|Two teams, one trophy.|
For each of the eight Only Connect finals to date I compared the round-by-round scoring of the eventual champions with that of the runners-up across their respective runs to the final. For each team I looked at the following six statistics (averaged across their games): total points scored, winning margin in those games, points scored in each of the first three rounds (connections, sequences, and walls), and points percentage in missing vowels.
These are hopefully self-explanatory except for the missing vowels 'percentage', which is a stat I concocted for comparing series champions in a previous post. Rather than looking at total points scored in missing vowels, which can vary in length from show to show, I instead compare how many points a team scored with the combined total of their and their opponents' score in the round. For example, if a team scores seven points in missing vowels while their opponents score three, their missing vowels percentage would be seven out of ten, or 70%. Not ideal, of course, but as most missing vowel clues are solved it's hopefully slightly more reliable than looking at raw score for that round.
For simplicity I decided to consider, for each of these metrics, whether or not the team that went on to become series champions had a better or worse record across the series than the eventual runners-up. The results of this can be seen in the figure below, where green indicates the champions had a better record, red a worse one, and orange that they were all square. (If you have trouble with red and green I've made a greyscale one, and if you want to see the full numbers you can check them out in an ugly table here.)
The first thing to note is that in every series so far the eventual champions have scored more points on their way to the final than the runners-up. Champions also tend to have better winning margins, with this being the case for seven of the eight. The round-by-round data, meanwhile, are rather less informative. Sequences are arguably the best indicator of potential champions, with six out of the eight champions demonstrating a better record on that front (and on the two other occasions things were very close), but beyond that you're usually looking at something of a coin toss.
|How the Series 9 finalists compare.|
The Relatives have the edge in four of the six metrics I've considered, including the apparently all-important 'average points' statistic, having averaged an additional 3.5 points per show. They only trail in winning margin and missing vowels, with the Europhiles' record in the latter particularly impressive. However, while the Europhiles have only conceded a grand total of six points to opponents in missing vowels, five of these came in their early encounter with the Relatives. If things are close going into the final round be prepared for some fireworks.
Now before I get a bunch of angry comments telling me to give back my PhD I should acknowledge that all this analysis is a little tongue-in-cheek given the size of the dataset I've worked with and the numerous assumptions I've had to make. The biggest question mark hangs over whether it's at all legitimate to compare team's head-to-head records, as these will be at least partly determined not only by which opponents they happened to meet along the way, but also by their route through the tournament. With later rounds supposedly featuring harder questions, for example, the Europhiles have presumably faced tougher sets on average given the Relatives' extra game during the theoretically easier first round. Then again, the Relatives pulled off an impressive 18-15 win in their semi-final while the Europhiles came home 11-7, but it's impossible to say how much of this is down to a better team, and how much is down to easier questions or just plain old luck of the draw.
Nevertheless, the stats alone suggest that while the Relatives may have stumbled in their first meeting they're now firmly the favourites on paper. My money, however, is on the Europhiles, who have impressed me a touch more over the series (and because you can rely on a statistician to hedge his bets). Either way we should be in for a cracking contest, and once you've seen it come back tomorrow where I'll be looking at the real Only Connect talking point: "is it just me, or was this series really, really hard?".