Following the Indian women’s cricket team’s stellar start to the 2017 Cricket World Cup, an inevitable question that has come up is whether they are better than the contemporary men’s cricket team. This question is not from a battle-of-the-sexes perspective, but more from the view of whether, relative to the current men’s team, the women’s team is performing better.
This is not an easy comparison to make, since the two teams literally play in different leagues. We could use the performance in a global tournament as a simple metric, but such tournaments are one-off and the knockout nature of the later stages of the tournaments implies that the best team doesn’t always win. Another way is to look at the win percentages of the respective teams, but that doesn’t take into account the relative quality of the opposition. Moreover, while the women’s team has had a stellar year (a victory in the Quadrangular Series in South Africa preceding the strong start in the World Cup), have they really been consistently better than the men’s team?
One good method to evaluate the performance of a team over a long period of time in its own peer group comes from chess, where for over 50 years now, the Elo ratings have been used to evaluate and rank chess players. Invented in 1960 by Arpad Elo, a professor who was also a chess master, the beauty of the Elo rating system is that it not only takes results into account but also the quality of the opposition.
For example, if a very strong player beats a very weak player, the increase in the former’s points is minimal. On the other hand, if a player defeats or draws with a player who is much stronger, the change in ratings is much larger. This way, a strong player or team cannot pad the rating by playing heavily against weak teams (a method certain cricketers have been accused of using to run up records).
The Elo rating system has not been without debate, not least because of certain underlying assumptions (such as the probability distribution to be used, as well as the extent to which a single game should affect the score). However, the system, irrespective of the precise variant used, has shown itself to be robust enough in chess that recently more sports have started adopting it. While official world sports bodies have been slow to adapt the Elo, it has found wide acceptance in board games and among independent sport analysts.
So what do the Elo ratings say about the respective performances of India’s men’s and women’s cricket teams? We will use about 10-12 years of data, going as far back as January 2006 (chosen without loss of generality). We will restrict our analysis to One-Day Internationals—women don’t play nearly as much in other formats—and look at each and every game played in this format in this time period. There are several off-the-shelf open source algorithms available to calculate Elo ratings—we will use the implementation that is part of the PlayerRatings R package (bit.ly/2uRzJI2).
In order to implement the Elo rating algorithm, we start with each team at the same rating (2,200 is a standard starting point). We look at each game played in sequence, adjusting the ratings of the teams following each game. The point adjustment after each game depends upon the points level of both participating teams just prior to the game. This way, we can get a kind of “current Elo ratings" for both the men’s and women’s game. The two tables in Figure 1 show the current ratings.
As we can see, the Indian men’s team is currently in fourth position based on this analysis, while the women’s team is in third position. These rankings may differ from the official International Cricket Council rankings, but actual positions and ratings are always a function of the method chosen, so there is no surprise in this inconsistency.
The current position of the teams’ rankings, however, is not as interesting as the path that the teams took to get to their current ratings. For this reason, it makes sense to see how the Elo ratings of the men’s and women’s teams have evolved since 2006 (see Figure 2). While the ratings are not strictly comparable since they are in different leagues, the shape of the graphs is instructive.
While the current Elo ratings of the two teams (according to our calculations) are not very different, the shape of the curves gives a clear indication of the attention that the women’s team is receiving. As the graph shows, since early 2015, the rating of the women’s team has been on a near-secular increase. And we can see that since the latter half of last year, there has been a sharp increase in the women’s team’s ratings.
The men’s team, on the other hand, has been largely consistent in terms of its rating in the last couple of years, and this is at a much lower level compared with where the team operated in 2013-14. It is, thus, natural that people perceive the women’s team is performing better!