AI vs. the Algorithm: Measuring Success on Twitch
Author:
- 26 August 2025
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Abstract
Games livestreaming has become an invaluable tool for game studios, supporting game discoverability, community building, and community management. Understanding the different forms of success on livestreaming platforms such as Twitch, along with the relevant metrics and target benchmarks, is crucial for maximising engagement. Similarly, gaining insight into how short-term success influences long-term performance can empower studios to strategically plan, design, and implement future content or game releases. However, the diverse ways in which games perform on Twitch present challenges for detailed analysis. To address this, this paper applies unsupervised machine learning techniques to identify and present seven archetypes of success, enabling studios to gain deeper insights into their game's performance.Keywords
This publication has 18 references indexed in Scilit:
- Power to the Spectator: Towards an Enhanced Video Game Stream Discovery Experience2023
- Anomaly Detection in Player Performances in Multiplayer Online Battle Arena Games2022
- Interpretable Real-Time Win Prediction for Honor of Kings—A Popular Mobile MOBA EsportIEEE Transactions on Games, 2022
- Archetypal Analysis Based Anomaly Detection for Improved Storytelling in Multiplayer Online Battle Arena Games2021
- DAX: Data-Driven Audience Experiences in Esports2020
- Win Prediction in Multiplayer Esports: Live Professional Match PredictionIEEE Transactions on Games, 2019
- Design Challenges for Livestreamed Audience Participation Games2018
- Real-Time Dashboards to Support eSports Spectating2018
- Viewers' Perception of Elements Used in Game Live-Streams2018
- Narrative Bytes2018