In yesterday's instant reaction thread, there was some discussion of the possible permutations of Arsenal's group, the underlying assumption of which was that Arsenal need to win their group. Being the contrarian that I am, I expressed a desire to understand whether the advantage that is supposedly conferred on a group winner over a group runner up as far as advancement and favorable next-round matchup is actually true.
I like data, I like being able to look at data, and I like drawing conclusions from that data, rather than relying on conventional wisdom, so here's what I did: Starting in 2003-04, which was the first year of the current 32-team, eight-group, one group stage format, I compiled the winners of each group and the runners-up, and in each stage I highlighted the group runners-up as they advanced through the tournament. I was not doing a probability analysis - this was and is not intended to be predictive of future results, or to calculate the odds of any particular runner-up advancing. This was merely an exercise to see how many group runners-up advanced further than group winners.
I found some pretty surprising stuff - surprising mostly in its consistency. I'm not sure what I was expecting to find, but what I found was kinda cool:
- Of the 80 participants in the quarter-finals since 2003-04, 23 were group runners-up, or 28%
- Of the 40 participants in the semi-finals, 11 were group runners-up, or 27.5%
- Of the 20 participants in the final, three were group runners-up, or 15%
- Of the ten winners, three were group runners-up, or 30%
So there is definitely an advantage to winning the group - historically there's been about a 70% chance of proceeding through the first two knockout rounds as a group winner. The semi-finals seem to have been a speed bump for runners-up, but if you're a runner-up and you make it to the final, your odds get back up to that 30% level - not great, but not hopeless either.
As I said, mostly what surprised me about these numbers is their consistency. There's obviously variance year-to-year - in 2005-06, only one runner-up made it to the quarters, and in 2007-08 five of the eight quarterfinalists were runners-up - but overall it seems like the whole pot/seeding convoluted thing is doing what it's intended to do - reward the higher seeds and make things harder for the lower seeds.