9. What happens in a ball?
Did you know that half the balls in an ODI are dot balls? Even in 2019.
Cricket data analytics company CricViz has offered broadcasters of the ongoing World Cup enhanced player profiles. Apart from standard statistics such as averages and strike rates, these player profiles also include shot profiles, scoring zones and other non-standard information.
While the use of additional statistics is welcome, the problem is that when presented without context, it is hard to interpret statistics. For example, what do we make of Mushfiqur’s 54% dot ball rate, and 8% boundary rate? If you were to ask a random cricket fan if these numbers are high or low, it wouldn’t be easy for them to answer. For the percentage of dot balls and boundaries are not information that cricket fans are used to tracking.
It is not in the scope of this newsletter to tell you how to represent data better (for that, you might want to check out my Bad Visualisations Tumblr), but this graphic from CricViz suggests it might be interesting to look at the distribution of events that can happen on each ball.
We will ignore extras for now and only focus on runs off the bat. For runouts where runs are scored, we will ignore the runs and simply consider them as wickets.
In terms of broad trends, there are no surprises in the data. Dot balls have become less frequent, from 58% in 2003 to about 50% in 2019 (so Mushfiqur’s 54% is on the higher side). There has been a secular increase in sixes. Fours were at the 6.5-7% mark until 2012, when there was a step change to about 7.5%, where it has remained since.
Singles also increased through the first decade of this century, and have since stabilised at around 31%. And since 2007, which can be considered the year when T20 cricket took off, the number of wickets in an innings went up from 7.8 to 8.1 - perhaps as a result of batsmen attacking more. Twos have decreased marginally but stabilised at around 5% of all balls.
It gets more interesting if we look at the same graphs based on the period of the game. While powerplay rules have been ever changing, it’s reasonably fair to divide the game into three periods - powerplay of overs 1-15 (yes I’m a nineties kid), middle overs from 16 to 40, and slog from 41 to 50.
Some pertinent observations from here (you can click on the image to open a full-size version):
Threes are far more common in the first 15 overs than in other periods of the game, but the incidence of them has dropped significantly
The difference in proportion of dot balls based on stage of the game is staggering. Even now, about 60% of balls in the first 15 overs are dots. The number for the last 10 overs has remained less than 40% since 2002.
There was a step-change in the number of fours hit in the middle overs from 2012 onwards, when the number of fielders outside the ring was restricted to 4 (it had been 5 until then)
Rotating the strike at the beginning of the innings wasn’t really a thing in the early 2000s (barely 15%). This has seen a significant and steady increase since then
The proportion of wickets has dropped marginally in the powerplay, and increased marginally in the middle overs. Also notice how the incidence of wickets in the last 10 overs is so much higher than in the first 40 overs.
The rate of six-hitting in the last 10 overs has increased much much more (albeit perhaps not on a percentage basis) than in the first 40 overs. Also there is a divergence developing between the rate of six-hitting in the first 15 overs, and in the middle overs.
The last two points together suggest that teams have been taking lesser risks in the powerplay this decade, and instead making up for it by faster scoring in the middle overs
I can go awn and awn and awn, but I’ll stop here.
As a parting note, I encourage you to read this note along with the edition on clustering batsmen. Given how much the distribution of events on a ball varies with the stage of the game, you can think of the proportion of dots, singles and boundaries hit by a batsman as more of a reflection of the stage of game he bats in rather than his inherent batting style.
Perhaps it will be a useful exercise to cluster batsmen after controlling for the stage of game they’ve batted in!
PS: I hope you’ve been enjoying the World Cup, though there have admittedly been few exciting phases in the tournament so far. My favourite piece of play from the little bit I’ve watched is Andre Russell’s spell against Pakistan where he consistently bounced them and got Fakhar Zaman and Haris Sohail to set up a collapse.
You can look at the stories of the games (even the live ones) in my web app.