---

The story of the college athlete is one of sacrifice, skill, and snap judgments. With so much at stake, every small advantage can tilt the game. One place those margins hide is in the data.

As the team statistician for Carolina men’s basketball, Conor Kerr had seen how much numbers could bend outcomes — and wanted to extend that winning edge. In 2022, he approached Mario Giacomazzo, a teaching associate professor in the Department of Statistics and Operational Research, with an idea: teach more students to do this work.

“I told him that if he wanted to make analytics available to other undergraduates, I’d help,” Giacomazzo reflects.

That conversation became a blueprint. Together, they began building Carolina’s first sports analytics hub — one that felt less like a class and more like a clubhouse for problem solvers. Today, the Sports Analytics and Intelligence Lab (SAIL) gives undergraduates the reins, inviting them to try research and leadership at the same time.

What began as one student’s passion project quickly gained notoriety, drawing eager peers from all over the academic landscape. With demand on the rise, they needed a new system.

“Over time, we realized our graduate students were outstanding at making sure things got done, so we made them the authority,” Giacomazzo says.

That shift opened the door for statistics PhD student Kendall Thomas to step in. A former Division I athlete, she arrived at Carolina determined to understand how data on female soccer player performance might contain clues related to injury prevention and ideal workload. Her motivation was personal.

“I got injured pretty quickly in my freshman year and wanted to find a way to impact my team without physically being able to be on the field,” she says. “That’s how I fell into sports analytics.”

Under Thomas’s leadership, SAIL found its stride. As members took on projects — some focused on player health, others on strategy — they began threading analytics through every corner of competition. Since its founding, SAIL’s members have shaped how games unfold, from courts to swimming pools, all in pursuit of one simple goal.

“Hopefully, you’ll see a better game,” says Gordan Tao, a Carolina junior and analyst in the lab.  

Practice makes perfect

Tao doesn’t just watch sports; he deciphers them. As jerseys intertwine and the crowd bursts into cheer, he scans the field for patterns within the motion. A tactful pivot, a calculated sprint — piece by piece, the path to victory is revealed.

“I’ve always liked the idea of noticing the little things, the ones that most people don’t,” he reflects.

Tao, a double-major in computer and data science, knows what he sees is just the tip of the iceberg. Behind every moment in the spotlight lie hundreds more spent training, experimenting, and unlocking the best in each athlete. So, when staff from the UNC-Chapel Hill football program approached SAIL with mountains of performance data, looking for answers, Tao set out to help them.

Staring back at him were thousands of recorded jump heights, accelerations, and impact forces. With the help of fellow student analysts, Tao filtered the stockpile of metrics down to a handful of key performance indicators — the stats most telling of an athlete’s game readiness — effectively creating a scorecard for practice quality.

Athlete fatigue became his number-one opponent. He analyzed the players whose injuries had them benched before the first kickoff, wondering if a more balanced schedule would have given them a better chance to get in on the action.

“If we’re getting the regimen right and athletes’ bodies aren’t too stressed leading up to competition, you may actually get to see some of your stars play,” he explains.

He hopes this research can unpack the stifling connection between stress and success, give coaches new insights into what their athletes are missing, and offer more players the opportunity to show what they’re truly capable of.

“We’re using math to make athletes realize their full potential,” Tao says. “It’s not about making a regimen and hoping it sticks, but deliberately working in small improvements so that they add up to something substantial.”

Predicting passes

Khushi Shah’s love for volleyball began well before she learned how to code. An avid player through high school, she found her new home with Carolina’s beach volleyball club soon after stepping onto campus in 2022. By the end of her first year, what began as a treasured pastime had nudged the determined premed toward an entirely different career path.

As a sophomore, the newly-minted SAIL analyst partnered with the UNC-Chapel Hill volleyball team to give them a deeper look into player trends. At the time, the team’s tracking software could catch moments of contact with the ball and record successful serves, but the competitor within Shah knew the coaches needed more.

Inspecting play-by-play game footage, she extracted everything from player positioning during a set to the mechanics of every effective block. The result was a rich dashboard of visualizations and heat maps, transforming motion into a nuanced narrative.

“We present the data so that coaches can see how their players are doing by the numbers,” she says. “Our goal is to understand their wants and needs and do whatever we can to fill the gaps.”

As Shah’s technical skills grew, so did her ambitions. Today, the senior statistics major is harnessing machine learning to forecast offensive plays before they unfold.

The anatomy of a volleyball play is poignantly simple: Hopping from the passer’s forearms to the setter’s fingertips, the ball is launched into the air as the hitter braces for a strike across the net. With practice, teams tend to fall into subtle patterns — a reliable step, a trusted angle, a comfortable jump height — unwittingly divulging their game plan.

“The more predictable the set, the less likely you are to score because the defense will be better prepared,” she explains.

Shah wanted to make these habits easier to spot. Computer vision — a form of artificial intelligence that allows machines to “watch” and interpret videos with human-like attention — had largely eluded volleyball in its sports analytics takeover, and she saw a perfect opportunity. By extracting visual details from thousands of clips and using them to train her machine learning model, she could compare every attempted play to the calculated average, reminding players of their tendencies.

“I can track the ball’s motion from the pass to the set, then from the set to the hit,” she describes. “If I graph those curves and consider features like speed, height, and starting zone, that could tell me the set’s predictability and its effectiveness as an attack.”

As her project evolves, Shah’s learning that every good research endeavor takes a team.

“It’s very easy to get things done when everyone around you has a passion for sports analytics and a genuine interest in knowing.”

Leveling up lineups

For most of his life, Yunus Mouline believed the magic of sports lived in split-second decisions. As his favorite Tar Heels took the court, he saw a basketball team guided by instinct and experience, prepared to deliver whatever the match demanded.

It was only when he became a Carolina student himself that he realized what felt spontaneous could be wisely cultivated.

“I didn’t realize how much analytics could help a team beyond good coaching intuition or player knowledge,” he recalls. “Having data to support your decisions is an important part of sports.”

Mouline knew, though, that not all data would lead to the best decisions and picking what truly mattered was its own challenge. It all began with a deceptively simple question: At any given moment, which five players would give the team the strongest edge? With every second of the game reduced to shot charts, ratings, and percentages, even the most devoted fans wouldn’t know where to begin.

Rather than guessing which decision felt “right,” the senior statistics and economics major decided to put them to the test. He scraped play-by-play basketball data from across the nation, building a treasure trove of evidence for his predictive model to understand how different player combinations might shift the flow of the game. Once trained, every new matchup becomes a pop quiz — a chance to tell the model where it can improve, sharpening it with each simulation.

“It’s trial and error,” he explains. “You’re running through everything that has happened in the past, and you use those observations to predict what we’ll see next.”

For Mouline, every improvement is a glimpse into the future of basketball — one where live updates could inform lineup decisions in real time. He continues to be amazed by the power of statistics to influence personal experiences and is grateful to be among others who share his excitement.

“At the end of the day, we’re friends who help each other, bouncing ideas even though we all work in different sports,” he says. “It’s been a fun journey, and I’ve loved every step of the way.”