Bespoke analytics engine
The customer is a leading provider of competitive gaming content for the global betting industry, with professional gamers playing titles around the clock, every match live-streamed to betting companies worldwide. The original system relied on a standalone desktop application prone to internet drops, machine crashes, and instability, losing multiple games every week; each loss forcing cancelled markets, voided bets, and returned stakes. Beyond stability issues, traders were tied to on-site machines, manually registering in-game actions by watching live streams, while scheduling hundreds of gamers across shifts was handled through emails and team chats. With retention below 80%, the platform was holding back growth.
The joint Zenitech-customer team rearchitected the core trading platform, reverse-engineering the legacy desktop client’s features, incorporating statistical pricing algorithms, and deploying everything onto a Kubernetes cloud cluster accessible via web portal. Over 20 stateless backend services enabled minimal-downtime deployments and endless scalability, while a fail-safe mechanism let another trader instantly take over if a connection dropped, eliminating the old single point of failure.
The team built a bespoke analytics engine using statistical models like Poisson and Negative Binomial distributions alongside a machine learning model predicting final scores, feeding live pricing updates every few seconds. For one title Zenitech partnered with a third-party pricing provider to launch quickly while working toward bringing pricing in-house. AI-driven video analysis eliminated manual data capture by translating on-screen actions into structured data automatically, and an automated Ops Portal handled scheduling for 1,000–1,500 games every 24 hours, integrating with HR systems and player ratings.
The results were dramatic: game retention jumped from below 80% to approximately 99%, a single trader can now supervise three games simultaneously (up from one), and the platform now supports 14 streams without additional trading staff. Competitive gaming is now one of their most profitable product lines.
Read the full case study below.