When data lies: The tape-checking story from a 26-year-old reporter
**Tác giả kiểm chứng băng hình 4 lần, phát hiện lỗi ghi rebound của Zion Williamson từ nguồn chính thức (NCAA 2019). Bài học: dữ liệu không đáng tin nếu chưa tái thẩm. | Key facts:** - Zion Williamson bị ghi thiếu 1 rebound trận Duke vs Virginia Tech (2019) do nguồn cấp dữ liệu nhầm. - Ném phạt NBA giảm 2,8% khi sân không khán giả, EuroLeague hầu như không thay đổi (luận án thạc sĩ 2020). - Ivan Perišić chạy 12,3 km/trận World Cup 2018, nhưng chỉ 31% hướng về khung thành đối phương. - Han Xu bị khai thác 14 lần/trận ở pick-and-roll dẫn đến 9 trận thua của New York Liberty (2023). | Nguồn: Kinh nghiệm cá nhân tác giả (Matthew Chen), xác thực chéo băng hình và dữ liệu Second Spectrum. | Related Q&A: **Câu hỏi liên quan:** - **Q: Làm sao biết số liệu chính thức đúng?** A: Phải tự tua băng kiểm chứng, không copy-paste. - **Q: Tại sao dữ liệu lớn vẫn sai?** A: Vì thuật toán không phát hiện ý đồ chiến thuật như chạy kéo giãn không gian.
I remember that February evening in 2026, sitting in a cramped dorm room in New York, rewatching the Duke vs. Virginia Tech game tape for the third time. On screen, Zion Williamson soared like a whirlwind, grabbing a rebound and slamming it home. I paused, slowed down, then paused again. Four times in a row. The officially published rebound count for Zion – 14 – kept echoing in my head. But the tape told a different story: 15. I counted the tape four times, and the error came from the source, not from me.
The 2026-19 NCAA season was full of superstar narratives: Zion, RJ Barrett, Ja Morant. But few noticed that the official stats system – managed by volunteer scorers courtside – often confused defensive and offensive rebounds, assists and hockey assists. For a 22-year-old sports intern, catching this error felt like uncovering a hidden treasure. My personal blog with 240 reads that day ignited a small debate in the analytics community: we place too much faith in numbers that have never been re-verified.
The core of the issue is not whether the number is right or wrong, but the laziness of the entire news chain.
That Duke-Virginia Tech game ended 74-64, with Zion scoring 22 points and 15 rebounds. But the official box score listed 14, and no major newspaper checked – they simply copy-pasted from the feed. Rebounds that the organizers mis-recorded still count — if you are willing to rewind. This raises the question: are tactical analyses based on flawed data worth reading? Pick-and-roll coverage, points in the paint, free throw efficiency – all start from raw stat sheets. If those sheets are wrong, the entire analysis building collapses.
This is where I see the gap between mass media writers and true analysts. An ordinary writer would take the number 14 and write: “Zion had an impressive rebound rate.” But an obsessive verifier – like me – would stop, count every step, use slow-motion timestamps. I realized that even game organizers make mistakes due to real-time pressure. And if I hadn’t counted again, the world would have believed a number one rebound short – a small deviation, but enough to shift per-game efficiency averages.
When the fans disappear, young free throws disappear with them — unless you are in the EuroLeague. My 2026 master’s thesis compared free throw percentages between the NBA and EuroLeague during the empty-arena period. The NBA dropped 2.8%, EuroLeague nearly held steady. The thesis committee said the sample was too small. But the lesson remained: data is not scripture, it is a puzzle piece that must be placed in context. During the 2026 World Cup match between Croatia and England, I sat counting Ivan Perišić’s 12.3 km run, of which only 31% were towards the opponent’s goal. Croatia is not the team that runs the most — they are the team that runs in the right direction.
If you only look at the total volume without watching the tape, you would believe Perišić was the hardest worker. The truth: he ran off-target 69% of the time.
Contrarian angle: People often believe that big data can fully replace the human eye. Wrong. Systems like Second Spectrum and SportVU provide hundreds of thousands of data points per game, but algorithms still cannot detect the tactical intent of a player running off-ball to stretch a defense. A player sprinting to the corner may never touch the ball, but he creates space for others – something no rebound stat sheet captures. This is the blind spot of our era: we worship numbers but forget that basketball is a sport of space and timing, not just totals.
Take the most recent case: New York Liberty in 2026. A 9-game losing streak began with Han Xu being exploited in pick-and-roll. I cross-referenced Second Spectrum with game tape and identified 14 instances per game. Head coach Sandy Brondello declined interview requests, but three weeks later, the defensive scheme changed. That was no coincidence. That was the result of rewinding tape and proving that the problem wasn’t overall defense, but one player being systematically targeted.
Takeaway: Never trust a single number unless you have seen it born from the tape. Every article, every podcast, every analysis should carry a footnote: “I rewound the tape X times.” Only then will the basketball world become more honest. And my final question to you: When was the last time you rewound the tape to verify a rebound?