Toss statistics lie, because of who tends to be winning anyway

The graphic showing sides who bowl first win more games mixes up the decision with the teams good enough to make it.

The graphic appears in the second over of every night match. Sides batting second have won so many of the last so many at this ground. Then a former captain nods and says the toss is worth twenty runs, and everybody moves on, and the number was close to meaningless.

The problem is who is in the sample. Captains do not choose at random. They choose based on the pitch, the dew forecast, the opposition and their own attack, which means the decision is correlated with everything that also affects the result. Add to that the fact that a stronger side wins more often regardless of the coin, and any raw split between batting and bowling first is contaminated before you look at it. You are measuring judgement and quality, then calling it the toss.

There is a cleaner way to look at it and hardly anyone shows it. Restrict the comparison to matches where the choice was forced or obviously conventional, control for the sides involved, and the advantage usually shrinks to something much smaller than the confident on-air figure. At some venues under lights it stays real, because dew is a genuine physical asymmetry rather than a preference. At most grounds in daylight it more or less disappears into noise.

I would go further. The obsession with the toss gives captains an excuse that ought not to be available. Lose after calling wrong and the analysis stops there, when the interesting question was why the middle order could not survive twelve overs of spin.

The place where the coin does decide matches is the multi-day game on a surface that changes character, and there the effect is not the toss, it is the pitch, and the pitch is a choice somebody made a week earlier.

It would take a producer an afternoon to build the better graphic. Same venue, same phase, split by whether the chasing side was the stronger one on paper, with the sample size printed underneath in a font people can actually read. Nobody does it because the honest version usually says the sample is thirty-one matches and the difference is inside the noise, and that is not a line anyone wants to deliver over a shot of the coin landing.

So put the graphic away, or put an error bar on it. One of those is a sentence a producer will never approve.