3 ms·
Their prediction numbers seem a little off to someone with a passing knowledge of rugby but not enough statistics knowledge to understand the analysis. eg NZ h
by antod 11y ago
Their prediction numbers seem a little off to someone with a passing knowledge of rugby but not enough statistics knowledge to understand the analysis.
eg NZ having a higher probability of beating Argentina, than NZ has of beating Namibia (or Georgia and Tonga too). Argentina is a much better team with a higher world ranking than Namibia.
The numbers were probably skewed by NZ not playing against Namibia very often (or even at all?), but they've played Argentina relatively often in the past.
- rangibaby 11y agoTheir data should have included RWC matches or tiers or both. Certain teams look quite good on paper; Japan have won about half their games (ever) but only won one match in the RWC (ever) before their "biggest-ever upset" victory against South Africa. I guess this proves the importance of making a good estimate. If something is obviously off with your result, it's worth checking your model.