Fenerbahce vs Lyon: What Ranked Scorelines Really Mean
Quick Answer
Why a second-ranked score matching a result does not prove model accuracy, using the earlier Fenerbahce–Lyon example and goals-table arithmetic.
The earlier Fenerbahce–Lyon article listed 2–1, 1–1 and 2–2 as its three ranked scores. It used an asserted 1–1 result to argue that the second or third choice “very often” succeeds when the first misses. That claim needs a complete record and cannot be established from one example.
What the archive contains
The article supplied a ranking but no individual score probabilities. Without those probabilities, we cannot calculate the chance assigned to the three-score set or how close the first and second choices were.
It also quoted these goals estimates:
| Market | Archived figure |
|---|---|
| Over 0.5 | 97.4% |
| Over 1.5 | 88.4% |
| Over 2.5 | 71.6% |
| Under 3.5 | 49.1% |
These are retained from the original text. A fixture-specific dated output and an official result tied to the precise match date have not been established here. This is an interpretation example, not a verified successful forecast.
What can be calculated
If the figures describe the same distribution, exactly two goals receive 16.8%: 88.4 minus 71.6. Over 3.5 is the complement of under 3.5, or 50.9%. Exactly three goals would then receive 20.7%: 71.6 minus 50.9.
The 16.8% for two goals includes 2–0, 1–1 and 0–2. It cannot all be assigned to 1–1.
How to evaluate a ranked list
Decide the selection rule before seeing the result. If the rule is “top score only,” a second-ranked score is a miss. If the rule is “any of the top three,” assess that same rule for every eligible fixture and disclose that it covers three outcomes.
A larger set naturally covers more possible results. That does not by itself demonstrate better calibration or profitability. A complete evaluation also records the assigned probabilities, sample period and missing results.
The correct-score software listing describes the related product. For a screenshot-based example, see the Bravo–Shkendija guide. The free website track record is a separate dataset.
Correction, 19 September 2026: Removed the unsupported claim that lower-ranked scores often rescue the top pick, and stopped treating an undated result and missing output as verified evidence.
Revised with AI assistance using the available article evidence. Predictions are estimates, not guarantees. 18+. Support for gambling-related harm.