The scoreboard, kept honestly.
Real periodic reports on consensus accuracy and league trends, computed directly from real finished fixtures — a dated summary of how the numbers actually held up, not a one-time claim.
During the week of September 14–20, the consensus hit rate across 694 graded fixtures stood at 51%, reflecting a near-even split between correct and incorrect predictions. The biggest upset of the week saw Hatta SC draw 2-2 with Shabab Al Ahli Dubai despite the latter carrying the highest confidence rating among matches that failed to deliver the expected result, with a pre-match consensus of 50%. On the Prime Pick front, 8 of 13 settled selections came through successfully for the week.
In August 2026, across 12,592 finished fixtures, the average goals per match stood at 3.15 for all tracked leagues. Romania's Liga 1 Feminin was the highest-scoring competition, recording 9.88 goals per match, well above the overall average. Spain's La Liga registered the highest share of consensus misses among all tracked leagues during the same period.
Past reports
For the week of September 7–13, the consensus hit rate across 1,452 graded fixtures with a real pre-match consensus came in at 49%, sitting just below the break-even threshold. The biggest upset of the week saw Ezra defeat Shan United 3-2, despite Shan United carrying the highest pre-match confidence among losing favourites at 50%. In the Prime Pick selection, 8 of 14 settled picks landed successfully over the course of the week.
During the week of August 31 to September 6, the consensus hit rate across 1,285 graded fixtures stood at 48%, meaning favoured outcomes failed to materialise at a slightly higher rate than they succeeded. The biggest upset of the week saw Brighton, who carried a 50% pre-match consensus, held to a 1-1 draw by Leeds. In the Prime Pick selections, only 5 of 12 settled picks proved correct.
What a report is for — and how it's different from everything else on this platform
Everywhere else on FDVS shows you a current snapshot — today's fixtures, this week's results, right-now ratings. A report is different on purpose: it's a fixed summary of a specific, closed period, generated once from real finished fixtures and left as a permanent record rather than continuously updated. Last week's report describes last week; it doesn't quietly change if this week goes badly.
Weekly versus monthly reports
Weekly reports track real consensus accuracy at a granular level — useful for spotting a short-term calibration wobble early, and only generated once at least a handful of real graded fixtures exist for that week. Monthly reports zoom out to real league-wide scoring and upset trends, which need a larger sample to say anything meaningful.
Why every number here is computed, not estimated
Every metric on this page — the hit rate, the fixture count, the biggest upset, the league scoring averages — is computed directly from real finished fixtures and their real pre-match consensus, the same underlying data behind the results and league statistics pages. Only the short summary paragraph is written with AI assistance, and even that is built strictly from the real numbers already computed, never from an invented statistic.
How this connects to the results page
Every fixture behind these reports is a result already shown individually on the results page, verdict and all — reports don't introduce new grading criteria, they aggregate the same honest wins-and-misses record into a period summary computed automatically.
Why reports don't get revised after publishing
Once a weekly or monthly report is generated, it's locked to that period permanently — the underlying system won't regenerate or overwrite it later, even if more data comes in. If something needs correcting, that happens as a new, separately dated note rather than silently rewriting history.
Coverage and cadence
A weekly report is generated for the most recently completed Monday-to-Sunday week once enough real graded fixtures exist; a monthly report is generated for the most recently completed calendar month once enough real finished fixtures exist. A period with too few real fixtures behind it is skipped entirely rather than published with a misleadingly small sample.