The final league round of the tournament was a good one for our model, getting 5 out of 7 correct, putting us just inside the top 30% of SuperBru pundits. This weekend sees the start of the short knockout phase of the tournament. There’s probably a totally different dynamic to these games, but we’ll stick with our model for the last few weeks before taking a look back and seeing where the model did well and where it fell short.

Here are the results for the final round of the competition, followed by our predictions for this weekend’s two big games.

Another week of slipping down the leaderboard for our model, getting 4 out of 7 correct last weekend when many on Superbru got the Hurricanes/Highlanders result right too. As usual when things don’t go our way we’ll try distract you with a graphic. It seemed to us that winning margins have been getting bigger in recent weeks, so here’s a plot of the average winning margin per round over the competition. Blue lines indicate lines of best fit for the first 8 round and next 9 rounds. We’ve cherry picked our split at round 8 for maximum effect, but it looks like there could be something there: let’s see what happens this weekend.

Here are the outcomes for last weekend followed by predictions for the weekend ahead: