Category Archives: T20 Cricket

The Toss and the T20 Blast

Does the toss in the T20 Blast give an advantage to the team who wins it? The % of games won by its winners  over the years in the competition under its various names is shown in the chart below, together with the % of field first decisions.

In eight of the fifteen seasons its winners have won more games than they lost, in the other seven years the opposite. The numbers bounce around from one year to the next, in most years alternate above and below 50%, which they could be expected to do if winning the toss had essentially no effect on the outcomes of matches. Overall since 2003 the win % is 49.8; a pointer suggesting an answer to the question of quite possibly not.

As to the field first decisions made, the big shift to a preference for fielding first is striking; a previous bat/bowl ratio of around 70: 30 reversed over the present decade; conventional wisdoms about how to apply pressure on the opposition old and new,  from  which a reasonable inference is that captains, and the supporting cast of analysts as maybe, presumably do think there is, or at least might be, an advantage to exploit.

In general terms the overall numbers sit comfortably enough with the simple observation that winning the toss could be decisive in very tight matches, and with the view that a bit of luck should be a factor in sport but that the extent of any advantage be simply not large enough, often enough, to impact the win% figures. But the numbers  do also prompt the question of why the current preference for fielding has gone as far as it has and also whether the decisions that are made are due for another shift?

The second chart shows the % of games won batting first and fielding first. In 2017 in rounded numbers the teams who won the toss chose to field 68% of the time, won 44% of the games, of which batting first they won 47%, fielding first 42%;  which might not unreasonably prompt the thought that there were too many decisions to have a bowl first. Standing a year ago and looking back on the then previous 2016 season, the respective numbers are 69%, 57%, 54% and 58%, which might not unreasonably prompt the opposite thought.

In other words relying on the numbers for just one year could be rather misleading; in the early years of the competition, when the decisions made were  towards batting first,  there was for a time some supporting evidence of teams winning proportionately more often batting first. In the last few years it is hard, or at least harder, to say the same for the current preference for bowling first; not only do the win% tend to alternate from year to year, but  also the breakdowns of wins when batting and bowling first as well.

It is, of course,  possible that there could be a systematic advantage from the current bowl first decisions, even with the numbers above, but that the influence is conflated with other general and/or in-play influences.  Some of which influences may also be measurable and possible to (statistically) model, but absent plausible evidence on this, the question is why is the toss anything more than a way of just starting matches?



T20 Blast Statistics 2017

If T20 is the future of cricket what to make of the data analysis that comes with it? Performance stats help decide how to bowl at opposing batsmen, set markers for the number of wickets during a powerplay and so on. Win the  toss and bat? Time was when this amounted to something like a conventional wisdom, although the Guardian newspaper this summer reported that 72% of teams in T20 matches in 2016 batted second and that 55% of teams chasing won;  a conventional wisdom overturned maybe.

Middlesex versus Hampshire at Lord’s 2017. Middlesex won the toss and batted, Hampshire won.

Underpinning much of this is the belief that T20 cricket produces a limited number of variations and, given many games, what statisticians measure will be stable for long enough to introduce an element of predictability. While games may mimic some of the features of an experiment, it would be fair to add that data analysts have not solved what the 18th century philosopher David Hume called the problem of induction. So, if scientific methods in T20 cricket might be useful up to a point, the question is which point or points?

The new wisdom of win the toss and bowl  seems to have permeated well in the T20 Blast; the bat: field first ratios being approximately 30:70 both last year and this. Prior to Finals Day, the number of games won batting first:second in 2016 was 52:63, something very similar to the (presumably more general) numbers given in the Guardian.

As to why there might be an advantage batting second, uncertainty about how the pitch will play and what constitutes a par score may be part of it. Limited overs games may be ‘moving on’ all the time, the effects of changing bat sizes, boundary ropes and mind sets that play the game. The average score of those batting first in games this year was 172,  seven more than in 2016; given this background, playing wait and see after winning the toss is an understandable decision.

Middlesex versus Kent at Richmond 2017. Kent won the toss and fielded, Middlesex won.

This seems believable enough for some games, although the numbers in 2017 after the quarter finals for those winning batting first:second were 58:53; which raises the question of whether the new  wisdom has moved on to the point where it tilts the odds in favour of the opposition.  Perhaps winning streaks in high profile matches have a disproportionate influence on thinking: in the 2016  T20 World Cup the West Indians won the toss, fielded first and won all six matches. Less commented on, maybe, was the experience of the 2016 T20 Blast winners, Northamptonshire, who won the competition on a losing streak of eight coin tosses.

It is, of course, possible that a more sophisticated look statistically, controlling for (measurable) other influences on winning might retrieve support for the current fashion for fielding first. It might do the opposite.  It might also just be that statisticians have been making available the benefits of a sugar pill, or a data placebo, for those minded to swallow one.