The Swinburne Computer has been predicting match winners, margins, odds, and final ladder order since 1981. Although its predictions are published early in the week, before team selections are made, this paper shows the computer’s performance compares favourably with that of media experts and the public. This suggests that far from making predictive use of extra information, humans do not individually process correctly the publicly available information. Combining their forecasts may produce better predictions. Evidence suggests the computer’s predictions can be used to exploit market inefficiencies, not only early in the week when prices are first set, but also late in the week when prices have adjusted following team selections.