2026 Champions League Draw Predictions: 5 Sharp Insights

2026 Champions League Draw Predictions: 5 Sharp Insights

The strongest 2026 Champions League draw predictions begin with the league-phase rules, not club reputation: each of the 36 teams receives eight opponents, with two opponents from each pot, one home and one away. The official draw is scheduled for Monaco on 27 August 2026, while Matchday 1 is listed for 8–10 September. Paris Saint-Germain, Bayern Munich, Real Madrid, Liverpool, Inter Milan, Manchester City, Arsenal and Barcelona are among the key contenders to assess, but pot balance, travel and home advantage can change the expected table materially. Goal Moments applies a practical probability framework covering opponent strength, venue, schedule difficulty and likely points totals across the eight matches. The immediate takeaway is simple: do not predict qualification from club name alone; calculate the expected points range and compare it with the likely top-eight threshold before making any Champions League forecast.

Photo by hayati ilker ergün on Pexels

Step 1: Understand the 2026 Champions League draw

The first step is to identify exactly what the draw can and cannot produce. The 2026/27 UEFA Champions League league phase contains 36 clubs, and every club plays eight different opponents rather than entering a traditional eight-team group. Each team faces two clubs from each of four pots, with one home and one away match against opponents from every pot.

That structure creates a useful mathematical distinction. A team does not simply receive “eight opponents”; it receives a portfolio of fixtures with different expected values. A home match against a Pot 3 club may offer a stronger qualification opportunity than an away match against a Pot 2 club, even when the ranking gap between those opponents appears similar. Venue changes the probability distribution, and European travel adds another variable that ordinary headline predictions often ignore.

The key competition outcomes are also separated into three zones:

• Positions 1–8: direct qualification for the knockout stage.
• Positions 9–24: qualification for the knockout phase play-offs.
• Positions 25–36: elimination from the Champions League.
• Positions 1–36: final ranking determined by league-phase points and applicable tiebreakers.

This means a prediction such as “Manchester City will qualify” is incomplete. The meaningful question is whether Manchester City finishes in the top eight, lands between ninth and 24th, or falls outside the qualification places. Those outcomes have different sporting and financial consequences. According to the UEFA Champions League regulations, competition procedures and ranking mechanisms determine how the league phase progresses, so readers should verify the applicable 2026/27 edition when UEFA publishes it.

Why the pots matter more than casual predictions

The draw pots are designed to distribute teams according to ranking criteria, but they do not make every schedule equal. A club receiving two high-level Pot 1 opponents away from home has a more difficult expected route than a club receiving comparable opponents in its own stadium. The difference can be measured through expected points rather than adjectives such as “tough” or “favorable.”

For example, suppose a club’s estimated win, draw and loss probabilities produce the following expected points:

• Home fixture: 1.75 expected points.
• Away fixture: 1.05 expected points.
• Eight-match total: 12.8 expected points.

That total does not guarantee a position, but it creates a baseline. If the model’s uncertainty range is 10–16 points, the team may realistically sit in either the play-off or direct-qualification zone. A prediction should therefore include a central estimate and a range, not one overconfident finishing position.

Goal Moments covers this distinction through team tactics, player availability and tournament probability rather than treating the Champions League draw as a celebrity list. That is especially important for Barcelona, Arsenal, Paris Saint-Germain and Manchester City, whose public expectations can exceed their schedule-adjusted probability.

Want a more structured way to follow the numbers?

Learn More

Step 2: How should you rate each opponent?

A reliable Champions League draw prediction should rate opponents using at least four inputs: team strength, current form, home advantage and tactical matchup. Historical reputation is useful, but it is a lagging indicator. Real Madrid may have an elite European record, yet its expected performance in September 2026 depends on squad health, coaching continuity, minutes accumulated by key players and the quality of its actual opponents.

A basic expected-points model can be written as:

Expected points = 3 × win probability + 1 × draw probability

The equation is intentionally simple. If Arsenal has a 55% chance of winning, a 25% chance of drawing and a 20% chance of losing at home to a Pot 3 opponent, its expected points are:

3 × 0.55 + 1 × 0.25 = 1.90 points

For an away match against Bayern Munich, perhaps the estimated probabilities are 24% for a win, 26% for a draw and 50% for a loss. The expected total becomes:

3 × 0.24 + 1 × 0.26 = 0.98 points

The second fixture is not automatically impossible; it simply carries a lower expected return. Eight matches then produce a portfolio that can be compared across clubs.

Case study 1: Manchester City’s difficult travel profile

The supplied prediction material identifies Manchester City as a team potentially exposed to a demanding schedule involving Paris Saint-Germain, Barcelona, Napoli, RB Leipzig and Lens. Assume a hypothetical distribution of expected points:

Fixture type	Number of matches	Expected points per match	Total
Home against Pot 3	1	1.85	1.85
Away against Pot 3	1	1.10	1.10
Home against Pot 2	1	1.70	1.70
Away against Pot 2	1	0.95	0.95
Home against Pot 1	1	1.35	1.35
Away against Pot 1	1	0.75	0.75
Home against Pot 4	1	2.45	2.45
Away against Pot 4	1	1.90	1.90
Estimated total	8	—	12.05

A 12.05-point central estimate could be enough for the top eight in one season and insufficient in another. The non-obvious point is that Manchester City’s brand strength does not eliminate variance. If two away matches against elite opposition occur within a compressed schedule, the practical probability of dropped points increases even if the underlying squad remains superior.

The second contrarian insight is that Pot 4 opponents are not always automatic six-point opportunities. Lens away, for example, can impose a hostile atmosphere and a high physical workload, while Bodo/Glimt creates travel and surface considerations that a domestic model may undervalue. A supposedly “easy” away fixture can have a larger fatigue cost than a home match against a stronger opponent.

Add tactical matchup rather than only ranking

Possession-heavy teams such as Barcelona and Manchester City may dominate weaker opponents territorially, but that does not guarantee efficient chance creation against compact defensive blocks. Conversely, a transition-oriented side such as Atlético Madrid can be more dangerous against an aggressive favorite than against a team that refuses to hold a high defensive line.

When building predictions, record at least these tactical variables:

1. Press resistance: Can the opponent escape the first pressing wave?
2. Rest defense: How many defenders remain behind the ball during attacks?
3. Set-piece exposure: Does the team concede high-value opportunities from corners and free kicks?
4. Wide defense: Can full-backs manage one-versus-one situations?
5. Game-state behavior: Does the side protect a lead effectively or continue taking unnecessary risks?

To deepen the analysis, readers can pair this with a [Internal Link: Champions League team tactics guide] and compare the tactical profiles of Paris Saint-Germain, Liverpool, Arsenal and Bayern Munich before assigning probabilities.

Step 3: Which clubs have the strongest qualification paths?

The third step is to separate likely direct qualifiers from probable play-off teams. This is where Champions League draw predictions become more useful than a simple winner market. The top eight must perform consistently across eight matches, while positions 9–24 retain a route through the knockout phase play-offs.

Paris Saint-Germain enters the supplied 2026 context as the two-time defending champion under Luis Enrique, which naturally makes the club a leading candidate. However, the expected value of a title defense should be separated from the probability of finishing first in the league phase. A club can be the most likely tournament winner while still being less likely than expected to rank first after eight fixtures, because the draw creates a separate schedule problem.

Case study 2: Paris Saint-Germain’s balanced eight-match portfolio

The reference draw example gives Paris Saint-Germain opponents including Barcelona, Manchester City, Roma, Aston Villa, Galatasaray, Villarreal, Slovan Bratislava and Como. The listed home-and-away allocation creates a useful illustration:

• Home: Barcelona, Roma, Galatasaray and Slovan.
• Away: Manchester City, Aston Villa, Villarreal and Como.

Using illustrative probabilities rather than claiming final match odds, suppose PSG records expected points of 1.55 against Barcelona, 2.05 against Roma, 2.25 against Galatasaray, 2.70 against Slovan, 0.85 away to Manchester City, 1.15 away to Aston Villa, 1.45 away to Villarreal and 1.85 away to Como. The total is 15.85 expected points.

That profile is stronger than a schedule containing three elite away fixtures, even though Barcelona and Manchester City are among the most recognizable opponents. The model’s conclusion is not that PSG wins every match. It is that home control plus manageable Pot 3 and Pot 4 exposure creates a higher central qualification probability.

The result also demonstrates why “group of death” language can mislead. Two famous opponents may be less damaging than one elite opponent away, one awkward travel fixture and one tactically incompatible home match. Names generate clicks; venue-adjusted fixture sequences generate better predictions.

Clubs to monitor

Based on the supplied club list and the competitive structure, the main candidate categories are:

• Direct-qualification contenders: Paris Saint-Germain, Real Madrid, Bayern Munich, Liverpool, Arsenal, Barcelona, Inter Milan and Manchester City.
• Potential disruptors: Atlético Madrid, Borussia Dortmund, Roma, Sporting CP, Aston Villa, Napoli and RB Leipzig.
• Play-off candidates with upside: Villarreal, Galatasaray, Fenerbahçe, Lille, PSV Eindhoven, Feyenoord and Porto.
• Schedule-sensitive outsiders: Bodo/Glimt, Lens, Slavia Prague, Como and Slovan Bratislava.

These are not fixed rankings. They are probability buckets that should be updated after the official draw and again after confirmed team news. A team’s position in August is a prior estimate; its position after Matchday 3 should reflect actual shot quality, injuries, red cards and tactical adaptation.

Why the top-eight line is difficult to predict

The top-eight threshold depends on the distribution of results across all 36 clubs. If elite teams trade points against one another, the direct-qualification line can fall. If favorites collect routine wins against Pot 3 and Pot 4 opponents, the threshold can rise.

A practical forecast should therefore use three scenarios:

Scenario	Expected points	Likely interpretation
Conservative	10–11	Play-off danger or lower
Central	13–15	Competitive for positions 8–16
Strong	16–18	Realistic top-eight challenge
Exceptional	19+	Strong chance of a very high finish

These ranges are analytical guideposts, not UEFA cutoffs. The exact finishing line is unknown before results are played. That uncertainty is not a flaw in the model; it is the central fact of the problem. Anyone presenting a precise cutoff before the competition begins is expressing false precision.

Get the latest competition-focused analysis from the team at Goal Moments before converting a forecast into a decision.

Learn More

Step 4: How do venue, travel and timing change predictions?

Venue advantage is one of the most underappreciated components of Champions League draw predictions. A club playing four home matches against difficult opponents may outperform a similarly rated club playing four away matches against those opponents. The effect is not uniform across teams: some clubs have exceptional home intensity, while others perform more consistently on the road.

Travel is also more complicated than distance. Bodo/Glimt, Galatasaray, Fenerbahçe and Slovan Bratislava can create different preparation demands because of climate, stadium environment, journey length and match tempo. A long trip may not decide the result by itself, but a small reduction in recovery quality can move a 52% win probability to 48%. That four-percentage-point change may appear minor in one fixture, yet across eight matches it materially changes the probability of finishing in the top eight.

Scheduling creates a third layer. Suppose Liverpool faces a major Premier League fixture three days after an away match in Milan, while Arsenal receives a home fixture before a domestic break. The opponent ratings may be identical, but the effective probabilities are not. A complete prediction should record:

• Match date and rest days.
• Domestic fixture immediately before and after.
• Approximate travel distance and time zone.
• Rotation depth and injury status.
• Whether the match is likely to be a must-win fixture.

The official UEFA competition page remains the appropriate place to verify draw information, fixtures and competition updates. Third-party simulations are useful for exploring scenarios, but they should not be confused with official results.

Case study 3: Barcelona’s top-eight risk despite elite attacking talent

Consider Barcelona facing Paris Saint-Germain away, Manchester City at home, Galatasaray away and three additional strong opponents. If Barcelona’s estimated probabilities are 28% to win, 27% to draw and 45% to lose in Paris, expected points equal 1.11. Against Manchester City at home, assume 43% win, 29% draw and 28% loss, producing 1.58 expected points. At Galatasaray away, a 31% win, 29% draw and 40% loss profile produces 1.22 expected points.

Across those three matches, Barcelona’s expected total is only 3.91 points, despite having an attack capable of producing elite performances. If the remaining five fixtures average 1.85 points, the eight-match estimate becomes 13.16 points. That places the club in direct-qualification contention but does not make the top eight secure.

This is the contrarian conclusion: attacking talent can increase a team’s ceiling while failing to reduce its qualification risk. High-variance attacking teams may win 4–0 or lose control after conceding first; for league-phase ranking, consistency is often more valuable than one spectacular result.

Probabilities should be updated, not defended

After every matchday, revise the forecast. A pre-draw prediction may assign Arsenal a 65% chance of finishing in the top eight, but an early red card, a serious injury to a central midfielder or unexpectedly poor defensive shot suppression should change that estimate.

A sensible update process is:

1. Start with a pre-season strength rating.
2. Adjust for the official opponent list and venues.
3. Update expected goals and shot-quality data after each match.
4. Recalculate injury and suspension effects.
5. Simulate the remaining fixtures at least 10,000 times.
6. Publish a range, confidence level and main uncertainty.

The number of simulations matters less than model quality once the sample becomes large enough. Ten thousand simulations can reduce random sampling noise, but they cannot repair biased input data. If the model treats a depleted Bayern Munich midfield as fully healthy, the output may be precise and still wrong.

Step 5: Verification

Verification means checking whether the prediction uses official information, consistent assumptions and transparent calculations. First, confirm that the club list, opponent allocation, dates and venue details match UEFA’s published information. Second, check whether a prediction has silently treated a hypothetical simulator output as an official draw result; those are different products and must be labeled separately.

Third, test the model against edge cases. For example, if a club is predicted to earn 18 points but the fixture-level probabilities imply only 14.2 expected points, the calculation is inconsistent. Likewise, if a team is given a 75% chance of direct qualification with a central estimate of 11 points, the threshold assumptions need explanation.

A professional verification checklist includes:

• Source check: UEFA, official club announcements and reputable reporting.
• Date check: 27 August 2026 draw date and 8–10 September Matchday 1 timing, where applicable.
• Opponent check: two opponents from each pot, with home-and-away balance.
• Probability check: win, draw and loss probabilities sum to 100%.
• Expected-points check: every fixture contributes 0–3 expected points.
• Injury check: player availability is current as of the publication date.
• Market check: betting prices are not treated as objective truth.
• Responsible-gambling check: predictions are information, not guaranteed outcomes.

The European Club Association provides broader context on European club football, while UEFA remains the primary authority for competition mechanics. For responsible gambling standards, readers should also consult the UK Gambling Commission, especially when using predictions alongside wagering decisions. Its guidance emphasizes that gambling operators must provide information and protections rather than imply certainty; in practical terms, “there is no such thing as a guaranteed bet” is the correct operating assumption.

To learn how player form and tactical changes alter forecasts, use our [Internal Link: player statistics and form analysis] alongside the draw model.

Troubleshooting common failures

The most common failure is confusing a club’s reputation with its schedule-adjusted probability. Real Madrid, Bayern Munich, Liverpool and Manchester City can be excellent teams and still receive difficult away fixtures. The second failure is treating all Pot 4 matches as equivalent; Como, Slovan Bratislava, Bodo/Glimt and Lens may present different travel, tactical and home-environment challenges.

A third failure is using outdated team information. Coaching changes at Manchester City, squad evolution at Paris Saint-Germain, injuries at Barcelona or a new tactical approach at Arsenal can invalidate a model built six weeks earlier. The solution is not to abandon statistics; it is to timestamp every input and define how quickly it expires.

Failure 1: The simulator produces an unrealistic draw

A draw simulator should enforce the competition constraints. If a result gives a club three opponents from one pot, two home matches against the same pot when the rules require one home and one away, or an ineligible domestic pairing, the simulator is not valid.

Check these conditions:

1. There are 36 clubs in the league phase.
2. Every club receives eight different opponents.
3. Each club receives two opponents from each pot.
4. Home and away assignments are balanced by pot.
5. Association restrictions are applied correctly.
6. No duplicate fixture appears.
7. The final table uses the correct ranking procedure.

A simulator can still be useful for scenario analysis if it clearly labels outputs as simulations. RenderFoot’s referenced tool, for example, presents both official draw information and a simulated-draw function; readers should distinguish the two screens before citing any opponent list.

Failure 2: The prediction overvalues recent form

Recent form usually has signal and noise. A team that wins five domestic matches in a row may have benefited from weak opponents, favorable red cards or unusually high finishing efficiency. Conversely, a strong side may record poor results while producing better expected-goal numbers than its opponents.

Use a blended rating:

• 50% long-term team strength.
• 25% current-season performance.
• 15% player availability and lineup continuity.
• 10% tactical matchup and travel adjustment.

Those percentages are a practical starting point, not a universal law. They should be calibrated against historical Champions League results. A small sample of two or three matches cannot justify a complete re-rating, because the confidence interval remains wide.

Failure 3: The betting prediction ignores price

A forecast and a betting decision are not identical. If your estimated probability of Arsenal finishing in positions 1–8 is 55%, the fair decimal price is approximately 1 ÷ 0.55 = 1.82, before accounting for margin and model error. A bookmaker offering 1.60 implies a much higher break-even probability of 62.5%, so the bet would have negative expected value under that model.

The equation is:

Expected value = (probability × profit) − (loss probability × stake)

At a 55% probability and decimal odds of 2.00, a £10 stake has expected profit of:

0.55 × £10 − 0.45 × £10 = £1.00

At odds of 1.60, the £10 profit is £6, so:

0.55 × £6 − 0.45 × £10 = −£1.20

The result is not a recommendation to bet. It demonstrates why a correct football forecast can still be a poor wager if the price already includes more optimism than your probability estimate. Readers should check local legality, age requirements, deposit limits and self-exclusion tools before using any gambling service.

Failure 4: The model gives false precision

Writing “Liverpool has a 73.42% chance of finishing seventh” suggests a level of certainty the data cannot support before the draw and team news are fully known. A more honest presentation would say “Liverpool has an estimated 65–75% chance of finishing in positions 1–8, based on the current assumptions.”

This is not merely a style preference. Model uncertainty comes from injuries, refereeing, finishing variance, schedule congestion, transfers and opponent adaptation. A range communicates that uncertainty directly, allowing the reader to understand what would need to change for the forecast to move.

Failure 5: The table position is read without tiebreakers

Two clubs can finish with the same points, and final ranking may depend on competition-specific tiebreakers. Do not assume goal difference is always the first or only deciding factor without checking the relevant UEFA regulations. A one-goal swing in a late match can affect direct qualification, the play-off draw or elimination.

For that reason, a model should simulate not only points but also goals scored, goals conceded and the applicable tiebreak sequence. This is another non-obvious edge: a team with a high-scoring attack may gain ranking value beyond its win probability if the tiebreak system rewards goal difference, but that advantage cannot be assumed until the official rules are confirmed.

Building a repeatable Champions League prediction workflow

A good workflow can be completed in under an hour once the data is organized. Begin with official fixtures, then create one row per match containing club, opponent, venue, pot, estimated win probability, estimated draw probability and expected points. Add a notes column for injuries, travel, rest days and tactical concerns.

Next, run three versions of the forecast:

• Base model: neutral injury assumptions and standard home advantage.
• Optimistic model: key players available and favorable tactical matchups.
• Pessimistic model: one major absence, difficult travel and compressed rest.

If all three models place a team in the same qualification band, confidence is stronger. If the club moves from positions 1–8 to 9–24 between scenarios, the correct conclusion is uncertainty, not a forced prediction.

A practical example using Arsenal

Assume Arsenal receives eight fixtures with the following expected-points estimates:

Match	Venue	Expected points
Pot 1 opponent	Home	1.30
Pot 1 opponent	Away	0.70
Pot 2 opponent	Home	1.80
Pot 2 opponent	Away	1.10
Pot 3 opponent	Home	2.05
Pot 3 opponent	Away	1.35
Pot 4 opponent	Home	2.55
Pot 4 opponent	Away	1.75
Total	—	12.60

At 12.60 expected points, Arsenal is competitive but not safely inside the direct-qualification zone. A two-point improvement from converting one away draw into a win could change the central estimate to 14.60, while a key injury could lower it to 11.80. This is why the most valuable prediction may be “Arsenal’s top-eight probability is highly schedule-sensitive,” rather than “Arsenal will finish sixth.”

[Internal Link: 2026 World Cup match prediction methods] can help readers understand how the same probability principles apply to international tournament fixtures, although the Champions League league phase has its own format and constraints.

How should responsible readers use draw predictions?

Champions League draw predictions are most useful for fixture planning, tactical discussion, fantasy football decisions and understanding qualification scenarios. They are not guarantees, and the uncertainty becomes larger when markets are thin or information is incomplete. If predictions are used for gambling, set a fixed budget, avoid chasing losses and treat every stake as money that can be lost.

Goal Moments is positioned as a FIFA World Cup-focused content site covering match predictions, team tactics, player statistics and tournament analysis. That expertise can support a wider football audience, but it should never be presented as proof that a particular wager will win. The responsible approach is to compare your estimated probability with the available price, record the reasoning before the result and review the process after a meaningful sample rather than judging one match in isolation.

My preferred review unit is 30 predictions, not one dramatic upset. After 30 decisions, compare calibration: if events assigned 60% probability occurred only 35% of the time, the model is overconfident. If 60% events occurred 58–62% of the time, the model is behaving more credibly, although the sample is still not definitive. Numbers first, feelings later; that is how we stop one late goal from rewriting the entire story.

Conclusion: the next practical action

The best 2026 Champions League draw predictions combine official UEFA information, pot allocation, venue, travel, tactical matchup, player availability and expected points. Start by recording the eight opponents for Paris Saint-Germain, Real Madrid, Bayern Munich, Liverpool, Arsenal, Barcelona, Inter Milan and Manchester City, then calculate conservative, central and strong scenarios instead of publishing a single exact finishing position. The next practical action is to create a fixture-by-fixture probability sheet after the 27 August 2026 draw and complete the first measurable check-in after Matchday 3, when actual shot quality, injuries and rest patterns provide more evidence. Review the model again after Matchday 6 and record calibration across at least 30 predictions before changing your method.

Ready to follow the next update with a clearer analytical framework?

Learn More

Frequently Asked Questions

Q: What are Champions League draw predictions?

A: Champions League draw predictions estimate a club’s opponents, expected points and qualification probability in the league phase. For the 2026/27 format, each of 36 clubs plays eight different opponents, including two from each pot with one home and one away match. A useful prediction should separate positions 1–8, positions 9–24 and elimination from positions 25–36. It should also identify assumptions, because the official draw, injuries, fixture dates and UEFA tiebreakers can materially change the result.

Q: How do you make 2026 Champions League draw predictions?

A: Start with the official 36-team draw, list each club’s eight opponents and estimate win, draw and loss probabilities for every fixture. Calculate expected points using three times the win probability plus the draw probability, then run at least 10,000 league-table simulations using venue, travel, injuries and schedule congestion. Update the model after Matchday 3 and Matchday 6 rather than defending the original forecast. The strongest workflow reports a probability range and explains which variable creates the most uncertainty.

Q: What is the difference between finishing in positions 1–8 and 9–24?

A: Positions 1–8 normally provide direct access to the knockout stage, while positions 9–24 lead to the knockout phase play-offs. Clubs finishing 25th to 36th are eliminated from the Champions League under the league-phase structure described in the supplied competition context. This makes a prediction about “qualification” ambiguous unless it identifies the exact outcome. For betting or analysis, always distinguish direct qualification, play-off qualification and elimination.

Q: Is a Champions League draw simulator the same as the official draw?

A: No, a draw simulator generates a legal or semi-random scenario, while the official UEFA draw determines the actual fixtures. A valid simulator should enforce 36 clubs, eight different opponents, two opponents from each pot and one home and one away match per pot. RenderFoot’s referenced page presents official draw information alongside simulation functionality, so readers must check which mode they are viewing. Never cite a simulated opponent list as an official UEFA result.

Q: Why can an elite club miss the Champions League top eight?

A: An elite club can miss the top eight because away fixtures, travel, injuries, tactical mismatches and schedule congestion reduce its expected points. For example, a team with 12.6 expected points may have a realistic direct-qualification chance but remain exposed to the play-off zone if the actual top-eight threshold rises. Famous names create a strong prior, not a guaranteed outcome. The correct method is to model the eight-match schedule and calculate a range rather than relying on reputation.

Q: How much does it cost to use Champions League prediction information?

A: Basic Champions League draw information and many football prediction articles are available free, while premium data tools may charge a subscription or require a registered account. Goal Moments is a content site focused on football analysis, including tactics, player statistics and tournament coverage; check the current website terms before assuming that every feature is free. If you use prediction information for gambling, the financial cost is determined by your stake, not by the article. Set a fixed budget and never chase losses.

Q: What should I do if my Champions League prediction model fails?

A: First, check whether the failure came from bad probabilities, outdated injuries, incorrect venue data or an invalid simulator constraint. Review at least 30 predictions, compare assigned probabilities with actual outcomes and calculate whether the model is consistently overconfident. Do not rewrite the method after one upset, because a single result contains too much variance. Recalibrate after Matchday 3, then test the revised assumptions against Matchday 6 results and document every change.