Who won the finals tipping competition?

Way back in February, before even a single game of the pre-season had been played, I asked The Arc readers to enter a competition to tip the finalists. The rules were simple: all you had to do was assign a probability, between 0 and 100%, to each team’s chance of making the finals. Easy! I haven’t got around to posting the results of the competition until now, because reasons, but by popular demand (ie. Greg Jericho keeps hassling me), I’m now getting around to it!

Every entrant has been ranked according to their Brier score. This score ranges from 0 to 1, with a lower score being better.  If you correctly picked all eight finalists, giving each of those teams a probability of 100 and all other teams a probability of 0, you’d receive a perfect Brier score of zero. If you assigned a 100 probability to eight teams that all missed the finals, you’d receive the worst possible score, one. All other scores fall somewhere between zero and one – the lower the better. For an explanation of how the scoring works, read this good Buzzfeed piece about the US election. But all you really need to know is that lower scores are better.

Now that I’ve mumbled through an explanation of the methodology…. the winner is… Sean and Cody from HPN! They got a Brier score of 0.1627, narrowly beating the second placed entrant Ben Phillips from ANU on a Brier of 0.165. Here’s the distribution of entrants’ Brier scores:

My own entries to the competition were based on simulating the season using my Elo ratings. At the start of one season, my ratings quite closely reflect teams’ final ratings from the season before, so the predictions aren’t that good; they get progressively much better as the season goes on. I’ll have another post on that soon, showing how the accuracy of my simulations changed over the season and how they compared to other forecasters.

The next graph shows what the competition entrants thought about each team. Richmond and Essendon were each given a low chance of making the finals by most entrants; the Bulldogs and Hawks were favoured to make the eight by most entrants.

The Bulldogs and Bombers were the teams that competition entrants were most wrong about – people didn’t rate the Bombers’ chances of making the finals, and did rate the Bulldogs.1 Brisbane was the team that entrants got the most right – almost no one gave them much of a chance of making the finals.

Here’s every entry in the competition, along with its Brier score (which I’ve multiplied by 100, so 0 is the best and 100 the worst possible score).2 Cool story!

  1. Measured as the absolute value of the result (1 = made finals, 0 = didn’t make finals) minus the median entrant’s probability (on a 0-1 scale).
  2. Some entrants didn’t enter probabilities that summed to 800 – in those cases, I scaled the entered probabilities so that they summed to 800.

Creator of The Arc