Article ID: | iaor20132499 |
Volume: | 24 |
Issue: | 3 |
Start Page Number: | 301 |
End Page Number: | 309 |
Publication Date: | Jul 2013 |
Journal: | IMA Journal of Management Mathematics |
Authors: | Maher Mike |
Keywords: | forecasting: applications |
The Ryder Cup is played for between the golfers of Europe and the USA and consists of 8 foursomes, 8 fourballs and 12 singles matches played over 3 days. Each match is worth one point; hence, to win, a team must gain at least 14½ points. In live coverage of the event, considerable attention is paid to the predicted outcome but typically this is assessed by the crude method of assuming that any match a team is currently leading is worth one point and any match currently level is worth ½ point. No distinction is made between a match where a team is one up after one hole and one where it is eight up with nine to play. A Markovian model provides a more well‐informed prediction of the outcome, in terms of the expected number of points to be won by each team given the current score in each match. This takes account of estimates of the probabilities of holes being won or halved in the different formats of fourballs, foursomes and singles, based on historical data from the Ryder Cup and other major professional golf tournaments. As well as an expected final score, the model also provides a probability distribution of the event outcome and could be used to estimate the implicit odds of any outcome for ‘in‐play’ betting.