PerformStars began as a simple, almost stubborn question I asked myself in 2020: can we turn ordinary match recordings into serious, objective sports analytics—fast enough to keep up with real play, and accurate enough to be trusted? From the start, the goal was clear: to treat every shot as a measurable event, capture the geometry and dynamics of the table, and convert the game into data that players, coaches, and organizers can actually use.

This became possible only because several technology waves finally converged. Cloud computing made heavy calculations accessible to small teams; modern microprocessors and GPU-class systems brought the needed throughput; neural networks pushed video recognition to a new level; and global streaming platforms normalized long, high-volume video broadcasts. As more billiard rooms and players share their games online, more of the sport becomes visible—and visibility, in our case, becomes measurable reality.

Billiards is a perfect discipline for this kind of work because it is deeply mathematical. Every shot contains geometry, dynamics, probability, and decision-making under pressure. Where a human eye “just sees” the line, the cut, the pace, and the outcome, our system formalizes it—turning intuition into a repeatable analytical instrument.

Technically, we built the project around real-time computer vision operating on large-scale computation. The system learns to understand the table as a physical object in space, aligns the playing surface, and reconstructs ball movement as a consistent numerical sequence. From there we derive spatio-temporal coordinates and other key parameters that allow us to describe each shot not as a subjective impression, but as a structured dataset.

Of course, the road to reliable recognition was not smooth. Internet videos vary wildly in quality, lighting, camera height, and recording style, and we had to make the model resilient enough to work “in the wild,” not only in laboratory conditions. Step by step, we turned these constraints into engineering requirements—until the system started behaving like a true universal tool rather than a fragile demo.

Once the recognition layer became stable, the next question was the one athletes care about most: how do we describe skill? We worked closely with experienced players and coaches and arrived at a practical framework where each shot contains two big worlds—pocketing and position play. For pocketing, we estimate complexity and outcome statistics with strict, objective criteria; for position play, we measure clear signals such as cue-ball travel distance while preparing the ground for more advanced predictive models based on large samples of elite play.

A major turning point for me personally was using the system in a closed training mode. Over roughly half a year of preparation, I recorded and uploaded every practice session, accumulated statistics, and received feedback that made weaknesses impossible to ignore. The result was tangible: my progress accelerated dramatically, and that experience shaped how we think about the product—not as “analytics for its own sake,” but as a real training accelerator.

We own the full intellectual property behind the system—both the core methods and the product layer—and we are building long-term protection for the technology on the international level, including patent coverage. That matters not as a formality, but as a foundation for scaling: a project like this only becomes valuable when it can be reliably developed, supported, and integrated across countries and partners.

In 2025 we took an important step into the global billiards community by presenting PerformStars at the Guangzhou Billiard Exhibition (GBE 2025), held May 10–12 in Guangzhou, China. We showcased our mobile application to the international market, demonstrated our tracking and analytics approach in practice, and—most importantly—spent time in direct conversations with world top players, industry representatives, and equipment stakeholders. Those discussions helped us validate what matters at the highest level: not “more numbers,” but actionable insights, trust in accuracy, and seamless integration into real competition and training routines.

In 2026 we plan to launch industrial operation for the international market with a subscription model: advanced players will be able to upload practice videos and receive a structured summary, an analytical breakdown, and personalized recommendations for improvement. Alongside the online service, we are preparing the next leap—our own hardware device: a dedicated camera designed specifically for billiards, capable of real-time capture and analysis, with instant shot results displayed on a local screen. This turns analytics from “after the match” into a live training and spectator experience.

After that, the roadmap expands toward full real-time online scoring for pool matches, with match-flow analytics and outcome forecasting based on objective performance data. Next comes adaptation to other billiards disciplines beyond pool, and then a broader transition to other sports where camera-based analytics can redefine training, judging, and broadcasting. The years of accumulated engineering and methodological experience put us in a unique position to do this better than anyone else in the world—and to turn that advantage into a new standard for sports performance analytics.

We are actively looking for partnerships with coaches and sports sections, tournament organizers, billiard rooms, federations, and sponsors who want to help shape the next generation of cue sports—where performance becomes measurable, progress becomes visible, and talent becomes easier to discover.