Matter of Stats

View Original

Leading and Winning in AFL

One of the bets that's offered by TAB Sportsbet is on which of the teams will be the first to score 25 points. After analysing scoring event data for the period 2008 to 2014 provided by Paul from afltables.com I was surprised to discover that the first team to score 25 points goes on to win the game over 70% of the time.

Home teams fare even better than this, winning over three-quarters of the time that they're first to 25 points, a feat they achieve about 56% of the time. In the other 44% of games, the Away team beats them to 25 points, these Away teams emerging victorious 64% of the time.

Whilst the absolute winning percentage for Home teams reaching 25 points is higher than for Away teams, the proportionate increase in Home teams' success relative to their base rate of winning, which is just over 58%, is actually significantly lower. The relative rate of victory for Home teams is 76.2%/58.3%, which is 1.3, while for Away teams it's 64.3%/41.7%, which is 1.5. In words, getting to 25 points first does more to increase the chances, proportionately, of an average Away team than it does to increase the chances of an average Home team. 

The choice of a 25 point finish line by the TAB is somewhat arbitrary. At right is a table summarising the results for the 25 point and for other thresholds. A few macro features are apparent:

  • as we increase the threshold, its significance as a correlate of victory also increases. For example, while the first team to score 10 points wins about 64% of the time, the first to score 30 points wins almost 75% of the time
  • analysing the implications for Home teams and for Away teams separately reveals that this same trend applies to both - a Home team that's first to 50 points wins more often than a Home team that's first to 25 points, for example
  • the separate analyses also show that the absolute probability of victory is higher for Home teams than for Away teams at any threshold. Home teams that are the first to score 40 points, for example, win more often than Away teams that are first to score 40 points

What's not immediately obvious from this chart, but something that can readily be calculated, is that the proportionate increase in Away teams' chances from being the first to score some threshold level of points is always greater than Home teams' chances at the same threshold. 

Earlier, we calculated the results for the 25 point threshold and found lift ratios of 1.3 for Home teams and 1.5 for Away teams. Using the data from the table above we can perform the same calculation for the 50 point threshold and estimate lifts of 86.2%/58.3%, which is 1.5, for Home teams, and 78.3%/41.7%, which is 1.9 for Away teams. Similar calculations can be performed for the other thresholds from which we can derive a general principle that Away teams' chances benefit more, proportionately, from being the first to some threshold score than do Home teams'.

Being the first team to attain some threshold score obviously boosts a team's chances of victory, but it seems logical that the extent to which it does this might be related to the lead that it had when it did so, with bigger leads conferring greater advantage.

That is, indeed, what the data suggests.

Teams that reach 25 points first and with a two goal or greater lead go on to win 78.5% of the time, which is about 16% more often than teams which attain that threshold score but with only a 6 to 11 point lead. Again, the analysis holds equally when we conduct it separately for Home teams and then for Away teams. Home teams that are the first to 25 points with a two goal or greater lead win 83% of the time, while those that lead only by 6 to 11 points win 66% of the time. For Away teams the equivalent percentages are 72% and 59%.

The table at right provides similar percentages for each of the thresholds we considered earlier, and also provides winning rates for teams attaining a particular threshold with only a 1 to 5 point lead. Some caution should be exercised in interpreting these latter figures because most are based on much smaller numbers of games. The results for Home teams being first to 50 points but with only a 1 to 5 point lead are, for example, based on only 42 games. Some of the results for higher thresholds and for leads of 6 to 11 points should also be treated with similar caution for the same reasons.

What this analysis ignores, of course, is the relative merits of the Home and the Away teams in a given contest. A 12 point lead held by a significantly weaker Away team might not be as detrimental to the Home team's chances as a similar lead surrendered to a stronger opponent. This is an aspect I plan to address in a subsequent blog.