conf=0.94)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“When the archetypes setup is symmetric (A vs. A), the proportions of levels that initially favor a single player are almost equal (50.1%).”
§ Level the Level: Balancing Game Levels for Asymmetric Player Archetypes With Reinforcement Learning > 5.1 Performance and Comparison to Baselines
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
10 more claims
Create a free account to see all 24 claims and every source verdict.
- Probably not a substantial claim. 7
- It may state two separate claims as one. 3
Citations
Cites
Proximal Policy Optimization Algorithms
Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017) · arXiv · doi:10.48550/arxiv.1707.06347 · Cited by 10,625 · cited 1×
“All PCGRL approaches use Proximal Policy Optimization (PPO) [18] as RL algorithm.”
PCGRL: Procedural Content Generation via Reinforcement Learning
Khalifa, A., Bontrager, P., Earle, S., & Togelius, J. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment · doi:10.1609/aiide.v16i1.7416 · Cited by 156 · cited 2×
“Recently, the Procedural Content Generation via Reinforcement Learning framework (PCGRL) has shown promising results in creating tile based levels [6].”
Neural MMO: A Massively Multiagent Game Environment for Training and Evaluating Intelligent AgentsPreprint
Suárez, J., Du, Y., Isola, P., & Mordatch, I. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.00784 · Cited by 99
Demonstrating the Feasibility of Automatic Game Balancing
Volz, V., Rudolph, G., & Naujoks, B. (2016) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/2908812.2908913 · Cited by 55 · cited 2×
“Evolutionary algorithms have been used by Volz et al. [20] to balance and build decks for card games, by De Mesentier Silva [9] to balance the meta of a collectible card game, by Morosan et al. [10] for real-time strategy (RTS) games, and for game economies by Rupp et al. [15].”
Dungeons & Replicants: Automated Game Balancing via Deep Player Behavior Modeling
Pfau, J., Liapis, A., Volkmar, G., Yannakakis, G. N., & Malaka, R. (2020) · IEEE Conference on Games (CoG) · doi:10.1109/cog47356.2020.9231958 · Cited by 53 · cited 2×
“For this reason, many works propose approaches to automate the process of game balancing [1, 7, 8, 10, 11, 15, 20].”
What is Game Balancing? - An Examination of Concepts
Becker, A., & Görlich, D. (2020) · ParadigmPlus · doi:10.55969/paradigmplus.v1n1a2 · Cited by 43 · cited 1×
“Game levels for competitive play need to be balanced in order to ensure player satisfaction and to avoid frustration or boredom [2].”
Learning Controllable Content Generators
Earle, S., Edwards, M., Khalifa, A., Bontrager, P., & Togelius, J. (2021) · IEEE Conference on Games (CoG) · doi:10.1109/cog52621.2021.9619159 · Cited by 43 · cited 1×
“Other works extend PCGRL for controllability [3], 3D levels [5], graph data [14], and scalability [4].”
Silva, F. D. M., Canaan, R., Lee, S., et al. (2019) · IEEE Conference on Games (CoG) · doi:10.1109/cig.2019.8847966 · Cited by 31 · cited 1×
“Evolutionary algorithms have been used by Volz et al. [20] to balance and build decks for card games, by De Mesentier Silva [9] to balance the meta of a collectible card game, by Morosan et al. [10] for real-time strategy (RTS) games, and for game economies by Rupp et al. [15].”
Evolving maps for match balancing in first person shooters
Lanzi, P. L., Loiacono, D., & Stucchi, R. (2014) · 2014 IEEE Conference on Computational Intelligence and Games · doi:10.1109/cig.2014.6932901 · Cited by 28 · cited 3×
“Limited work [1, 7, 15], however, has been published to automatically balance asymmetric games, i.e. players with different abilities and stats.”
Learning Controllable 3D Level Generators
Jiang, Z., Earle, S., Green, M. C., & Togelius, J. (2022) · International Conference on the Foundations of Digital Games (FDG) · doi:10.1145/3555858.3563273 · Cited by 27
Research works printed but not linked: 1
- 10.1007/978-3-319-55849-325 · doi:10.1007/978-3-319-55849-325
Cited by
Jang, T.-H., Lee, H.-U., & Kim, H. (2026) · IEEE Access · doi:10.1109/access.2026.3730445
“For example, simulation-driven approaches automatically balance tile-based levels by training agents to optimize win rates, thereby reducing the manual design effort [29], [30], [31], [32].”
Metadata
- Authors
- Rupp, F., & Eckert, K.
- Affiliations
- Technische Hochschule Mannheim (DE)
- Venue
- Proceedings of the 20th International Conference on the Foundations of Digital Games
- Publisher
- ACM
- Published
- 2025-04-15
- Pages
- 1-4
- OpenAlex
- W4410236929
- Cited by
- 1
- References
- 20
- Citations by year
- 2026: 1
- FWCI
- 0.95
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- International Conference on the Foundations of Digital Games (FDG) (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Funders
- This research was supported by the Volkswagen Foundation (Project: Consequences of Artificial Intelligence on Urban Societies, Grant 98555). (98555)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
- Social Sciences › Sociology and Political Science › Digital Games and Media
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Keywords
- procedural content generation, reinforcement learning
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-01, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-01 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Level the Level: Balancing Game Levels for Asymmetric Player Archetypes With Reinforcement Learning. https://app.cochat.ai/papers/10.1145/3723498.3723747
- The paper
- Rupp, F., & Eckert, K. (2025). Level the Level: Balancing Game Levels for Asymmetric Player Archetypes With Reinforcement Learning. Proceedings of the 20th International Conference on the Foundations of Digital Games. https://doi.org/10.1145/3723498.3723747
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.