NRL Predictions for Round 9
Team Ratings for Round 9
The basic method is described on my Department home page.
Here are the team ratings prior to this week’s games, along with the ratings at the start of the season.
| Current Rating | Rating at Season Start | Difference | |
|---|---|---|---|
| Storm | 9.39 | 9.29 | 0.10 |
| Panthers | 5.79 | 8.50 | -2.70 |
| Sharks | 4.90 | 5.10 | -0.20 |
| Roosters | 3.99 | 7.44 | -3.50 |
| Sea Eagles | 3.48 | 2.97 | 0.50 |
| Cowboys | 3.46 | 4.11 | -0.70 |
| Bulldogs | 2.76 | 0.07 | 2.70 |
| Broncos | 1.24 | -1.82 | 3.10 |
| Raiders | -0.40 | -3.61 | 3.20 |
| Dolphins | -0.83 | -1.96 | 1.10 |
| Warriors | -1.34 | -1.68 | 0.30 |
| Rabbitohs | -3.71 | -4.35 | 0.60 |
| Dragons | -3.75 | -4.55 | 0.80 |
| Knights | -3.82 | -0.05 | -3.80 |
| Eels | -6.47 | -3.02 | -3.50 |
| Titans | -6.99 | -5.50 | -1.50 |
| Wests Tigers | -7.71 | -10.97 | 3.30 |
Performance So Far
So far there have been 64 matches played, 38 of which were correctly predicted, a success rate of 59.4%.
Here are the predictions for last week’s games.
| Game | Date | Score | Prediction | Correct | |
|---|---|---|---|---|---|
| 1 | Broncos vs. Bulldogs | Apr 24 | 42 – 18 | -0.60 | FALSE |
| 2 | Roosters vs. Dragons | Apr 25 | 46 – 18 | 9.10 | TRUE |
| 3 | Warriors vs. Knights | Apr 25 | 26 – 12 | 5.10 | TRUE |
| 4 | Storm vs. Rabbitohs | Apr 25 | 24 – 16 | 17.00 | TRUE |
| 5 | Cowboys vs. Titans | Apr 26 | 50 – 18 | 11.70 | TRUE |
| 6 | Panthers vs. Sea Eagles | Apr 26 | 10 – 26 | 7.30 | FALSE |
| 7 | Raiders vs. Dolphins | Apr 27 | 40 – 28 | 2.50 | TRUE |
| 8 | Wests Tigers vs. Sharks | Apr 27 | 20 – 18 | -10.80 | FALSE |
Predictions for Round 9
Here are the predictions for Round 9. The prediction is my estimated expected points difference with a positive margin being a win to the home team, and a negative margin a win to the away team.
| Game | Date | Winner | Prediction | |
|---|---|---|---|---|
| 1 | Sharks vs. Eels | May 02 | Sharks | 11.40 |
| 2 | Roosters vs. Dolphins | May 02 | Roosters | 1.80 |
| 3 | Rabbitohs vs. Knights | May 03 | Rabbitohs | 0.10 |
| 4 | Warriors vs. Cowboys | May 03 | Cowboys | -4.80 |
| 5 | Wests Tigers vs. Dragons | May 03 | Dragons | -4.00 |
| 6 | Titans vs. Bulldogs | May 04 | Bulldogs | -9.70 |
| 7 | Panthers vs. Broncos | May 04 | Panthers | 1.50 |
| 8 | Storm vs. Raiders | May 04 | Storm | 9.80 |
David Scott obtained a BA and PhD from the Australian National University and then commenced his university teaching career at La Trobe University in 1972. He has taught at La Trobe University, the University of Sheffield, Bond University and Colorado State University, joining the University of Auckland, based at Tamaki Campus, in mid-1995. He has been Head of Department at La Trobe University, Acting Dean and Associate Dean (Academic) at Bond University, and Associate Director of the Centre for Quality Management and Data Analysis at Bond University with responsibility for Short Courses. He was Head of the Department of Statistics in 2000, and is a past President of the New Zealand Statistical Assocation. See all posts by David Scott »
Hi there, for your model’s predictions (namely the Super Rugby, NRL, and AFL), is it possible to see the standard deviations for the predicted scores anywhere or is this information that you don’t make available?
Thanks,
Jay
2 months ago
If you follow the link to the method provided at the top of the post, you will see that there can’t be a standard deviation for the predicted score difference. The prediction uses the difference in ratings with a home advantage adjustment (which used to include an is_a_different_country to the away team component). I suppose that the rating could have something like a standard deviation because of the smoothing algorithm but that should be in the ratings table rather than the game predictions.
2 months ago
I don’t often reply to or even publish many comments but this one is interesting.
Yes, I am able to provide standard deviations for the predicted scores.
However, I don’t think they would be useful for what I think are your purposes, namely to provide prediction intervals for the predicted score differences (margins) and/or probabilities for the score differences lying in certain ranges.
The major reason for my being of that opinion is that the distribution of scores around the predicted central value is not normal. Instead the distribution has heavy tails: essentially the winning team has a tendency to win by a lot.
As it happens I have a fair amount of experience in dealing with heavy-tailed distributions, to the extent of having a number of R packages dealing with distributions having those characteristics. (Try googling GeneralizedHyperbolic for one example.)
The distributions I am familiar with can provide good fits to the error distributions for the score differences. Again there is an however. My suspicion is that any prediction intervals obtained by this process will be subject to high variability. I would have to do substantial analysis before I would put my name to any results of this kind.
My reasons for producing and publishing these sports forecasts are firstly didactic:
1. Simple statistical techniques based on limited data can provide useful estimates and have the benefit of not being subjective personal opinion.
2. Randomness is real and needs to be taken into account: sometimes things work out and predictions seem to work well, but at other times they don’t.
Aside from these didactic purposes, being a statistics and computing nerd, I enjoy the work of producing forecasts in an efficient manner.
1 month ago