Tuesday, 30 October 2012

University Challenge - Round Two Preview

That was the first round, then. We’ve already had a look at the playoffs , but now let’s extend that a little. Here’s the raw details of the teams still in the competition: -

TeamScore OppositionMargin
UCL 260 85 175
Durham 245 70 175
New Oxford 230 145 85
Imperial225 80 145
Magdalene, Oxford 205 125 80
Pembroke, Cambridge 200 140 60
York 185 100 85
Manchester 180 175 5
St. George's London 175 145 30
Lincoln, Oxford 175 180 -5
Warwick 175 100 75
Jesus Oxford 150 120 30
Kings Cambridge 145 175 -30
Homerton Cambridge 145 230 -85
Lancaster 140 200 -60
Bangor 125 105 20
Bath 125 110 15
Bristol 120 105 15

* NB - Italics indicates a team in the play offs

So looking at the table it seems that our view that this is something of a lower scoring series than some of recent years seems to be borne out by the facts. There’s no score over 300, for a start. Looking at the other end of the table the lowest winning score is quite a bit lower than last year’s. So it’s quite possibly true that the questions have generally been a notch harder.

As regards the teams themselves, well, let’s cover ourselves with the caveat that first round form can be unreliable as a guide, and then let’s pick out some of the teams who are likely to do well. UCL, Durham, New College Oxford and Imperial all seem likely candidates for progression to the quarter final round. In particular I point you towards New College who managed 230 despite their opposition scoring 145 of their own points, and earning their own place in the play offs. The other three teams I mentioned all racked up very large winning margins against teams who , to be harsh but frank, couldn’t really manage to live with them. There’s less than 100 points separating the top 11 teams , which seems to promise some close matches in round 2. In fact I think you can go all the way down to Lancaster – who are underdogs in their play off against Lincoln college – before you get to teams who look unlikely to progress further. Just look at last year. Losing finalists Pembroke actually had the 11th best performance in the first round.Reigning champions Manchester are some way down the table, and yet they have a fantastic record of getting through to at least the quarter finals. So it’s difficult to pick winners now.

One thing you can say, though, is that any team which can improve upon its bonus conversion rate from the first round will be in with a pretty good chance. It’s all up for grabs.


jim360 said...

When I told my Mum that we'd made the high-scoring loser playoffs on a 145, she said that it's been a low-scoring series then. Thanks, Mum!

But it has been. A few careless mistakes probably cost our team about 10 or 15 points of bonuses that we might have got, and I'm sure we weren't the only team like that. I wonder why it's been low-scoring. Fewer questions asked? Just harder ones? None of your Guttenplans or Trimbles?

In terms of score margins, one thing I never really appreciated until actually being behind the buzzer was how fine margins can be to lead to a huge swing. On at least two questions in our first round match I was a fraction of a second too slow to the buzzer. And each of those close calls can be as much as a 50-point swing if you get all the bonuses.

Also for that matter the way the questions fall can hurt team scores since you might know the full set of bonuses that the other team got, and vice versa. I swear we kept getting all the medical questions and the medics of St. George's the Maths and Physics ones!

I've probably said far too much so I'd best shut up...

Welshguy said...

I like to think Bangor could have pulled off a better score in that first round. Maybe 135-140 or so. Still, as I said before, a win is a win, and as the first victory by Bangor since about 1995 we're already heroes as far as the university is concerned!

Andrew B. said...

I don't think "margin of victory" is a very good measure of a team - I'd be more impressed by a team than won 190-180 than one who won 200-50 (given that all shows are roughly the same length).

Londinius said...

Hi Everyone

Hi Jim. As long as you don't give away any results of any matches, then feel free to say what you like here !

Hi Welshguy

Well said ! I think I should warn you, though, that you will probably be ( have been ) battling aganst the power of support from the Clark sofa in your next match.

Hi Andrew

I don't think that I did say that it is a good measure of a team . I don't say that it is any measure of a team , I simply put it on the table on the principle that more information builds up a more complete picture of the teams involved. I'm not in any way qualified to lecture anyone about the use of statistics.

Still, what you yourself have said is actually as good a reason as any for including it. I tend to agree with you - a team which has won a high scoring match has obviously had to contende with another good team, and so is therefore possibly stronger than a higher scoring team that won a walkover. Which means that it is the margin of victory which showed you this and enabled you to make this judgement.

If you mean that you don't think that a huge margin of victory is a good measure of a team, well, yes, that I would possibly agree with. Or at least, I would agree that it does not necessarily indicate that the team are extremely strong.

Welshguy said...

Do you know who we face next? Obviously, I do, but how do you know if you do? Is there a pattern w/regards who plays whom in the next round - I'd assumed it was just random.

Welshguy said...

The way to get a better understanding of informative first-round results are would be to analyze past competitions. It'd be interesting to see how series champions did in their first rounds.

Looking at the wikipedia stats for the past five years, it seems that the winning teams scored 255, 340, 165 (lost), 330, 245; so doing well in the first round is certainly a strong indicator of future success.

Londinius said...

Hi Welshguy.
No, I don't know who you've got next. The reason why I'm saying you'll have my support is a) You're representing a Welsh university - and b) You're a contributor to LAM. That's a heady combination, and I can't see whichever team you face having a greater claim to my support ! Mind you, that's a poisoned chalice if ever there was one.

Andrew B. said...

Yes, sorry for suggesting that you were suggesting something you weren't (if that makes sense!)

I think maybe "total score" is a clearer indicator than "margin of victory" (though of course they are related).

dxdtdemon said...

I'll be tabulating various statistics about UC tonight (in my time zone), in order to come up with something resembling the formula that NAQT decides which universities from the US, Canada, and UK qualify for their "national" tournament, but with a bunch of modifications and assumptions that make the rankings very hand-wavy.

jim360 said...

I'd be interested to see those stats - where would I be able to find them?

I think the pairings are essentially random, so that New College Oxford could play UCL as likely as playing Bath or Bangor. I have no idea how true those are, I'm just picking names, so if that turns out to be an actual match it was a lucky guess - honest!

dxdtdemon said...

My internet connection was down for most of the night, so I'll have to make the spreadsheet tomorrow. Sorry about that.

dxdtdemon said...

tl:dr?, then click on the link at the bottom
Well, I knew that the formula wasn't going to be that great of a predictor because some of the variables had to do with how well your opponents did. Yet, I didn't realize how utterly useless it would be when teams played only one opponent like they do in University Challenge as opposed to 10+ opponents in NAQT until I found that Durham would be ranked 14th and UCL 16th. Anyway, I copied some of the more sensible stats into a different spreadsheet and ranked the teams by bonus conversion instead. I would like to thank LAM user Jack so that I could double check some of the stats and to David Clark whose detailed descriptions of the matches helped solve some discrepancies in the stats without having to try to find the episode on YouTube and watch it again.

I think the stats are pretty straightforward. Dspsh stands for dead starters per starter heard. I thought it might be a useful indicator of teams that might have some knowledge holes going forward. I did not count starters that were still being read at the gong as going dead. I also did bonus conversion by 15*(bonus parts answered correctly/total bonus parts), so that teams weren't penalized for having the gong go off in the middle of their bonus.

Some things I noticed were that the average points per bonus for all teams was 6.72, which is about 44.5% of all bonus points asked in the first round. Most question companies aim to have at least 50% bonus conversion for their target audience, so they did make the questions a little harder than they intended. Assuming that University Challenge maintains the College Bowl tradition of packets of 28 starters and 28 bonuses per round, only one match used the whole set of starters, although some of the 27's may have had the 28th being read at the gong.

Best of retrospective luck to all teams that are still alive.

Here is the link to the spreadsheet:https:

dxdtdemon said...
This comment has been removed by the author.
dxdtdemon said...

I'm not sure why the link isn't hypertexting.

Welshguy said...

Interesting stats there. Glad to see that they somehow rank us higher than any other measure would...

Londinius - I was a bit confused because your post seemed to suggest that you would be supporting our opponents; but now I realise you merely meant that your support is a curse... thanks anyway though!