Bowlers win tournaments

In limited overs cricket, it is a common refrain that batsmen win matches but bowlers win tournaments. Pakistan seemed to prove this point true in the Champions Trophy this year. With the group stages over, we saw the most explosive batting line-up in the tournament fizzle out against a resurgent Pakistan bowling attack. Even without Mohammad Amir, they quickly removed the top order and then proceeded to tame the previously untameable Ben Stokes, who lasted 64 balls but without hitting a single boundary

In the final, Pakistan continued to demonstrate the importance of top bowling. Amir returned and triumphantly dismissed Sharma, Dhawan, and Kohli - the highest run-scorers in the tournament. The brilliant Hasan Ali was fittingly crowned as the player of the tournament

I used my Win Probability model to test this idea in T20 cricket. The first step was to identify top match-winning performances. To illustrate the method, let's take an all too familiar performance near the top of the list: Carlos Brathwaite in the 2016 World Cup final

We may all remember his contribution with the bat but Brathwaite had also bowled well in the first innings. He took 3 important wickets and was a huge factor in limiting England to just 155. It was enough to earn 0.26 Win Probability Added (WPA) - his teams chances improved by 26% while he was bowling.

 Top ten tournament performances based on Win Probability Added (WPA)

Of course, the real match-winning was achieved with the bat. When Ben Stokes began his ill-fated final over, my model rated England as 91% favourites. Brathwaite annihilated this advantage with 4 consecutive sixes and added a further 0.91 WPA to his total. Taking into account his entire innings and also his feats with the ball, Carlos Brathwaite earned 1.13 WPA over the course of the entire match

How does this help us to decide whether batsmen or bowlers win matches and tournaments?

I looked a past tournament matches and found the top WPA performances for winning teams and determined whether it was by a batsman or bowler. Where players made positive contributions with both bat and ball, I classified performances into either Mainly bat, All-round, or Mainly Ball

 Batsmen win matches. 50% of match-winning perfromacnes were by a batsman vs. 33% bowlers

Carlos Brathwaite rightly won Man of the Match honours for his efforts and, in this case, he is classified as Mainly Bat. He contributed in both departments but more than 75% of his contribution came from his bat

As the adage predicts, batsmen win T20 matches. They produced 49% of match-winning performances as opposed to 33% by bowlers. If we relate the distribution of match-winning performances to the balance of a T20 team, then this equates to about 5.5 batsman, 2 all-rounders, and 3.5 out-and-out bowlers. This may seem fair but I would argue that bowlers are under-represented - 50% of the game is completed with the ball in hand

The next question is clearly whether this trend holds true for tournaments as a whole. I completed the same analysis as before but, instead of looking at individual matches, I aggregated performances across the entire tournament. The result was an even stronger skew towards batsman - just one bowler was their team's top performer across the duration of an entire tournament

 Batsmen win tournaments too. Top performers in T20 tournaments are almost exclusively batsmen

Based on this evidence, it would seem that batsmen win both matches and tournaments. But there are some caveats. Firstly, there are some players in the list above who were included in their teams to bowl, Carlos Brathwaite being the number one example. Bowlers often bat towards the end of innings when Win Probability can can change very quickly and this can boost their batting WPA and lead to them being classified as an all-rounder. Even so, the fact remains that the list is dominated by batsmen (of course, what you really want is someone who does both)

The dynamics of batting and bowling are also very different. Teams generally have 5-6 batsmen capable of delivering a valuable innings but only one or two need to actually deliver. Batsmen are like lottery tickets - some won't win you anything but it doesn't matter... as long as one hits the jackpot. It is hard for a batsman to have a particularly negative effect on the game, aside from 'pulling an Amla' and batting very slowly for a long period of time

Bowlers, on the other hand, are more like chainmail. Bowling attacks are as strong as their weakest link and batsmen often target that link. If one bowler is hit for 24 in a single over then that can completely undo the efforts of his teammates

 Distribution of permances as measured by WPA

We see this difference when we look at the distribution of performances with the bat and compare it to the distribution with the ball. The average of both distributions is the same, zero, but the distribution of batting performances is negatively skewed: there are many failed innings which have a small negative impact but enough brilliant innings to cancel out. The distribution of bowling performances is more even: there are as many poor performance as good performances

So it is unsurprising that match-winning performances are more likely to be produced by batsmen. The winning team almost always features at least one success innings from the top 6 but it may not always be the same player. Tournaments are short enough that the same individual could produce a series of match-winners

Considering that bowling attacks are theoretically as weak as their weakest link, maybe we should consider them as a whole. Individual bowlers do not win matches but maybe bowling units do

We can look at whether teams win matches with their batsmen or their bowlers. I calculated the total WPA accrued by each team over the course of an entire tournament, broken down between bat and ball. Every team was then ranked in both departments. To account for the differing number of teams between tournaments, I adjusted the rank by dividing by the total number of teams. The eighth best bowling attack in the Big Bash (8 teams) is the worst bowling attack but the eighth best bowling attack at the World Cup (16 teams) is about average

 Showing the relative quality of bowling and batting units for tournament winning teams. table

In the end, it doesn't matter how you rank the teams. 54% of tournament-winning teams have the #1 ranked bowling attack in the competition. Only 21% have the #1 ranked batting attack

This speaks to the relative consistency of bowling over the hit-or-miss nature of batting. If your best players are bowlers then you can expect value from them in every match, from the group stages to the knock-outs. Royal Challengers Bangalore discovered that even the greatest T20 batsmen can't guarantee victory in every match: they lost against the number one ranked bowling attack in the final in 2016 and flopped spectacularly the following year throughout the entire tournament

Carlos Brathwaite won the World Cup for the West Indies with his bat in 2016. That team seem to be one of the exceptions to the rule, ranked #2 with the bat but only #5 with the ball

The 2016 West Indies may not follow the rule but they do demonstrate why the rule generally works. That last over was such a massive game-changer that it allowed the West Indies to jump from #5 to #2 in the batting rankings based on that over alone. Having to rely on such a spectacular performance is unsustainable - especially when your match-winning batsman is actually a bowler

Conclusion: Batsmen win matches, bowlers win tournaments