This blog is the first in an ongoing series of blogs – the blog series will last the time of the Rugby World Cup 2019!
I love rugby and cannot wait for the Rugby World Cup to start. 4 hours to go!
As a product manager I am focusing on the Augmented Analytics capabilities of SAP Analytics Cloud.
Back from the summer vacation & fully energized, I thought about leveraging the power of Smart Predict to predict the game winners and hopefully top the RWC predictor game…
This was the starting point for this blog back in early September.
After a few sleepless nights gathering team statistics across the past years, writing fancy SQL and polishing the data and the predictive models, I am very happy to introduce you with my just-in-time predictions powered by SAP Analytics Cloud.
A few words on the first phase of the Rugby World Cup:
- There are 20 teams in total, divided in four pools of five teams each.
- The top two teams from each pool progress to the quarterfinals.
- This represents 40 games to predict.
Let’s focus on the predictions above and what they would mean for the pool results.
Pool A (Ireland, Scotland, Japan, Russia, Samoa)
Pool Winners: Ireland, Scotland
- Ireland is predicted to win its 4 pool games, including Scotland (93% chance) and Japan. Right now Ireland is the #1 nation according to World Rugby rankings and will likely move to the quarterfinals. (side-note: I do have a LOT of Irish colleagues and wishing them the very best for the team!). Jonathan Sexton, Ireland’s fly-half, is currently the best player in the world, according to World Rugby.
- Scotland is predicted to win 3 out of their pool 4 games beating Japan, Russia and Samoa. In my opinion beating Japan will be hard due to the incredible determination of the “Brave Blossoms” and the local fan support. Scotland is given a 77% chance of beating Japan.
- Japan is likely to beat Russia & Samoa but lose against both Ireland & Scotland. This would be a big disappointment for Japan as they would not go beyond the pool stage. (side-note: I am fond of this outsider team and I am secretly hoping Japan will create the surprise but the stats seems to be against them).
- Samoa would lose all games but the one against Russia.
- Russia would lose all pool games. They are one of the weakest nations in the tournament, their chances of winning are reduced.
Pool B (New Zealand, South Africa, Italy, Namibia, Canada)
Pool Winners: New Zealand, South Africa
- South Africa is predicted to win the 4 pool games and likely top the pool. This came as a surprise to me that the model predicts a victory against New Zealand. This being said the win probability is 57% for South Africa and the results of the very last games between these two nations were always super close. South Africa won the last Rugby Championship and this is a serious psychological advantage.
- New Zealand will face little significant opponents in pool B apart from South Africa and is predicted to beat the remaining 3 nations.
- Italy will be beaten by South Africa (with a probability of 93%) and New Zealand (81%). They should beat Canada & Namibia with not too much difficulties.
- Canada would beat only Namibia (with a probability of 77%).
- Namibia would end up being the last team in pool B. As per World Rugby rankings, Namibia is the weakest nation among the participating teams.
Pool C (England, France, Argentina, USA, Tonga)
Pool Winners: England, France
(I must confess this is the pool I was the most excited to predict as: 1. France is part of it 🙂 and 2. It’s sometimes called the “death pool” with 3 top rugby nations and only 2 that will receive a ticket for the next stage)
- England is predicted to win its 4 pool games with 66% chance against France and 92% against Argentina. Topping the pool C would put England in a great position for the rest of the competition.
- France aka “Les Bleus” would also make it to the quarterfinals (Cororico!). The key game is happening tomorrow (September 21), there they would beat Argentina, later on USA & Tonga and qualify as the second team in pool C.
- Argentina would stop at pool stage despite two victories on USA & Tonga.
- USA would beat Tonga. USA and Tonga would be the last two nations for Pool C.
Pool D (Australia, Wales, Georgia, Fiji, Uruguay)
Pool Winners: Wales, Australia
- Wales would top its pool by winning all 4 games. Their probability to beat Australia is estimated at 80% by the predictive model.
- Australia would be the second nation in pool D, due to their defeat against Wales and despite three victories on Georgia, Fiji and Uruguay.
- Fiji would end up in the third place after beating up both Uruguay & Georgia.
- Uruguay would make a surprising number 4 in the pool, as they are predicted to beat Georgia, however with 54% chance.
- Georgia would lose all 4 games and end up being last – however in both games against Fiji and Uruguay they have a 46% chance of winning.
Something I find interesting in the predicted pool results is that it corresponds to the top 8 nations as listed in the current World Rugby rankings. (You might have guessed that World Rugby rankings form an important part of the predictive model).
There is always uncertainty in sport (one of the reasons why I love it 🙂 ).
While predictive models can estimate the reality to a degree, there might be in some cases an X-factor that cannot be captured. As George Box famously said “all models are wrong, but some are useful”. It could be Japan getting the bonus point of being the hosting nation, or emerging players like Romain Ntamack taking the chance of their life…
I am excited to think about these 40 games (yum yum), fingers crossed for France!
I’ll cover in upcoming posts the step by step move from the initial idea to the predictions delivered in SAP Analytics Cloud:
If I can do it, I am sure you can do it too. I am hoping these blog series will make you want to use Smart Predict for your own questions!
Stay tuned in the coming weeks for more Rugby World Cup & Smart Predict!