Data Never Lies: My 10-Year Journey Learning to Listen to the Numbers in Vietnamese Billiards

I began my career in billiards analysis with a naive belief: data would answer every question. Ten years later, I still hold that belief, but I have learned that asking the right question matters more than getting a quick answer. This article is an honest report on what I have learned from thousands of hours of observation, hundreds of matches, and countless moments of self-review. THE OPENING MOMENT In 2026, at age 17, I first applied the xG model to Vietnamese football. I used Understat data for the match between CLB Hai Phong and Sanna Khanh Hoa in Round 18 of the V.League: Hai Phong generated 2.8 xG, while the opponent had only 1.0. I confidently predicted Hai Phong would win 3-1. The match ended 0-1, and Sanna Khanh Hoa goalkeeper Tran Buu Ngoc made 7 saves, destroying my entire model. FIRST LESSON: DISTRUST SINGLE DATA POINTS That night, I sat in front of my computer screen, looking back at the data table I had trusted with arrogant confidence. 2.8 xG versus 1.0 – a dominant gap. But I had missed a crucial variable: goalkeeper form. xG does not account for that, especially in matches with a low defensive block, where goalkeepers face more shots from distance rather than clear-cut chances. I began manually recording 20 consecutive matches to cross-reference the metrics. For each match, I recorded shot position, shot angle, the number of defenders between the ball and the goal, and most importantly – the goalkeeper's position at the moment of the shot. Results after 20 matches: the xG model predicted outcomes correctly only 55% of the time. But when I added the "goalkeeper position" variable, accuracy rose to 68%. Data never lies, but I once misheard it. I misheard because I only listened to a single source. From then on, I established a principle: never use a single metric to draw conclusions. Every article of mine begins with a list of "conditions to verify" – at least two data sources before making a judgment. SECOND LESSON: THE CROWD LAUGHED. THE DATA DIDN'T. World Cup 2026. The match Mexico 2-1 Germany in the group stage. I had just turned 18, just entering university for journalism. While the world debated the shock, I sat writing a blog analyzing Mexico's pressing. Germany had 66% possession, completed 613 passes – but Mexico's PPDA was 8.4. That means Germany was only allowed to complete an average of 8.4 passes before being interrupted. I wrote: "Germany will be eliminated soon." The blog was ridiculed. Readers believed ball possession mattered more. Two weeks later, Germany lost 0-2 to South Korea and was eliminated. I received 12 emails from readers admitting I was right. The crowd laughed. The data didn't. A year later, I revisited that article. I cross-checked every prediction against actual results. There were places where I was right, and places where I was wrong. But the most important lesson: publicly reviewing my own work is the only way to improve. From then on, I developed the habit of citing data sources within each paragraph. I use the phrase "the data shows" instead of "certainly." Every article of mine includes footnotes explaining how metrics were calculated, sample sizes, and analysis limitations. THIRD LESSON: WHEN HOME GROUND IS NO LONGER A FORTRESS In 2026, at age 20, during the COVID-19 pandemic. The Bundesliga returned with 81 matches without spectators in the final 9 rounds of the 2026/20 season. I collected all the data: home win rate dropped from 44.7% to 33.3%, average away xG increased from 1.15 to 1.32. I proposed reducing the home advantage coefficient in betting models to 0.18 goals per match. A forum administrator criticized me for the small sample size. I conducted a chi-square test with p = 0.045, published the results with a limitations warning. That model helped me win 62% of Asian handicap bets during that period. When home ground is no longer a fortress, I learned to listen to the empty stands. An empty stadium does not kill football. It only strips away my adjustment layer. I realized that many "truths" in sports are actually products of context. When context changes, those truths collapse. APPLYING TO VIETNAMESE BILLIARDS From these lessons, I began applying this methodology to Vietnamese billiards. Billiards is the perfect sport for data analysis: every shot can be measured, every ball has a defined position, every decision can be evaluated. I started collecting data from domestic tournaments. I recorded: winning percentage when a player breaks first, winning percentage when the opponent makes an error in the deciding frame, safety shot completion rates in difficult positions. After 6 months, I had a substantial database. One surprising finding: Vietnamese players' winning percentage on new tables was 12% lower than on their regular practice tables. This figure is much higher than international players. The reason: Vietnam's training system is often tied to a specific table, and players rarely get opportunities to adapt to diverse table surfaces. I wrote an analysis about this. The article sparked debate. Many said I relied too much on numbers, ignoring psychological factors. But my data came from 200 matches, not from intuition. I continued tracking and updating. FOURTH LESSON: CORRELATION IS NOT CAUSATION In 2026, World Cup Qatar. The match Japan 2-1 Germany. Japan had only 26% possession, created significantly lower xG. But they won. Many analysts concluded: high pressing is the key. I disagreed entirely. I cross-referenced data from 50 matches with under 30% possession at World Cups and Euros. Result: winning rate was only 18%. Japan was the exception, not the rule. The conclusion "high pressing is the key" was based on one match – exactly the kind of single-data-point conclusion I had learned to avoid. Correlation is not causation. Just because Japan pressed high and won does not mean high pressing creates victory. Perhaps Japan won because Germany was weak, or because of luck, or because of other tactics. I apply this principle to billiards. When a player wins 5 consecutive matches, I do not rush to conclude they are in peak form. I examine: who were their opponents? Under what conditions did they win? Were there any lucky factors? Only when I have sufficient data from multiple angles do I make a judgment. CONFRONTING MY OWN LIMITATIONS I am not someone who never errs. I have been wrong many times. In 2026, I predicted a young player would soon enter the world top 10 based on his impressive junior tournament data. He never achieved that. I had overlooked the pressure factor of professional competition – a variable that cannot be measured with numbers. One goalkeeper dropping a ball is a mistake. Three goalkeepers dropping balls is a signal. But one young player failing at the professional level could just be an outlier, not a signal. I learned to distinguish between noise and true signals. I also learned to publish my limitations. In every article, I note: what the sample size is, what the significance level is, what variables I cannot measure. This makes my articles less flashy, but more honest. I DON'T WRITE TO CONVINCE ANYONE The model knew from October. I only had the courage to believe in May. This sentence summarizes how I work. I build models, collect data, test hypotheses. But I do not rush to publish conclusions until I have sufficient evidence. I don't write to convince anyone. I write so that data has a witness. When I publish an analysis, I do not expect readers to agree immediately. I expect them to verify, to question, and if they find errors, to point them out. That is the only way for Vietnam's sports analysis community to progress. THREE THOUSAND MATCHES TAUGHT ME THAT ONE MATCH CAN TEACH MORE THAN ALL OF THEM After 10 years, I have watched over 3,000 matches. But the match that taught me the most remains the 2026 Hai Phong versus Sanna Khanh Hoa match – the first match where I applied xG and failed miserably. That match taught me humility. It taught me that data is a tool, not an idol. It taught me that every number has limitations, and a good analyst is one who knows their own limitations. Vietnamese billiards is developing. Young players are reaching out to the world. But to go far, we need a solid analytical foundation. We need people who know how to ask the right questions, collect data properly, and publish conclusions honestly. I do not know what the future of Vietnamese billiards holds. But I know that if we continue to learn from data, continue to test hypotheses, and continue to remain humble before the complexity of this sport, we will go far. Data never lies. But we need to learn how to listen correctly.