How KickOracle predicts the World Cup
KickOracle turns public football data into win probabilities for all 48 World Cup 2026 teams. This page explains exactly how that happens — which inputs feed the model, what our Chemistry Index measures, how predictions move during the tournament, and, just as importantly, what the model does not yet do.
Last reviewed:
Most World Cup coverage hands you a verdict and asks you to trust the byline. KickOracle works the other way around. Every probability we publish is the output of a documented, repeatable formula, and this page is the formula written out in full. If you disagree with a prediction, you should be able to point at the exact input you think we have weighted wrong — that is the standard we hold ourselves to.
We are an independent analytics project, not a tipster service and not a sportsbook. Our goal is narrative-first football intelligence: clear explanations of why one squad is favored over another, grounded in numbers we are willing to show. Nothing on this page is advice to wager money. It is a description of how a prediction model reasons about a football tournament.
Data inputs and sources
KickOracle's team strength score is a transparent weighted blend of five inputs. None of them are secret, and none of them require proprietary scouting access. Each team carries a value for FIFA ranking, squad chemistry, morale, tactical stability, and tournament familiarity, and the model combines those five into a single power score that drives the rankings and feeds into match probabilities.
The underlying data comes from public sources: the official FIFA/Coca-Cola World Ranking, publicly announced squad lists and call-ups, historical international match results, confederation qualifying records, and openly reported news on injuries, coaching changes, and camp morale. KickOracle then derives its own composite metrics — chemistry, morale, stability, and familiarity — on top of that public foundation. We do not buy proprietary scout reports or paid expert tips, and we do not treat any single pundit's opinion as a model input.
The five inputs and their weights
The largest single input. A team's position in the official FIFA World Ranking is converted to a 0–100 baseline — the higher the rank, the higher the score. It is an externally maintained, slow-moving measure of long-run results, which makes it a stable anchor for everything else.
KickOracle's estimate of how well a specific squad functions as a unit: shared club connections, settled partnerships, and continuity from one camp to the next. This is the metric that most often separates two similarly ranked teams.
A read on recent sentiment around the camp — qualifying momentum, the mood after warm-up results, and openly reported friction or unity. Morale captures the short-term swing that a static ranking misses.
How much the coaching staff and first-choice lineup have churned. A team that changed manager late, or that cannot settle on a spine, is penalised relative to a side that has been consistent for months.
Coach tenure and how much of the player cohort has been retained across cycles. Tournament-tested groups that have been together through multiple campaigns score higher than freshly assembled ones.
These weights are deliberately public and are frozen before the tournament's first kickoff so that nobody can claim we quietly tuned the model to fit results. The power score is a 0–100 figure: a settled, in-form squad ranked just outside the global top ten can comfortably out-score a higher-ranked side carrying turmoil, which is the model behaving exactly as designed.
The Chemistry Index
The Chemistry Index is KickOracle's distinctive metric and the thirty-percent slice of the power score that you will not find in a plain ranking table. International football is unusual: a manager assembles a squad from players who spend most of the year at rival clubs, then has only a handful of training sessions to forge them into a team. The Chemistry Index is our attempt to quantify how successfully a given squad has crossed that gap.
A high chemistry score does not mean a team has the best individual players. It means the parts fit. Two nations can sit within a place or two of each other in the FIFA ranking and still be separated clearly by the model because one has a cohesive, continuous spine while the other is rebuilding mid-cycle. When a lower-seeded team is favored in one of our match cards, the Chemistry Index is usually the reason — and pointing readers to that reason is the entire purpose of publishing it.
What feeds the Chemistry Index
- • Shared club affiliations and established on-pitch partnerships within the squad.
- • Continuity of the core group from qualifying through to the final squad list.
- • Coaching stability and the length of the current manager's tenure.
- • Retention of tournament-experienced players from previous World Cup and continental cycles.
How predictions update
Predictions are not static. The five inputs behind every team's power score change as real-world news arrives — a key injury, a settled-on starting eleven, a confidence-shifting warm-up result — and the published probabilities move with them. Server-side data refreshes on a short cycle measured in minutes, and a daily editorial review captures the softer signals: squad-list changes, fitness updates, and shifts in camp morale that the raw feeds do not flag on their own.
There is one hard cut-off. The model formula and its five input weights are frozen before the tournament's opening match. After kickoff, the inputs keep updating but the recipe that combines them does not. Freezing the weights is a discipline check: it stops anyone, including us, from quietly re-tuning the model to flatter results once games are being played. All predictions are pre-match — we do not adjust a probability while a game is in progress.
What is modeled now vs. at tournament launch
Honesty about scope matters more than a polished pitch. KickOracle is pre-launch, and we would rather state plainly what is and is not running today than imply a finished system. The power score described above is fully live. The deeper match-engine components are still being built, and until they ship, a few values on the site are clearly-labelled placeholders rather than live model output.
Running on real, documented logic
- • The five-input power score that drives team ratings and the Power Rankings page.
- • The squad Chemistry Index and the morale, stability, and familiarity composites.
- • Player ratings (OVR and component scores), derived deterministically from squad data.
Shipping during the tournament build-up
- · The live Elo and recent-form pipeline — current Elo figures are a placeholder band mapped from FIFA rank.
- · Rank-movement trend indicators, which stay blank until a real rank-history feed exists rather than showing invented arrows.
- · Last-five match form, intentionally left as a 'pre-tournament' placeholder because fabricating recent results would mislead.
Limitations and what we exclude
A prediction model is only useful if you know where its edges are. KickOracle deliberately leaves several things out, and there are real-world factors no pre-match model can capture. We would rather you treat our numbers as a well-reasoned estimate than as certainty.
- Probabilities are not predictions of certainty. A 60% favorite is still expected to lose around four times in ten, and a single-elimination knockout match compresses a season's worth of variance into ninety minutes.
- The model excludes in-match luck — deflected goals, woodwork, VAR overturns, red cards, and penalty-shootout randomness are not something a pre-match estimate can foresee.
- We do not use live market prices, do not weight social-media sentiment, and do not factor in referee assignments. Those are deliberate exclusions, not oversights.
- Some inputs are estimates derived from public information rather than measured ground truth, and squad data can change late when final lists are confirmed or injuries strike on the eve of a match.
How we track accuracy
A model that never grades itself is just an opinion with extra decimal places. KickOracle publishes a separate, public accuracy record that scores the prediction model retrospectively against completed tournaments — Euro 2024 and Copa América 2024 — using Brier score, top-1 accuracy, and log loss, the same metrics professional forecasting desks use to grade their own books.
That page also shows a calibration curve, the most honest accuracy check there is: when we say a team has a 60% chance, that group of predictions should win close to 60% of the time. We report it alongside FiveThirtyEight and devigged bookmaker closing odds so you can judge KickOracle against credible benchmarks rather than against our own marketing.
See the methodology in action
Now that you know how the model reasons, explore the live predictions and team ratings it produces.