September 28, 2015

Rugby World Cup Predictions, 29 September 2015 to 03 October 2015

It is a little difficult to choose a period for the predictions. I have predicted up to October 3 because Tonga plays a further game on October 4, and I want to have updated their rating based on the result of the Tonga-Namibia game before predicting their game against Argentina.

Team Ratings at 29 September

The basic method is described on my Department home page.

Here are the team ratings prior to 29 September along with the ratings at the start of the Rugby World Cup.

Rating at 29 September Rating at RWC Start Difference
New Zealand 27.38 29.01 -1.60
South Africa 21.98 22.73 -0.70
Australia 20.47 20.36 0.10
Ireland 17.72 17.48 0.20
England 17.29 18.51 -1.20
Wales 14.10 13.93 0.20
France 11.53 11.70 -0.20
Argentina 9.03 7.38 1.60
Scotland 6.00 4.84 1.20
Fiji -3.23 -4.23 1.00
Samoa -3.54 -2.28 -1.30
Italy -6.93 -5.86 -1.10
Tonga -7.23 -6.31 -0.90
Japan -10.61 -11.18 0.60
USA -15.69 -15.97 0.30
Georgia -17.57 -17.48 -0.10
Canada -18.06 -18.06 -0.00
Romania -20.20 -21.20 1.00
Uruguay -31.10 -31.04 -0.10
Namibia -34.63 -35.62 1.00

 

Performance So Far

So far there have been 19 matches played, 16 of which were correctly predicted, a success rate of 84.2%.
Here are the predictions for previous games.

Game Date Score Prediction Correct
1 England vs. Fiji Sep 18 35 – 11 29.20 TRUE
2 Tonga vs. Georgia Sep 19 10 – 17 11.20 FALSE
3 Ireland vs. Canada Sep 19 50 – 7 35.50 TRUE
4 South Africa vs. Japan Sep 19 32 – 34 33.90 FALSE
5 France vs. Italy Sep 19 32 – 10 17.60 TRUE
6 Samoa vs. USA Sep 20 25 – 16 13.70 TRUE
7 Wales vs. Uruguay Sep 20 54 – 9 51.50 TRUE
8 New Zealand vs. Argentina Sep 20 26 – 16 21.60 TRUE
9 Scotland vs. Japan Sep 23 45 – 10 14.40 TRUE
10 Australia vs. Fiji Sep 23 28 – 13 24.10 TRUE
11 France vs. Romania Sep 23 38 – 11 33.30 TRUE
12 New Zealand vs. Namibia Sep 24 58 – 14 64.00 TRUE
13 Argentina vs. Georgia Sep 25 54 – 9 24.60 TRUE
14 Italy vs. Canada Sep 26 23 – 18 12.50 TRUE
15 South Africa vs. Samoa Sep 26 46 – 6 23.90 TRUE
16 England vs. Wales Sep 26 25 – 28 11.20 FALSE
17 Australia vs. Uruguay Sep 27 65 – 3 50.30 TRUE
18 Scotland vs. USA Sep 27 39 – 16 21.40 TRUE
19 Ireland vs. Romania Sep 27 44 – 10 38.80 TRUE

 

Predictions for 29 September to 03 October

The prediction is my estimated expected points difference with a positive margin being a win to the first-named team, and a negative margin a win to the second-named team.

Game Date Winner Prediction
1 Tonga vs. Namibia Sep 29 Tonga 27.40
2 Wales vs. Fiji Oct 01 Wales 23.80
3 France vs. Canada Oct 01 France 29.60
4 New Zealand vs. Georgia Oct 02 New Zealand 44.90
5 Samoa vs. Japan Oct 03 Samoa 7.10
6 South Africa vs. Scotland Oct 03 South Africa 16.00
7 England vs. Australia Oct 03 England 3.30

 

avatar

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 »

Comments

  • avatar

    Hi there, prof Scott!

    I stumbled upon your predictions for the RWC2015 Week 1, and has since been a true believer!

    I would like to use your method in other rugby competitions, but the details you provide on your homepage lack some specifics – especially for somebody who had Stats 114 and Stats 144 at university 20 years ago… Is it possible you can explain your method in more detail, please?

    Thanks in advance, and keep those predictions coming! I don’t know if you play SuperBru (www.superbru.com), but I definitely know you’ll kick serious butt if you do! (Since I’ve been using your predictions, I’ve gone from “Top 85%” to “Top 5%” on average in the groups I play in!)

    Best regards,

    Jaco
    (Wellington, South Africa)

    9 years ago

    • avatar

      I believe some of my other readers go in for SuperBru. There might be diminishing returns if more players use my predictions as a guide.

      As for the method, you need to bone up on exponential smoothing. But you start with all teams with equal rating some years before you want to make a prediction (hence you need a lot of past data). Then for each game you predict using ratings and a home ground advantage. After you get the result you update the ratings of each team by taking a fraction of the difference between your prediction and the actual (that difference is the error, hence error-correction exponential smoothing). Do that for many, many games, and for different values of the home ground advantage and the fraction used in the updating. Choose the best values of the home ground advantage and the fraction.

      There are some tweaks to that but that is all it is. There is art in choosing the parameters and some additional parameters involved in the tweaks that I just mentioned, and there is computing skill needed to do things efficiently so I can predict four different competitions at the moment without spending a lot of time on it.

      9 years ago

  • avatar
    Steven Bott

    David,

    Unless I am mistaken, your success rate is not accurate.

    For example;

    France v Romania – prediction 33.3 – result 27

    NZ v Namibia – prediction 64 – result 44

    Arg v Georgia – prediction 24.6 – result 45

    And the list goes on.

    I may have misunderstood the prediction concept.

    Please clarify.

    Thanks

    9 years ago

    • avatar
      Thomas Lumley

      I think you’ve misunderstood the summary of results: a prediction is counted as correct if the result is in the same direction as the prediction.

      9 years ago

    • avatar

      As Thomas said, the accuracy relates only to whether I predicted the winner correctly. Predicting margins, even within a few points would be ridiculously hard.

      9 years ago

    • avatar

      I agree with the two gentlemen. Also remember that there is absolutely no way you can account statistically for ever-changing team selections, injuries, one bad bounce of the ball, a bad decision by a captain (Eng vs Wales comes to mind!) and many other factors that can influence the outcome and/or scores.

      All three outcomes that were “wrong” up to this point were seen as massive upsets by the rugby world, and if you look at those games, the factors I mentioned above all played a role. Except with the SA v JAPAN game – those Springboks were just poor! And yeah, I’m a Bok supporter.

      9 years ago

      • avatar
        Mark Dalton

        Jaco, I don’t agree that Wales over England was a massive upset. As a former UK resident and hence follower of the intense rivalry between those ‘home nation’ teams, it was always on the cards. I have to say, that I did pick Wales by 7. Cheers, and good picking.

        9 years ago

        • avatar

          Mark, I hear what you’re saying. The thing with the Six Nations competition is that any of those teams can win on the day (even Italy has provided an upset or two in the past).

          But on this day, with all the things going in favour of England (Twickenham, England RWC2015, Six Nations results, injured Wales players, etc.) the expectation surely was that England only needed to show up for 80 minutes. And it almost worked for them – Wales scored a try in the final ten from what could easily be described as a lucky bounce of the ball, England went for a try instead of kicking for goal (and failed in getting any points), and Wales slotted a penalty kick from a rather difficult field position.

          On SuperBru, only 16% of the Brus predicted that outcome – a major upset, then.

          Now, Prof Scott is predicting that England will beat Australia. From where I’m sitting and based on current form, that might also be seen as a major upset. This might just be the game of the tournament! I’ve made all my picks for this round, but this is the one that I’m not sure about. There are just too many factors that can screw things up! ;-)

          9 years ago

  • avatar
    Xavier Wing

    Hi David,

    I stumbled upon your predictions and wanted to start learning how to make my own predictions statistically. What would I need to acquire knowledge in? I have read your basic method but still am still confused as I dont come from a maths background.

    9 years ago

    • avatar

      The method is simply the error-correction form of exponential smoothing, plus home ground advantage. You need a basic understanding of time series forecasting and the ability to carry out repetitive numerical calculation. It could be done in Excel, but that would be tedious and error-prone. I program in R.

      You also need to be able to assemble years of data to initialise ratings and to enable appropriate choice of parameters. That usually requires data munging ability because entering data by hand is also tedious and error-prone.

      9 years ago

  • avatar

    Any predictions for Sundays games.
    4th October

    9 years ago

  • avatar

    Hi David, what is your prediction for tonights game Canada Vs Romania? Cheers

    9 years ago

  • avatar
    David Vin

    The question is are the All Blacks holding back ?

    9 years ago