Kindy Analytics 2026 FIFA World Cup is an independent data science project exploring how statistical modelling and machine learning can be used to forecast international football.
The project combines football analytics, tournament simulation, and predictive modelling to estimate match outcomes, progression probabilities, and World Cup champion chances throughout the tournament.
Rather than aiming to predict every match perfectly, the goal is to build a transparent forecasting system, evaluate its performance in real time, and openly share both the successes and the mistakes.
About Kindy Analytics
Independent, transparent, and tournament-focused
Kindy Analytics 2026 FIFA World Cup is an independent data science project exploring how statistical modelling and machine learning can be used to forecast international football.
The project combines football analytics, tournament simulation, and predictive modelling to estimate match outcomes, progression probabilities, and World Cup champion chances throughout the tournament.
Rather than aiming to predict every match perfectly, the goal is to build a transparent forecasting system, evaluate its performance in real time, and openly share both the successes and the mistakes.
Why I Built This
A football project with public accountability
I'm a data scientist and lifelong football fan who created this project to combine two passions: international football and predictive analytics.
Major tournaments provide an ideal environment for testing forecasting models because predictions can be published before matches are played and objectively evaluated as results unfold.
Instead of keeping the model private, I chose to make the forecasts, methodology, and evaluation publicly available so others can explore how the model evolves throughout the tournament.
Every prediction shown on this website is generated before the outcome is known and preserved for later evaluation.
How the Forecast Works
A four-step forecasting pipeline
The forecasting pipeline consists of four stages.
01
Estimate Match Probabilities
The model estimates the probability of a home win, draw, and away win for every fixture using historical football data, engineered features, team-strength metrics, and tournament context.
02
Tournament Simulation
Those match probabilities are used to simulate the tournament thousands of times.
- progression probabilities for every remaining round
- champion probabilities
- expected tournament paths
- likely future matchups
03
Forecast Updates
After completed matches, the latest results are incorporated into the published forecast.
- fewer possible tournament paths remain
- team strength evolves
- new information becomes available
- uncertainty decreases
The dashboard aims to adapt as the tournament progresses. That is as each tournament stage concludes, the dashboard shifts from forecasting that stage to evaluating it.
04
Publish Forecasts
Once a forecast refresh completes, updated probabilities are published across the dashboard.
- Original historical predictions are never overwritten and remain available for evaluation.
Estimated Scorelines
Illustrative outcomes, not the primary prediction target
The model is designed to estimate match outcome probabilities (win, draw, or loss), not exact scorelines. Estimated scores are derived from the underlying probability distribution and represent one plausible outcome rather than a precise prediction.
Football is a low-scoring sport where many nearby scorelines often have similar likelihoods. As a result, the estimated score should be interpreted as an illustrative forecast, while the win, draw, and loss probabilities are the model's primary outputs.
How Performance Is Evaluated
Forecasts are scored against the predictions that were actually published
Forecasts are evaluated continuously throughout the tournament using the original prediction that existed before each match or stage began.
The dashboard tracks several measures of forecasting performance, including:
Match prediction accuracyBrier ScoreLog LossQualification forecasting accuracyTournament progression forecastsChampion probability evolution Historical forecasts are preserved rather than replaced, making it possible to compare what the model predicted with what actually happened.
What the Dashboard Shows
The public interface is designed to explain forecast change over time
Throughout the tournament the dashboard provides:
Match predictionsChampion probabilitiesKnockout progression probabilitiesTournament simulationsForecast evaluationHistorical forecast changesModel-versus-market comparisons These views are designed to help explain how the forecast changes over time, not simply to display a single prediction.
Transparency & Limitations
Public beta means the strengths and mistakes stay visible
Kindy Analytics is an independent public beta project.
Forecasts are probabilistic estimates, not guarantees.
Some provider data may occasionally lag during live matches, and published forecasts reflect the latest completed forecast refresh rather than continuous real-time recomputation.
The model will sometimes be wrong. That is expected.
The objective of this project is not perfect prediction, but building a forecasting system whose strengths and weaknesses can be measured openly throughout the tournament.
Kindy Analytics is not affiliated with, endorsed by, or associated with FIFA.
View the full dashboard
Expanded FAQ
Common questions about the project and the model
These answers mirror the same public-beta principles used across the dashboard: preserve the original forecasts, explain how probabilities are generated, and be clear about what the model can and cannot do.
How are champion probabilities calculated?
Champion probabilities come from thousands of tournament simulations using the latest match probabilities and tournament state. The reported percentages represent how often each team wins the tournament across those simulations.
Why do forecasts change during the tournament?
Forecasts update after completed refreshes as match results, tournament progression, and team-strength context evolve. As uncertainty decreases later in the tournament, probabilities naturally become more concentrated.
Why compare the model with betting markets?
Reference market probabilities provide an independent benchmark because bookmakers aggregate vast amounts of football information into implied probabilities. The model does not copy those prices. Instead, market probabilities serve as one input among many and provide useful context for comparing the model's conclusions with the broader consensus.
What does confidence mean?
Confidence reflects how consistently the model's simulations support a predicted outcome. It is intended to help interpret the forecast, not to guarantee that a result will occur.
Can I see how previous forecasts performed?
Yes. Historical predictions are preserved throughout the tournament so the dashboard can evaluate both match forecasts and tournament predictions using the forecasts that were actually published before the results were known.
Is this an official FIFA forecast?
No. Kindy Analytics is an independent research project created for educational and analytical purposes. It is not affiliated with FIFA or any tournament organizer.