Sunday, 3 February 2013

University Challenge - Second Round Review and Quarter Final Preview

Let's have a look at the relative performances of the 8 teams that have made their way through to the quarter finals: -

Team Round 1 score Margin Round 2 score Margin Average ScoreAverage Margin
UCL 260 185 215 35 237.5 110
Imperial 225 145 245 130 235 137.5
Pembroke, Cambridge 200 60 255 180 227.5 120
New College, Oxford 230 85 215 70 222.5 77.5
St. George’s 175 30 230 90 202.5 60
King's, Cambridge 205 45 185 15 195 30
Manchester 180 5 200 110 190 57.5
Bangor 125 20 175 10 150 15

Now, a word or two of explanation about the table. It takes two aspects of each team's performance into account - their overall scores, and their margins of victory. Also, rather than use King's first round score, I have used their score intheir repechage match, and this needs to be considered when assessing their chances.

So what conclusions, if any, can we draw from the table? Well, firstly, that it seems very close. No less than 7 of the teams' average scores are separated by less than two full sets of starters and bonuses. You can't look at the table and say that any of the teams are going to have a relatively easy passage into the semis. Yes, UCL, Imperial and Pembroke all have impressive average margins of victory, but that doesn't tell the whole story. Take St. George's, for instance. In their first round they had to defeat King's Cambridge, who were also good enough to get to the quarters in their own right. Likewise, if we go by the statistics alone, Bangor seem to be very much the outsiders. Yet only this week we saw Bangor defeat a very good Durham team who were widely expected to get to at least the quarter final stages themselves.

So basically anything can happen. A lot will depend on the draw. If we get two of the 'top' teams from the table drawn against each other in their first match, then one will be playing for survival in their second and third, and that can change things. We saw some surprising quarter final results last year, and to be honest, a repeat of that this year is the only thing I can predict with any great confidence.


jim360 said...

I gave my own analysis, or the results of it, over at Jack's Online Writings. In summary, in my rankings UCL and New come top, then St. George's, King's, Manchester and Pembroke, Imperial, and Bangor in that order. I've included the first-round loss for King's but actually if I ignored that we'd have come third. Go figure. I'm not sure whether or not we really ought to be ahead of most teams or not. But then stats never lie...

Everything is very close outside the top two of New College and UCL, so yeah there could be some surprises ahead and, hopefully, some very good matches too as the top teams start to clash head-on. Manchester v. Imperial tomorrow could be a cracker, and UCL v. Bangor will follow that when I hope Bangor can cause yet another upset, but we'll see.

Londinius said...

Hi Jim

I'll be honest, I didn't really want to go down the route of comparing the opposition that each of these teams have faced, because there are variables to take into account. Did a team only post a modest score because they didn't have a great level of knowledge, or was it just because they possessed slower buzzer fingers? Answer, you can't be sure. So I'm really just contenting myself with the stats. What I am convinced of is that any of these teams is capable, given the right set of questions, of beating any of the others. So it promises to be a fascinating set of quarter final matches

jim360 said...

TO be sure I don't believe that my stats are going to be anything other than mildly interesting. It was mainly an exercise for myself to see if I could produce a realistic statistical measure that would help to reflect what actually happens, whatever that is. Of course the chances of that "prediction" being 100% accurate are very small since so far each team has only played 2 or 3 matches which is far too little to go on. Maybe if there were some sort of group stage so that each quarter-finalist had ended up playing 5 matches or something before reaching the quarters then we'd have more to go on than just vague indications. So yes, my stats are very likely to be fairly meaningless in the end.

What they do show is that, apart from outliers at either end, everything is very close, so that it's likely to be too close to call most of the coming matches.

Maybe including more statistics such as bonus conversion rates and starter interruptions speeds/ success might be even more revealing. But far too much effort and to be honest I doubt it will be that much of a predictor. The variation in question sets is enough to make any team look good or bad on a given day, so that again two matches per team is just not enough data for a proper analysis.

Instead, I'll just enjoy the matches. Even being in the series there are several matches I've not seen so I'm looking forward to the coming weeks.

jim360 said...

A small correction by the way. Manchester's 2nd round score was 220 not 200.