Australia v South Africa
What the record holds about this fixture, beside what the model says about it.
27 – 21
30 October 1999Allianz Stadium, London · neutral venue
What the model says about Australia v South Africa
- Kick-off
- 13:00 UTC
- Elevation
- 7 m
- Temperature
- 13.5 °C
- Rainfall
- 3.3 mm
- Wind
- 32 km/h
Daily means at the ground, from the ERA5 reanalysis via Open-Meteo.
Beside a name: points of margin the side has beaten its own rating by with him in the squad, against the same period without him. 10 of the 44 named here have enough matches on both sides to measure.
Australia
- 1Richard Harry
- 2Michael Foley
- 3Andrew Blades
- 4David Giffin+0.8
- 5John Eales(c)
- 6Matt Cockbain
- 7David Wilson
- 8Toutai Kefu
- 9George Gregan
- 10Stephen Larkham
- 11Joe Roff-6.5
- 12Tim Horan
- 13Dan Herbert
- 14Ben Tune-0.3
- 15Matt Burke
Replacements
- 16Jason Little
- 17Nathan Grey+10.3
- 18Chris Whitaker
- 19Mark Connors
- 20Owen Finegan
- 21Rod Moore
- 22Jeremy Paul+10.0
South Africa
- 1Os du Randt+3.9
- 2Naka Drotske
- 3Cobus Visagie
- 4Krynauw Otto
- 5Mark Andrews
- 6Johan Erasmus
- 7Andre Venter
- 8Bobby Skinstad+2.3
- 9Joost van der Westhuizen(c)
- 10Jannie de Beer
- 11Pieter Rossouw
- 12Pieter Muller
- 13Robbie Fleck
- 14Deon Kayser
- 15Percy Montgomery+0.2
Replacements
- 16Stefan Terblanche-3.6
- 17Henry Honiball
- 18Andre Vos
- 19Albert van den Berg-6.0
- 20Ollie le Roux
- 21Werner Swanepoel
- 22Chris Rossouw
Measured against the side's point-in-time rating, so opposition and venue are already removed, and over the years between a player's first and last cap rather than his side's whole history. None of the 778 players assessed has a gap larger than chance explains once the number tested is corrected for — a rating already contains the players who usually play for a side, so what a team sheet can add to it is small. Team selection is not an input to the forecast on this page. The full table and the method.
Context beside the forecast, never inside it. Weather, team selection and match statistics are not model inputs, so a published probability does not move because any of them changed.