As the tournament goes on, it has become more apparent that batting first seems to be the way to go. Regardless of any lingering overhead conditions, teams batting first were generally able to enjoy 20-25 overs on a good wicket before it started to slow down. By this point they usually had plenty of runs on the board, and the chasing team often struggled to keep up with the run rate on an already slow pitch. In total, 27 out of the 41 group games - 66% - were won by the team batting first.
For the first half of the group stages, many of the teams were electing to field first - the winner of the toss elected to field first in 71% of the first 28 games, but these teams only managed to win 40% of these games - an early indicator that batting first may have been the way to go. The turning point in this regard came in game 28, England vs Australia. England, who had elected to field, struggled to take early wickets and ultimately lost the game by a comfortable margin. It appears at this point that the majority of captains had taken note of the trends in how the wickets were playing, with only 2 out of the remaining 13 toss-winners electing to field first for the remainder of the tournament.
So how big of an advantage was batting first? Luckily we have just the tool to quantify this. The Bradley-Terry model, which we use for our ranking system, can be modified to include an "order effect". This order effect can be seen as any external factor that might be influencing the results - typically it would represent something like Home Advantage, which we know can have a strong influence on results.
When we run the model an include such an order effect, it will output estimated team abilities based on the results in question and controlled for the influence of order effect. If we take Home Advantage as the order effect, for example, this means that if both Team A and Team B each have identical records, but Team A has played 90% of their games at home, and Team B have played 90% away, then Team B will have the higher rating.
The other output - and the key metric for this particular analysis - is the order effect coefficient. This is a measure of to what extent a team's 'ability' is boosted when the order effect is in their favour, i.e. how much of an advantage is gained by playing at home or, in this case, by batting first.
Without further ado, let's take a look at the model output, which takes in all the results from the group stage, using batting first as the order effect:
The model shows batting first as boosting a team's ability by 1.0012. To explain the implications of this, let's consider England, who are rated as 3rd best in our model. They are just behind Australia, and so our model would initially consider Australia favourites if these two were to meet again. However, this would change if England were to bat first, as their ability would be boosted by 1.0012 and would go ahead of Australia. Similarly, England would be initial favourites against both Pakistan and New Zealand, but the odds would flip against them were they to field first.
The batting first coefficient is a remarkably high figure - for comparison the equivalent figure for home advantage for ODIs in the four years preceding the tournament sits at around 0.4
An interesting side note here is that Sri Lanka are rated below both South Africa and Bangladesh, despite finishing above them in the standings. This is largely due to controlling for the advantage gained by batting first - an advantage that Sri Lanka enjoyed in 6 of their 7 matches, whereas as both South Africa and Bangladesh batted first just 3 times each.
The raw figures in the table above are admittedly somewhat abstract and difficult to interpret on their own, but to better understand the effect that batting first can have we can convert these into win percentages for each possible matchup:
The table shows winning probabilities for the team in the 'Batting first' column, based on their chances of victory if they were to bat first against each of the other teams. Afghanistan have been excluded from the table since their grand total of 0 wins mean that the model gives them a 0% chance in every possible match.
Note that these win percentages shouldn't be taken as accurate predictions for what is to come - they are only based on results during the group stage of the competition and there are, of course, dozens of other factors that haven't been taken into account. They could perhaps be seen as the predicted outcome if the two sides in question put in an average of their performances in the tournament. In any case, the main goal here is to look at how heavily these win percentages swing depending on who would be batting first.
The table shows some dramatic swings in win likelihood based on who is batting first. Let's take the upcoming semi-finals as examples. If India bat first against New Zealand, the model gives them a very generous 90% chance of victory, but if New Zealand bat first, India's win likelihood drops to 56%. In that game India are still favoured regardless of the outcome of the toss, but it's a different story for the other semi-final. If England bat first, their win likelihood is 66%, making them fairly strong favourites, but this swings dramatically to 79% in favour of Australia if they bat first.
An easy way to demonstrate the effect of batting first is to look at the respective win probabilities of two otherwise evenly matched teams. In this model, England and New Zealand both come out as fairly even, and when running the win probabilities without the influence of batting first they are given a 50-50 chance each. This then jumps to 73% in favour of the team that gets to bat first.
What does this all mean? Well, it basically shows that throughout the tournament that batting first has been a significant advantage - enough to make a potentially even matchup into a heavily one-sided affair. It likely means that whoever wins the toss in the semi-finals will elect to bat first, and we can only hope that the games can still be competitive and won't be effectively be decided by the flip of a coin.
Having said that, if England win the toss, bat first, and cruise to victory in both the Semi and the Final, you'll hear no complaints from me
Comments