# How Our Rugby League World Cup 2026 Model Works

Alphr predicts every game of the 2026 Rugby League World Cup by rating each player in the named 17s from his club football and Tests on one scale, then adding each nation's international record and home ground. In testing on games it had not seen, it picked 74% of winners between the top seven nations. No bookmaker prices are used anywhere in the model.

## What data does the World Cup model use?

Official match-centre stats from the NRL, the NSW Cup and Queensland Cup (2021 to 2026, about 1,760 games), Super League (2015 onwards) and full internationals: Tests since 2015, the Pacific Championships from 2023 to 2025, and the 2008, 2013, 2017 and 2021 World Cups. 224 of the 230 players in the ten 23-man squads have recorded games in that data.

## How can it compare a Super League player with an NRL player?

Each game becomes one production score from the stats every competition records (tries, try assists, line breaks, tackle busts, metres, offloads, tackles, errors and penalties), compared with players in the same position in the same league and season. Each league is then placed on the NRL's scale using players who have played in two leagues. On that scale an average NSW Cup player sits about 0.4 standard deviations below an average NRL player, a Queensland Cup player about 0.5 below and a Super League player about 0.4 below. A player's rating weights his most recent games most, over the last 18 months.

## What else goes into a prediction?

A nation's strength is the sum of its 17 players' ratings, with the bench at half weight. An international rating built from every Test, Pacific Championships and World Cup result since 2008 adds what club form misses, and a venue term covers home ground, worth about eight points in international rugby league. The model uses no bookmaker prices.

## How accurate is it?

Tested out of sample on the 77 games between Australia, New Zealand, England, Tonga, Samoa, Papua New Guinea and Fiji from 2015 to 2025, with each season predicted by a model fitted on the other seasons, it picked 74% of winners with an average error of 13.7 points on the margin.

## When are tips published, and can they change?

A game's tip is published once both teams have named their 17 and is never edited afterwards. It is graded at full time, and the record is public. Before team lists are named, the tournament outlook rates each nation's best 17 from its 23-man squad.

## How are the title chances worked out?

The rest of the tournament is simulated 20,000 times with the model's margins, the results so far and the 2026 format: the top two in Pool A and the top two in the combined Pool B and C table reach the semi-finals. The outlook refreshes after every result.

## See it in action

- Predictions for every game, the pool tables and who will win: https://alphr.com.au/rugby-league-world-cup
- Teams and power rankings: https://alphr.com.au/rugby-league-world-cup/teams
- Verified accuracy across every sport: https://alphr.com.au/accuracy