One of the most popular and fun markets to bet on throughout the AFL season is undoubtedly the first goal scorer market. Even though it is essentially a market designed for less serious betting, especially due to the number of players on field being relatively large for the AFL, is there too much randomness in the result or can some research provide an insight into how to be smarter on this ‘novelty’ market?

There seem to be two key factors in the way bookmakers shape their markets, in that forwards tend to be allocated lower odds than midfielders, with key forwards generally being lower odds than general forwards and that the odds are scaled lower relative to their teams odds of winning. For example last season GWS players would have had, all else equal, lower odds to kick the first goal against Essendon than against Sydney as the bookmakers make the connection that the team whom is considered better (as defined by their odds of winning the match), will have a higher chance of kicking the first goal.

So these two factors are the ones that will be analysed. Firstly what are the chances of a player given their position on ground of kicking the first goal and how can these chances be used to assess potentially betting on the market? Secondly, is there a relationship between a team’s pre-game betting odds and their likelihood of kicking the first goal?

In assessing the first factor, the last two season’s first goal scorer for each match were recorded and their position listed for 413 matches. This gave the following results chart.

table2

The positioning was determined through my own judgement, thus a few caveats must be placed on the results. The number of midfielders who kicked the first goal is likely overstated, this is because some players considered midfielders occasionally start up forward and therefore are mislabeled. General forwards are slightly understated due to the previously mentioned midfielder starting forward issue. Key forwards are likely understated due to the fact ruckmen would have a subset that would have started inside 50 due to their team selecting two ruckmen.

Thus the above table can be used as a rough guide as to what players represent value in the first goal scorer market. If a player whom will be playing key forward is priced clearly above $12.78, it is likely going to be a bet with a positive expected return. For example from this week’s matches, Jonathon Griffin will be playing up forward to begin the match, likely very close to goal due to Aaron Sandilands being almost certain to contest the ruck at the start of the game. His odds to kick the first goal are as high as $26 at some bookmakers, clearly above the key forward fair price per player calculated for the past two seasons. Similar logic can be applied to Brett Eddy ($23), Braydon Preuss ($41), Dan McStay ($25 and surprising has kicked the first goal of the match 3 times the past two seasons) and Troy Menzel ($34).

Now for the second factor, to attempt to test the hypothesis that a team’s chances of kicking the first goal is linked to their pre-game odds (as a proxy for a team’s relative expected performance), I compiled the betting odds for each of the last two season’s games. With this data I compiled the below table.

table1

As can be seen, there was negligible difference in the first four bands of teams, with all hovering between 55% to 60%. This leads toward the conclusion that the more fancied team tends to have a slightly higher chance of kicking the first goal, regardless of their exact chances. That is that all that seems to matter is just that they are indeed the favourite with the magnitude of their favouritism being irrelevant. However the last column with a percentage chance of 43% is not consistent with that logic. It could be that there just isn’t a large enough sample size to approximate the true percentage chance appropriately or that the betting odds when teams are considered that closely matched become less relevant. Regardless 39 matches in this band is simply too low to attempt to draw any conclusions from, whilst the bands with 74 and 76, although consistent with the two lower odds bands also likely need another season or two of data to really be able to draw solid conclusions from.

Despite this it seems quite likely there is very little relationship between the magnitude of a team’s betting odds and their chances of kicking the first goal. Thus given bookmakers do intend implement this form of scaling in their markets, tending toward choosing players whom are playing on a team who is heavy underdog would also seem to be a smart idea. This is assuming the bookmakers scaling toward the more fancied team kicking the first goal is 58% or higher, which especially for matches between teams with big quality differences (For example GWS and Essendon last season) is probably true.

Another way to look at this is the scatterplot between a team’s match win rate and first goal scorer win rate. This shows a correlation of 0.51, implying a decent strength, positive, linear association between these two variables. Thereby it is stating that approximately 25% of the variation in first goal scored can be explained by their win rate, thus there seems to definitely be an association between the two, however other factors seem to be much more important.

graph 1

So even though the first goal scorer market is indeed one in which tends to thought of as a ‘novelty’ form of betting, there is definitely opportunity to bet and profit in the long run by following a few basic principles. Although it is clearly a very volatile and relatively unpredictable market, so it is in my opinion best for everyone’s sanity to not take it too seriously and just use these statistics as a way to select players who aren’t terribly under the odds when feeling like a flutter!

Advertisements