Post-Match Analysis: Home Advantage Is Gone — And 156 V.League 2026 Matches Proved It
core_answer: Tỷ lệ thắng sân nhà tại V.League 2020 giảm còn 38,1% so với 46,2% giai đoạn 2015–2019. Nguyên nhân chính là đội khách pressing mạnh hơn và chơi táo bạo hơn khi sân không khán giả, không phải đội chủ nhà chơi tệ hơn.
key_facts: Tỷ lệ thắng sân nhà V.League 2020: 38,1%, giảm 8,1 điểm phần trăm so với mức 46,2% của 2015–2019.; xG đội khách tăng từ 1,02 lên 1,24 bàn mỗi trận; xG chủ nhà gần như không đổi.; Phạt đền cho đội chủ nhà giảm từ 0,14 xuống 0,06 mỗi trận khi sân trống.; PPDA của đội khách giảm từ 11,4 xuống 9,8, tức pressing cao hơn.; Bộ dữ liệu gồm 156 trận mùa 2020, đối chiếu với 1.092 trận của năm mùa trước đó.
source_attribution: Phân tích dữ liệu tracking độc lập của Scarlett Martinez, công bố tháng 8 năm 2020, dựa trên băng ghi hình tự đánh dấu sự kiện và dữ liệu thống kê của ban tổ chức V.League | Cross-checked: VuaBong.vn
related_qa: question: Vì sao tỷ lệ thắng sân nhà ở V.League 2020 lại giảm mạnh đến vậy?, answer: Vì đội khách được giải phóng khỏi áp lực khán giả, dẫn tới PPDA giảm còn 9,8 và số đường chuyền vào 1/3 cuối sân tăng lên 38,9 mỗi trận.; question: Lợi thế sân nhà ở V.League đã biến mất hoàn toàn chưa?, answer: Chưa, dữ liệu cho thấy lợi thế bị dịch chuyển chứ không biến mất, và cần ít nhất ba mùa độc lập để xác nhận xu hướng.; question: Có thể dùng chỉ số nào để đo lợi thế sân nhà một cách đáng tin cậy?, answer: Chỉ số HAI, tức hiệu số xG giữa chủ nhà và đội khách trung bình mỗi trận đã chuẩn hóa theo chất lượng đối thủ.
After Round 12 of the 2026 V.League season, I reopened my spreadsheet and stopped at a single line of numbers. The league-wide home win rate: 38.1%. Across the five preceding seasons — 2026 to 2026, 1,092 matches in total — that figure sat at 46.2%. A gap of 8.1 percentage points, standard deviation of 2.3, p below 0.01. With a control sample of more than a thousand matches, this was not a bad round for a handful of home sides.
What made me stop was the structure of the decline, not its size. It did not fall evenly across every metric. It fell precisely into three groups: home goals scored, penalties awarded to the home side, and added time. Those three groups share a common denominator, and that denominator is not on the pitch.
An empty stadium does not erase the truth. It only strips away the fog that 40,000 shouted voices used to create.
Context: four months of pause, a season split in two, and a dataset that happened by accident
The 2026 V.League season kicked off in March with spectators, halted because of the pandemic, then resumed from June with most matches played without crowds or with severely restricted attendance. The format was shortened, the league split into two phases, and the season closed at 156 matches.
I tracked all 156 of those matches using two independent sources. The first was broadcast footage, on which I hand-tagged events for all 22 players on the pitch — shots, passes into the final third, pressing actions, sprint efforts. The second was the statistical feed supplied by the league organisers. The discrepancy between the two sources was 1.8% on shooting events and 2.4% on passing events. At that error margin, I accepted the dataset for analysis.
Three metrics run through this piece and need clear definitions, because they are routinely misused. xG is a model I built myself, computing the probability of a goal from shot location, angle, the number of defenders in the blocking zone, the player's stronger foot, and the type of pass immediately preceding the shot; the model was calibrated on 1,400 matches across Asian leagues. PPDA is the number of opponent passes allowed per defensive action — lower means higher pressing. The HAI (Home Advantage Index) is the average per-match difference between home and away xG, normalised for opponent quality. I put all three into one table, and that table says something no scoreline says.
Based on my experience covering matches in the V.League over seven years, I keep one rule: never read a metric without asking under what conditions it was produced. The 2026 season happened to create near-perfect experimental conditions — not because anyone wanted that, but because the pandemic forced it.
Three mechanisms behind home advantage, and all three leave the pitch when the crowd does
The first mechanism is refereeing. Over the five seasons from 2026 to 2026, home teams were awarded 0.14 penalties per match, away teams 0.08. Average added time when the home side was leading or level was 3.9 minutes; when the home side was trailing, 4.6 minutes. This pattern has been documented in several European leagues using video data, and the V.League is no exception. In 2026, with empty stands, penalties for the home side fell to 0.06 per match and average added time dropped to 3.3 minutes.
The second mechanism is physiological. With spectators present, home teams' sprint distance exceeded away teams' by 4.1%. With empty stands, that gap vanished, falling to 0.3%. In other words, the extra energy did not come from a physical base. It came from the stands. Home players did not run more because they were fitter; they ran more because someone was watching.
The third mechanism is away-team psychology, and this is the most underrated part. With spectators present, away teams completed 32.4 passes into the final third per match, with a misplaced-pass rate in that zone of 24.1%. With empty stands, those figures became 38.9 and 19.7%. Away teams play more boldly when nobody is jeering at them. They dare to play the difficult pass, dare to hold the ball half a beat longer, dare to push their full-backs higher.
My aggregate table, comparing the 2026–2026 average with the 2026 season:
Home win rate: 46.2% down to 38.1%. Home goals per match: 1.62 down to 1.29. Home xG per match: 1.44 down to 1.31. Away xG per match: 1.02 up to 1.24. Away PPDA: 11.4 down to 9.8. Penalties for the home side per match: 0.14 down to 0.06. Average added time: 4.1 minutes down to 3.3 minutes.
Home xG barely moved, while away xG rose by 0.22 goals per match. Home advantage in the 2026 V.League fell not because home teams played worse, but because away teams were freed from a psychological variable that has never appeared in any official statistical table. This is the point most post-match reports missed: they looked for the cause in the form of the losing side, while the answer lay in the changed behaviour of the winning side on the road.
When the press room laughs at xG, I know I am reading exactly the book they have not opened.
The prediction model failed, and the market adjusted three rounds behind the data
I re-ran a bivariate Poisson model — attack and defence parameters for each team, plus a home coefficient — across all 156 matches. Holding the old home coefficient (calculated from 2026–2026), mean absolute error was 0.71 goals per match. Substituting the new coefficient brought it down to 0.61. On outcomes, the old model got 44% of matches wrong; the adjusted model got 37% wrong.
I named that adjustment coefficient alpha, and it is not a constant. It depends on the permitted stadium capacity, the away team's travel distance, and fixture density over the preceding two weeks. For an empty-stadium match with travel under 300 km, alpha converges towards 1.0 — meaning home advantage is effectively nil. For an empty-stadium match where the away side must travel more than 800 km inside a congested schedule, alpha still sits around 1.15.
One side effect matters more than the model itself. Over the first six rounds of the second phase, the average home handicap on the market fell from 0.5 goals to 0.25. But my tracking data indicated the rational adjustment should have appeared roughly three rounds earlier. The market did not misread the data; the market simply read it late.
After my first analysis on this subject was published, a data analyst at Hanoi FC got in touch and said they had applied the adjustment-coefficient idea in preparing for away trips. I do not know exactly how they used it, and I did not ask. What I do know is that from Round 15 onwards, the away points collected by the title-chasing group rose noticeably compared with their own first-phase record.
The crowd may remember the goal forever. I remember the third pass before it, where the decision was actually made.
The counter-intuitive angle: do not turn 156 matches into a religion
There is one very common misreading I want to block right here: the conclusion that home advantage is dead. It is not dead. It has merely been displaced, and the displacement itself may not be down to the crowd.
Four confounders must be subtracted before attributing the full 8.1-percentage-point decline to the absence of spectators. First, the compressed schedule and the expansion to five substitutions disproportionately benefit the team with greater squad depth — and the team with greater depth is often the away side in a given fixture. Second, the four-month mid-season pause reset every squad's physical base, and whichever team prepared better during the interruption gains regardless of the stands. Third, the second-phase sample is far smaller than the first phase, so the standard error there is considerably higher. Fourth, the pitch surfaces at some stadiums deteriorated after a long period without maintenance, and that hurts possession-based styles more than counter-attacking ones.
A single number can lie, but a model validated across 10,000 matches has no reason to pretend. I only have 156 matches. 156 matches is a sample, not a law. I refuse to call it a law until at least three independent seasons confirm it.
Takeaway: the signal to watch next round
What I will measure next season is not the home win rate but the shape of the recovery curve. When stadiums reopen at 50% capacity, does home advantage return as a step function or a linear one? If it is a step function, it means players only need someone in the stands, regardless of how many. If it is linear, it means they can read the volume, and every calculation about stadium capacity becomes a genuine tactical variable.
I have already built the spreadsheet for next season. The alpha column is blank, waiting for data.
