If you want to look at the spreadsheet yourself, it’s here – or alternatively, read the post and this’ll be at the bottom again.
A few months back, I bought the BBL Elo Ratings over to The Flag from my old blog. As complicated as Elo ratings are, a cricket league is one of the best sports to do it on – ties are highly unlikely, and each team has the same season length. Adapting it to internationals would be a whole new challenge, with the added difficulty of uneven schedule strengths – we don’t have any India v Nepal matches to sort out team’s ratings.
We took on that challenge.
We’ve been over the Elo calculations before, so this blog post isn’t going to be the place to do it. Instead, we’ll discuss the one problem we had at the start: setting initial team strengths.
You may wonder why we had to do this, and why we couldn’t just start every team at 1500. The answer to that is simple: Afghanistan. Most of their games have been against nations like Scotland and the Netherlands, who are weaker opposition than (for example) Australia and South Africa. Playing these poor teams, who would have fairly average ratings, would overly inflate their score.
We decided on what’s almost certainly not the solution, but it seems pretty good to us and we had to choose something. To start, we looked at the inaugural ICC T20 ratings, released by world cricket’s governing body in 2011. We then added an even amount to each team’s rating, which took the average to 1500 (what it should be in an Elo system).
That gave us ratings for the top teams, but there were others who weren’t ranked. To give them rankings, we looked at the World Cricket League divisions at the time.
Canada, Ireland, Kenya, the Netherlands, and Scotland were in Division 1, so we gave them a rating 100 below the other teams, then re-averaged to get to 1500. Teams from Division 2 got a rating 100 below that, Oman (division 3) was another 100 down, and Nepal (division 4) ended up bottoming out.
With that in mind, we then looked at every T20 international from October 2011 to today, adjusting the Elo ratings after each one. After all of that, we ended up with these ratings:
Scotland and the Netherlands, although they have a higher rating than Bangladesh and Zimbabwe, are not ranked because of small sample sizes – teams need to have played five T20Is in the last year to qualify for these ratings.
Because Elo ratings can be used for predictions, too, we’ve also included a section on the spreadsheet where you can see upcoming T20Is and our winning chances. Take a look.
The Spreadsheet – http://bit.ly/2FJNzpY
The Flag 