Page 1 of 1

Stattopalooza

Posted: Sun Oct 28, 2012 5:43 pm
by Graeme Cole
Following on from this, here are some more statistics. We can do things like this because of the work done by a number of people over the years, mainly Charlie Reams, Mike Brown and James Robinson. Countdown's probably unique in the world in this regard - can anyone think of another game show for which such comprehensive data is published?

These figures include all ordinary series, Championships of Champions, Supreme Championship, Junior Championship, Ladies' Championship, Countdown Masters and Specials. Excluded are unbroadcast episodes or episodes we don't have details for. Series 67 is included up to and including this episode from last week. Also excluded is this game - contestants changing mid-game, and Gyles buzzing twice on the conundrum (you can't do that!) leads me to believe this game was deliberately set up to annoy anyone trying to put all the Countdown games in a relational database 25 years later.

Where a contestant offered more than one word, which seems to have been quite common in the earlier years, I've only taken the first one. Incidentally, the contestant guidelines now specifically tell you not to offer more than one word. This is presumably because if you smugly offer three words and it turns out they're all the wrong length, you look a bit silly.


Comebacks

In a CoC IX last-16 match in 1997, Pete Cashmore won despite being 20 points behind with two rounds left. Perhaps surprisingly, this is the only time that's ever happened.
The largest deficit overturned in the last three rounds of any game is 21. Two people have managed it: Lee Graham in his "decider" special against Nicki Sellars last year, and Dan Webster against Tuck Broadbent in series 54.
The largest ever overturned deficit was 40, by Richard Pay in series 51.


Words

Most frequently offered words
TRAINED 118
RATION 118
TRAILED 99
RATIONS 97
ROASTED 91
PAINTED 87
POINTED 85
LOITERS 78
COATED 74
PANTIES 73

Most frequently offered words as maxes
RELATION 55
TRAINED 42
COASTED 38
TRAILED 37
DONATES 37
POINTED 37
FLOATER 35
PAINTED 34
STRAINED 34
SPRAINED 33
RATIONED 33

Most frequently disallowed words
This only counts words disallowed for being unacceptable words, not for any other reason. Note that BAITER, which used to be invalid, is now allowed.
entrail 15
gaited 13
staired 12
corgies 11
outlaid 11
baiter 10
casted 10
pattie 10
pilates 9
failers 9
moulter 9

Talking of disallowed words, here are the reasons words can be disallowed, and how many times it's happened:
Unacceptable word: 5683
Not in selection: 1217
Misdeclared: 292
Not written down: 6

"Misdeclared" includes when someone declares a length but then doesn't offer a word when asked, as well as offering a word that doesn't match the length declared.

Breakdown of misdeclarations
Misdeclared as 4: 6
Misdeclared as 5: 32
Misdeclared as 6: 85
Misdeclared as 7: 116
Misdeclared as 8: 50
Misdeclared as 9: 3

Most common words to be disallowed because they aren't in the selection
EXILED 7
ROUTER 3
GREATER 3
NEGATED 3
NATION 3
DUSTED 3
... and another 20 words with two occurrences.

I don't know what it is about EXILED, but in the last 30 years it's been offered 11 times, and on only four of those occasions has it actually been in the selection. :shock:


Numbers

CECIL's favourite number is 609 - it's come up 29 times in 14,123 numbers rounds for which we know the target. The least common is 100, which has only come up four times, but that's probably because CECIL used to be able to pick 100 but not any more. The next least common targets are 931, 929, 268 and 836, with 6 occurrences each.

Frequency of picks
6 small: 1164 (8.2%)
1 large: 10053 (71.2%)
2 large: 1996 (14.4%)
3 large: 378 (2.7%)
4 large: 528 (3.7%)

Percentages are rounded to 1 decimal place. The counts don't add up to 14,123 because there were three rounds where we know the target but not all the numbers.


Conundrums

Conundrum buzz time frequency.

Image

Everyone who has Superstats on apterous has a graph that looks like this. But what if we split it into correct and incorrect buzzes?

Image
Image

Predictably enough, quite a few people buzz in with a wild guess near the end that turns out to be wrong. And no, I don't know what happens at 10 seconds that suddenly makes people think of the right answer. It's possibly a mistake somewhere, as the graphs on Superstats don't seem to show such a thing, but I can't see one.

The contestant subjected to the most crucial conundrum games was series 37 champion Ray McPhie. Of his 13 games, 10 were decided on the final conundrum.

Finally, tiebreaks: 69 games have ended in a tiebreak. Eight contestants have been subjected to two tiebreak games: Clive Freedman, Satbir Gupta, Dennis Fearon, Gerry Pearce, Phil Cooke, Michael Calder, David Hoskisson and Denise Smith. Of those, Freedman, Fearon, Calder and Hoskisson won them both.

Re: Stattopalooza

Posted: Sun Oct 28, 2012 5:56 pm
by James Robinson
Bloody hell, Graeme. This is a statistician's fantasy :!: :!: :!:

Major kudos to you my friend. ;) :) :D :geek: :ugeek: :mrgreen: 8-) ;) :) :D :geek: :ugeek: :mrgreen: 8-) ;) :) :D :geek: :ugeek: :mrgreen: 8-) ;) :) :D :geek: :ugeek: :mrgreen: 8-) ;) :) :D :geek: :ugeek: :mrgreen: 8-)

Re: Stattopalooza

Posted: Sun Oct 28, 2012 6:07 pm
by James Robinson
Graeme Cole wrote:Most frequently offered words as maxes
RELATION 55
TRAINED 42
COASTED 38
TRAILED 37
DONATES 37
POINTED 37
FLOATER 35
PAINTED 34
TRAINED 34
SPRAINED 33
RATIONED 33
Although, they look like the same word to me...................... :?: :?

Re: Stattopalooza

Posted: Sun Oct 28, 2012 6:12 pm
by Graeme Cole
James Robinson wrote:
Graeme Cole wrote:Most frequently offered words as maxes
RELATION 55
TRAINED 42
COASTED 38
TRAILED 37
DONATES 37
POINTED 37
FLOATER 35
PAINTED 34
TRAINED 34
SPRAINED 33
RATIONED 33
Although, they look like the same word to me...................... :?: :?
Whoops. Mispasted. The second one is STRAINED.

Re: Stattopalooza

Posted: Sun Oct 28, 2012 6:13 pm
by Jennifer Steadman
This is brilliant. Great work, Graeme! :)

Re: Stattopalooza

Posted: Sun Oct 28, 2012 7:10 pm
by Jon O'Neill
Awesome. Well done sir.

Re: Stattopalooza

Posted: Sun Oct 28, 2012 9:24 pm
by Mike Brown
Massive LIKE. Is there some kind of Graeme Cole fan club we can join? And I hope you haven't stolen too much of Charlie's thunder! (hasn't he been planning this sort of thing for years?)

Re: Stattopalooza

Posted: Sun Oct 28, 2012 9:32 pm
by Graeme Cole
Mike Brown wrote:And I hope you haven't stolen too much of Charlie's thunder! (hasn't he been planning this sort of thing for years?)
I did think about this, but my impression is that what Charlie's planning is an ongoing thing that gets linked to the recap writer and whatnot so people submit games straight into it and the whole thing gets updated. This is just analysis of a static data set which is already slightly out of date (it doesn't include the last five episodes) and is only going to get more so.

If this is a photograph, Charlie's plan is a high-definition live video feed, which I'm still looking forward to seeing. Have I understood that right, Charlie?

Re: Stattopalooza

Posted: Sun Oct 28, 2012 9:39 pm
by Mike Brown
Graeme Cole wrote: If this is a photograph, Charlie's plan is a high-definition live video feed, which I'm still looking forward to seeing. Have I understood that right, Charlie?
Ha ha, nice analogy. And I wasn't suggesting any impropriety on your part, Graeme, just didn't want to see any feathers ruffled!

Re: Stattopalooza

Posted: Mon Oct 29, 2012 12:08 am
by Jon Corby
Graeme Cole wrote:And no, I don't know what happens at 10 seconds that suddenly makes people think of the right answer. It's possibly a mistake somewhere, as the graphs on Superstats don't seem to show such a thing, but I can't see one.
Isn't this most likely to be recording accuracy? As in Robbo might time it to the 1/100th of a second, but most recappers'll just go "yeah, that was about 10 seconds". The fact that it's a nice round number makes me more inclined to think this.

Great work btw.

Re: Stattopalooza

Posted: Mon Oct 29, 2012 12:27 pm
by Gavin Chipper
Mike Brown wrote:
Graeme Cole wrote: If this is a photograph, Charlie's plan is a high-definition live video feed, which I'm still looking forward to seeing. Have I understood that right, Charlie?
Ha ha, nice analogy. And I wasn't suggesting any impropriety on your part, Graeme, just didn't want to see any feathers ruffled!
So what you're saying is that you're a bit scared of Charlie "The Axe Man" Reams?

Re: Stattopalooza

Posted: Mon Oct 29, 2012 5:53 pm
by Mike Brown
Jon Corby wrote:Isn't this most likely to be recording accuracy? As in Robbo might time it to the 1/100th of a second, but most recappers'll just go "yeah, that was about 10 seconds". The fact that it's a nice round number makes me more inclined to think this.
Surely not? I reckon most people would at least post to the nearest second, but maybe I have overly-high expectations. Personally, I'm in the 'nearest quarter-second' brigade. :)

Re: Stattopalooza

Posted: Mon Oct 29, 2012 6:44 pm
by James Robinson
Mike Brown wrote:
Jon Corby wrote:Isn't this most likely to be recording accuracy? As in Robbo might time it to the 1/100th of a second, but most recappers'll just go "yeah, that was about 10 seconds". The fact that it's a nice round number makes me more inclined to think this.
Surely not? I reckon most people would at least post to the nearest second, but maybe I have overly-high expectations. Personally, I'm in the 'nearest quarter-second' brigade. :)
I'd like to state that I'm in Mike's brigade too. :D

Re: Stattopalooza

Posted: Mon Oct 29, 2012 11:15 pm
by Thomas Carey
James Robinson wrote:
Mike Brown wrote:
Jon Corby wrote:Isn't this most likely to be recording accuracy? As in Robbo might time it to the 1/100th of a second, but most recappers'll just go "yeah, that was about 10 seconds". The fact that it's a nice round number makes me more inclined to think this.
Surely not? I reckon most people would at least post to the nearest second, but maybe I have overly-high expectations. Personally, I'm in the 'nearest quarter-second' brigade. :)
I'd like to state that I'm in Mike's brigade too. :D
I think most recappers (me included) go for the nearest second. I always look at the Countdown clock and see which second the hand seems nearest to.

Re: Stattopalooza

Posted: Mon Oct 29, 2012 11:18 pm
by Graeme Cole
Thomas Carey wrote:
James Robinson wrote:
Mike Brown wrote:Surely not? I reckon most people would at least post to the nearest second, but maybe I have overly-high expectations. Personally, I'm in the 'nearest quarter-second' brigade. :)
I'd like to state that I'm in Mike's brigade too. :D
I think most recappers (me included) go for the nearest second. I always look at the Countdown clock and see which second the hand seems nearest to.
I do this too, because the recap writer only accepts two characters in the field. I might sometimes change it in the posted version to make it more precise though.

Re: Stattopalooza

Posted: Tue Oct 30, 2012 3:16 pm
by Adam Gillard
Could you find out the record for most incorrect letters declarations in a 15-round game? I just want to know if Mavis holds a record (http://wiki.apterous.org/Episode_5188)!

Re: Stattopalooza

Posted: Tue Oct 30, 2012 5:57 pm
by James Robinson
Adam Gillard wrote:Could you find out the record for most incorrect letters declarations in a 15-round game? I just want to know if Mavis holds a record (http://wiki.apterous.org/Episode_5188)!
There was a guy called Omair Azam in Series 55 who did one better :!: http://wiki.apterous.org/Episode_4163

Re: Stattopalooza

Posted: Tue Oct 30, 2012 7:23 pm
by Graeme Cole
James Robinson wrote:
Adam Gillard wrote:Could you find out the record for most incorrect letters declarations in a 15-round game? I just want to know if Mavis holds a record (http://wiki.apterous.org/Episode_5188)!
There was a guy called Omair Azam in Series 55 who did one better :!: http://wiki.apterous.org/Episode_4163
14 people have had five words disallowed in one game, four have had six, and one has managed seven.

Re: Stattopalooza

Posted: Tue Oct 30, 2012 7:30 pm
by Adam Gillard
Thanks Graeme!

Re: Stattopalooza

Posted: Thu Nov 01, 2012 2:44 pm
by Jon Corby
Graeme Cole wrote:I do this too, because the recap writer only accepts two characters in the field. I might sometimes change it in the posted version to make it more precise though.
Well, yeah, but do you think everyone else has always done that ever? Can't see another explanation that makes sense to me.

Re: Stattopalooza

Posted: Thu Nov 01, 2012 3:16 pm
by Nick Deller
Can you do a graph of buzz times for each five year period or something along those lines? If the 10-second peak only occurs in the 1980s and maybe 1990s, it would definitely suggest recording error.

Re: Stattopalooza

Posted: Sat Nov 03, 2012 12:23 am
by Graeme Cole
Nick Deller wrote:Can you do a graph of buzz times for each five year period or something along those lines? If the 10-second peak only occurs in the 1980s and maybe 1990s, it would definitely suggest recording error.
Image

Looks like we see the 10-second peak in 1987-1992, 1997-2002 and 2002-2007. Not so much 1992-1997.

Not sure what to conclude from it, but there's the graph anyway.

Re: Stattopalooza

Posted: Sat Nov 03, 2012 9:31 am
by Charlie Reams
This looks pretty cool. My long-term goal has really been to get all the games into machine-readable format so we can do whatever we like with them, including this sort of thing. The next step would be to make that data more widely available, which I'm working on now -- it looks like Graeme and I probably duplicated a lot of effort in doing this and the Classic games update.