The 2014 Grand Final : When the Coin Flipped

The Sydney Swans were deserved pre-game favourites on Saturday according to most pundits (but not all - congratulations to Robert and Craig for tipping the winners). At some point during the course of their record-breaking loss that favouritism was handed to the Hawks. In this blog we'll investigate when.

Proud Hawks and rueful Swans fans might suggest that the rot started the moment the Swans lost the toss for choice of ends, and certainly the Swans looked off their game from very early in the contest, but the model I created a couple of weeks ago for estimating the chances of the home team in-running affords us a chance to arrive at a more impartial view of the crucial moment.

That model takes as one of its inputs the TAB Bookmaker's pre-game estimate of the Home team's victory chances on the Wednesday before the game, using the Risk-Equalising approach to map Head-to-Head prices to probabilities. On Wednesday the market was Sydney $1.58 / Hawthorn $2.45, which gives a pre-game probability estimate for the Swans of 1/1.58 + (1/1.58 + 1/2.45)/2 = 61.2%.

Plugging that value into the in-running model gives an initial assessment of the Swans' victory probability of 62.8%, the small ratcheting up of the probability being the model's way of adjusting for the TAB Bookmaker's historical and small Away team bias. Interestingly, on Betfair's Australian site the Swans were at 62.9% (Sydney $1.59 / Hawthorn $2.70) at the time of the first centre-bounce, which suggests that the in-running model's assessment was reasonable.

According to the model, the Swans' assessed victory probability reached its peak at the 9:22 mark of the 1st term when Dan Hannebery's behind took their lead to a game high 6 points. At that point they were 68.5% favourites.

Puopolo's goal about a minute and a half later drew the scores level and cut the Swans' probability back near to its pre-game level, before Franklin's reply boosted their chances above 68% again for the last time in the match.

The next six-and-a-half minutes saw Hawthorn kick 2.1 to the Swans 0.0, giving the Hawks a 7-point lead and causing the model to dramatically reassess the two teams' chances. Sydney were now only 52% prospects and the coin was poised to flip.

And flip it did less than a minute later with Gunston's goal stretching the Hawks' lead to 13 points and driving the Hawks into favouritism as far as the in-running model was concerned.

It was a status the Hawks never relinquished.

By Quarter Time the Hawks' lead was 20 points and the in-running model assessed the Swans as about 2/1 outsiders. On Betfair, though the market was thin, punters were giving the Swans a little more credit for their recuperative powers. They had the market at Sydney $2.18 / Hawthorn $1.81, which implies Sydney were about 45% chances. Though it's hard - probably impossible - to be objective in retrospect, that does seem like a generous assessment of the Swans' potential at that point.

Ten minutes later it was all but irrelevant anyway since by then the Hawks had established a 30-point lead and the Swans were about 16% chances. Five minutes after that, the lead was near 50 and the Swans' chances had more than halved.

Sydney rallied, or at least stemmed the bleeding, for the remainder of the term, going into the main break down by 7 goals and rated about 6% chances by the in-running model. Betfair punters had them as 16% chances (Sydney $6 / Hawthorn $1.17) but by that point there was so little liquidity in the Betfair market that you'd have been hard-pushed to get change for a $100 note.

The only interest provided by the 3rd Quarter was when the in-running model would rate the Swans as 100/1 chances (15:21 into the term with Matt Suckling's goal) and when 1,000/1 chances (late in the term with the Swans trailing by 9 goals, with the passage of time rather than a particular scoring event dragging the probability over that threshold).

At Three-Quarter Time the model had the Swans as about 2,000/1 chances and the Betfair market was little but the echoes of retreating footsteps and closing doors as punters left to look elsewhere for value and interest.

Hawthorn's first goal less than a minute into the final term pushed the Swans to 10,000/1 chances and, after that, it doesn't make much sense to attempt to measure how unlikely was a Swans' victory. Probabilities that small, at least in the sporting context, lack meaning.