The start of the AFL season brings with it one additional weekly commitment for thousands across Australia. As footy tipping competitions spring up across workplaces, in pubs and among groups of friends, many will be asking themselves:
Did I remember to enter my tips?
Predicting the outcome of each of the 198 home-and-away matches is no small task. However, in the age of internet forums and 24-hour news coverage, there’s never been more opinion written about who will win each game. But which of these sources are reliable? And which are simply unfounded speculation, cliché or guesswork?
I have examined a dataset of all home-and-away AFL games since 2010, and applied a few different tipping strategies to find which would have produced the best predictions. Some appear to be more reliable than others at a first glance. But which stand up to scrutiny against historical results?
The laziest method
The simplest system for generating tips is simply flipping a coin for each game.
Assuming you never tipped a draw, this would return on average around 49.6% of correct tips – exactly half of the matches that do not end up a draw.
There are clearly better methods to be used.
Home sweet home
Another simple system would simply be to tip the home team in each game, assuming there is some competitive advantage to home sides.
This is complicated somewhat by the number of teams sharing stadiums. The nine Melbourne teams share just two home grounds. In both South Australia and Western Australia, a single stadium is shared by two clubs. Nonetheless, each game has a designated home team, and this strategy involves tipping that side.
History suggests that this is superior to the naïve 50/50 choices, but not by as much as some people might think. Over the past six seasons, just over 55% of home teams have won.
Tip the leader
Another commonly employed method involves a simple glance at the ladder each week and tipping the side in the higher position. This method is widely used, especially among the sport’s less dedicated followers.
For simplicity’s sake, I have assumed Round 1 games are tipped according to a simple coin flip but all subsequent rounds are predicted by the ladder.
Unsurprisingly again, this method has delivered better results than either of the two previously discussed. Over the past six seasons, more than twice as many games have been won by the higher-ranked side. The success rate with this method was more than 70% in each of the 2011, 2012 and 2013 seasons.
Melbourne’s Herald Sun newspaper assembles a panel of experts to tip each round. This group of 28 (in 2015) or 29 (in 2014) consists primarily of AFL journalists, broadcasters and current and ex-players. It would be easy to argue that few people could predict results better than those who have played or write about the game for a living.
Across the board, though, the results are perhaps less impressive than you might think. In both years, the average expert called less than 63% of games correctly.
While some pundits did especially well in some years, their status as predictive geniuses is perhaps undermined when looking at both years together. 2015’s tipping champion was in the middle of the pack a year prior. 2014’s leader backed it up by dragging down the panel’s average in 2015.
Call on the statisticians
Since 1981, a computer program developed by Swinburne University’s Stephen Clarke has generated predictions for every AFL game. His work has been much discussed over the years and frequently claims to outperform most experts.
Although the exact formulae within the model have been tweaked over the years, it has always primarily relied upon two main factors – home-ground advantage and each team’s form, with heavy weighting on their most recent matches. From this, Clarke estimates that the home ground can be worth as much as a ten-point advantage in some games.
The tipping algorithm’s results are impressive. With more than 69% of correct tips in 2015 and more than 70% in 2014, it comfortably outperforms the panel of experts. Not one pundit was able to beat the computer in both seasons.
Bet on the bookies
Many tipsters will not enter their predictions without a quick glance at the latest bookmakers’ odds.
Relying on bookies’ prices is essentially a hybrid of several of these approaches, along with what statisticians sometimes call the wisdom of the crowd. Computer models and expert opinions inform the odds initially offered, but they frequently change in response to money wagered by the general public. If many punters start to back a team at a given price, their odds will soon shorten.
The simple strategy of believing that the bookmakers know best delivered better results than any other method I’ve discussed. More than 71% of “favourites” – the side with the shortest odds – in each of the past five seasons have won.
In cases where both teams were equally priced, I assumed a simple coin-flip scenario.
None of the other methods reached this benchmark in even a single season, apart from backing the higher placed side in the 2013 season. Even this was surpassed by the bookmakers method, which called almost three-quarters of tips correctly that year.
Tips for success
If you’re looking to win your colleagues’ and friends’ respect – and possibly their money – you can do a whole lot worse than simply tuning out of the countless hours of TV and radio analysis and averting your eyes from the numerous column inches in print.
The arguments made may sound informed and eloquent, but they are not guaranteed to prove correct. History suggests that you are more likely to out-think yourself and overlook the obvious.
When it comes to consistent tipping success, a history of premiership wins, All-Australian honours or thousands of hours of airtime don’t count for that much. Perhaps, betting slip in hand, the humble figure in your local TAB outlet could tell you more than the “experts”.
Authors: The Conversation Contributor