About a decade ago, I discovered that my insistence to play Bowser on every version of Mario Kart EVER was a giant, giant mistake. I'd got sucked into the implied equivalence that each character would be as fast as the last - but somehow I kept getting smoked by that little Toad douche. Turns out some characters are just flat out faster than others. To this day I still use Bowser (coz fuck Yoshi), but i'm resigned to battling it out with Wario and DK, and I'm comfortable with my lot.
There's an implied equivalence too at the start of cricket matches (T20 and ODI). With the issue of pitch degradation largely a test-match cricket phenomenon, we're lead to believe that neither team has an advantage with respect to batting or bowling first. But with only the simplest of data analysis, it appears that batting first is the Bowser of coin toss decisions. Teams that bat first, win less often. It's statistically significant, and it's happening in both short forms of the game (T20: odds ratio* 0.840, p-value* 0.0082. ODI: odds ratio 0.942, p-value 0.0357). The cruel irony is that teams have been very slow at picking this up. This is particularly true of the Australian Big Bash League where only in this most recent season (2016/17) have they figured this out to any relevant degree. The following charts show the tendency of captains in the BBL and IPL leagues to elect to BOWL first. While it should (by rights) be close to 100% given this information, it appears only to have surged up in the right direction more recently.
Proportion of "BOWL FIRST" decisions by captains who win the toss in the Big Bash League (n = 204), by season.
Proportion of "BOWL FIRST" decisions by captains who win the toss in the Indian Premier League (n = 574), by season
In One Day Internationals (ODIs), the trend may be only slightly positive and we are still below 50% in the most recent year. This means that more captains are happy to give the other team the advantage!
Proportion of "BOWL FIRST" decisions made by captains in One Day Internationals (n=3658, since 1971), by year. Only years since 2000 shown.
Why is this happening? Why are captains in T20 only JUST waking up to this, and those in the 50-over format not at all? Well in the video below I discuss the notion of Action Bias. Psychologists will tell me it's to do with the notion of regret theory and diversions from accepted "norms", but for our purposes here we can consider that cricketers are biased towards taking ACTION to win the game. To let the opposition bat first is seen as a passive (maybe even weak) option, so naturally captains will head out to SET THE PACE of the game and elect to bat first.
Here's the problem; When you bat first, you really don't know what kind of daemon you're facing. Let me explain. There's a fairly defined trade off between batting conservatively (protecting one's wicket) and batting aggressively. Think of it as aggression slider that each batter has in their head when they arrive at the crease. To the left, they're William Scotton. To the right, AB de Villiers. The team that bats first really has NO idea where best to place that slider. Most likely they'd push it into the de Villiers direction, taking on more risk than maybe due. For the team batting second though, each batter that comes to the crease knows EXACTLY where they need to put that slider. In other words, batting second gives the other team the opportunity to SUCK first. And knowing that the other team sucks is really helpful when trying to figure out how much risk to adopt in your innings. Expect most coin tosses in T20 cricket (at least) to lead to bowl first decisions.
* An odds ratio is calculated by dividing the number of wins to teams that bat first by the number of wins to the teams that bowls first. If there were no advantage to batting or bowling first, the odds ratio should be 1. You'd expect SOME variation in a random sample of course, but that's where the p-value comes in. The lower the p-value, the less likely the difference is due to random variation. P-values less than 0.05 are seen by the scientific community as being statistically significant.
