Page 11 of 30
Re: Ask Graeme?
Posted: Tue Nov 05, 2013 11:32 pm
by Graeme Cole
Jordan F wrote:*Again, by R-M% stat, who was the best contestant to never make the finals? To make life easier, I'll also add the criteria of at least 5 games won.
Top 10 non-finalists by raw score divided by max score, where the player won at least five games:
Code: Select all
Marie Hayden 83.64
Greg Hayhurst 81.81
Matt Croy 81.56
Judith Armstrong 81.43
Phyllis Styles 81.05
Dee Voce 80.48
Clive Johnson 79.62
David Franks 79.27
Ian Graham 79.21
Richard Harris 78.60
Also, of anyone who made their Countdown debut and lost, who got the most maxes?
In 15 rounds,
David Morgans,
Paul Lyne and
Christine Smith lost their first games with 10 maxes. In 9 rounds,
Tom Bradshaw and
Gareth Fuller maxed seven rounds on their debuts and lost.
Re: Ask Graeme?
Posted: Tue Nov 05, 2013 11:39 pm
by Graeme Cole
Jack Worsley wrote:How many people have offered their own name (can be first name or surname) on the show?
I think it happened in the James Martin 952 episode but have there been any more cases? Thanks.
It's happened twice. One was the word
FOSTER by Gerald Foster in the episode you mention. The other was
BROWN by Chris Brown.
Re: Ask Graeme?
Posted: Wed Nov 06, 2013 12:05 am
by Graeme Cole
Dave Preece wrote:Who are the best players, say top five, not to make a series finals. Going by the current finals qualification rules IE players are sorted by number of wins then by number of points. Points from both wins and losses count.
Top 10 non-finalists in 15 rounds:
Code: Select all
WINS PTS
David Franks 7 744
Peter Coutts 7 662
Phyllis Styles 6 576
Ryan Loughborough 6 554
Marie Hayden 5 593
Sheri Evans 5 551
Ian Graham 5 547
Greg Hayhurst 5 540
Matt Croy 5 538
Richard Harris 5 535
Note that Michael Chadwick did qualify for the series 63 finals with 7 wins and 706 points, but as I recall he ran off to Singapore at very short notice.
Graham Hill qualified for the series 64 finals with 6 wins and 577 points, but then booked a holiday over the finals or something like that.
Top 10 non-finalists in 9 rounds:
Code: Select all
WINS PTS
Dee Voce 5 356
Kay Powick 5 318
Rita Bean 5 310
John Morrison 5 305
Daniel Holloway 5 300
Paul Kerry 5 281
Graham Bucknall 4 290
John Snedden 4 282
Stephen Dodds 4 281
Denis Allan 4 281
Sheila Mann, whom some of you might remember from the Edinburgh Countdown Club, was #4 seed in series 2 with 6 wins and 299 points. She didn't reach the quarter finals, but she was defeated in the last 16, or "qualifying preliminaries" which they had in series 2.
Re: Ask Graeme?
Posted: Wed Nov 06, 2013 12:24 am
by Graeme Cole
sean d wrote:Has anyone ever declared a numbers solution of over 1,000? There are at least a couple of combinations that yield a 'max' solution of just over 1,000 for a high target just under 1,000 (eg 1 3 4 4 7 7 -- 999. Best solution is 1,001)
There have been occasional joke declarations. Peter Medhurst declared 5200 in
this game. This was the record until very recently when Peter Etherington declared
a million and one. At the other end of the scale, there have been two declarations under 100, neither of them entirely serious: David Affleck declared
25 and Tim Davies declared 2 in the last numbers round of a
series 7 quarter-final.
Other than the above, there have only been two declarations greater than 1000, and they came in the same round of
this CoC game. One player declared 1001 and the other declared 1008. Tim Morrissey's declaration of 1008, while not a scoring declaration in this case, is 63*16 and 72*14, so it can be a more useful number than some might think.
Re: Ask Graeme?
Posted: Wed Nov 06, 2013 12:32 am
by Dave Preece
Ironically the 5,200 declared-game was VERY easy to solve, even for me!
Re: Ask Graeme?
Posted: Wed Nov 06, 2013 7:39 am
by Ben Wilson
Graeme Cole wrote:
Top 10 non-finalists in 9 rounds:
Code: Select all
WINS PTS
John Rawnsley 7 409
Geraldine Hylands 7 355
Craig Richardson 6 352
Phil Bennett 6 321
Weren't these four the finalists of series 45? Though I suppose technically they never reached the
quarter-finals...
Re: Ask Graeme?
Posted: Wed Nov 06, 2013 9:07 am
by Graeme Cole
Ben Wilson wrote:Graeme Cole wrote:
Top 10 non-finalists in 9 rounds:
Code: Select all
WINS PTS
John Rawnsley 7 409
Geraldine Hylands 7 355
Craig Richardson 6 352
Phil Bennett 6 321
Weren't these four the finalists of series 45? Though I suppose technically they never reached the
quarter-finals...
Well spotted. The query I ran just checked if they'd appeared in a quarter final. I'll edit the table.
Re: Ask Graeme?
Posted: Wed Nov 06, 2013 9:35 am
by Jon Corby
Dave Preece wrote:Ironically the 5,200 declared-game was VERY easy to solve, even for me!
This is the risk you take with joke declarations; if you just go "nope sorry, lost it", you might have been nearly there, you might have actually had it but got a bit confused at the last second, whatever. By declaring 5,200 you're sort of going "I haven't got it, and it's ridiculous for you to even ask", so you do end up with a bit of egg on your face if it solves easily. IMO.
Re: Ask Graeme?
Posted: Wed Nov 06, 2013 4:50 pm
by Dave Preece
Graeme Cole wrote:Dave Preece wrote:Who are the best players, say top five, not to make a series finals. Going by the current finals qualification rules IE players are sorted by number of wins then by number of points. Points from both wins and losses count.
Top 10 non-finalists in 15 rounds:
Code: Select all
WINS PTS
David Franks 7 744
Peter Coutts 7 662
Phyllis Styles 6 576
Ryan Loughborough 6 554
Marie Hayden 5 593
Sheri Evans 5 551
Ian Graham 5 547
Greg Hayhurst 5 540
Matt Croy 5 538
Richard Harris 5 535
Note that Michael Chadwick did qualify for the series 63 finals with 7 wins and 706 points, but as I recall he ran off to Singapore at very short notice.
Graham Hill qualified for the series 64 finals with 6 wins and 577 points, but then booked a holiday over the finals or something like that.
Top 10 non-finalists in 9 rounds:
Code: Select all
WINS PTS
Dee Voce 5 356
Kay Powick 5 318
Rita Bean 5 310
John Morrison 5 305
Daniel Holloway 5 300
Paul Kerry 5 281
Graham Bucknall 4 290
John Snedden 4 282
Stephen Dodds 4 281
Denis Allan 4 281
Sheila Mann, whom some of you might remember from the Edinburgh Countdown Club, was #4 seed in series 2 with 6 wins and 299 points. She didn't reach the quarter finals, but she was defeated in the last 16, or "qualifying preliminaries" which they had in series 2.
Thanking you Graeme!
An unusually high percentage of ladies in both lists, or not?
Re: Ask Graeme?
Posted: Wed Nov 06, 2013 7:09 pm
by Andy Platt
What are the mode and mean margins of victory?
And would it be possible to compare that to apterous?
Thanks
Re: Ask Graeme?
Posted: Sun Nov 10, 2013 12:20 pm
by David Williams
Graeme Cole wrote:sean d wrote:Has anyone ever declared a numbers solution of over 1,000?
there have only been two declarations greater than 1000, and they came in the same round of
this CoC game. One player declared 1001 and the other declared 1008. Tim Morrissey's declaration of 1008, while not a scoring declaration in this case, is 63*16 and 72*14, so it can be a more useful number than some might think.
Is it just me that is somehow very impressed that someone actually scored points with a declaration outside the range of possible targets?
Re: Ask Graeme?
Posted: Sun Nov 10, 2013 3:21 pm
by Gavin Chipper
What about declarations under 100?
Re: Ask Graeme?
Posted: Mon Nov 11, 2013 9:43 am
by Charlie Reams
Andy Platt wrote:What are the mode and mean margins of victory?
And would it be possible to compare that to apterous?
Thanks
For apterous, considering only bog-standard 15 Rounders (new format) in human-vs-human games and ignoring extra conundrums (i.e. if it's tied after 15 rounds, take the winning margin to be zero), the stats for winning margin are:
Mean: 23.85
Median: 20
Mode: 3 (1146 times)
Re: Ask Graeme?
Posted: Mon Nov 11, 2013 4:07 pm
by Dave Preece
I've never understood why mode is used, it can be very inaccurate compared to the mean?
Re: Ask Graeme?
Posted: Mon Nov 11, 2013 4:21 pm
by Gavin Chipper
Dave Preece wrote:I'e never understood why mode is used, it can be very inaccurate compared to the mean?
It has its uses.
Re: Ask Graeme?
Posted: Tue Nov 12, 2013 2:23 am
by Charlie Reams
Dave Preece wrote:I've never understood why mode is used, it can be very inaccurate compared to the mean?
http://en.wikipedia.org/wiki/Mode_(statistics)#Use
Re: Ask Graeme?
Posted: Tue Nov 12, 2013 12:46 pm
by Andy Platt
Nice work. Didn't think it'd be so close to be honest.
I had a feeling 17 might be a good shout for mode (slightly better player wins conundrum and one letters/tricky numbers). Interesting stuff
Re: Ask Graeme?
Posted: Tue Nov 12, 2013 2:04 pm
by Charlie Reams
Andy Platt wrote:
I had a feeling 17 might be a good shout for mode (slightly better player wins conundrum and one letters/tricky numbers). Interesting stuff
I think you're overweighting expert-level games. In evenly matched games of average-level players, they miss a lot of stuff and score a lot of winners. So it's very volatile. The mode is not very relevant here anyway, I only included it because you asked for it -- the next most common margin is 10 and then it's all over the place after that.
Re: Ask Graeme?
Posted: Tue Nov 12, 2013 2:59 pm
by Graeme Cole
Gavin Chipper wrote:What's the latest in the game that the first E has appeared? Today (massive spoiler), I think the first E appeared as the last letter of the third letters round, so the 27th letter. Are there any that beat that?
Episode 577 holds the record. The first E was the 33rd letter to come out.
In
episode M87 it was the 32nd letter.
In
episode 800 it was the 30th letter.
In
episode 1870 it was the 29th letter.
Episodes
560,
782,
1448 and
4002 match what we saw the other day, with the first E being the 27th letter out.
Re: Ask Graeme?
Posted: Tue Nov 12, 2013 3:12 pm
by Graeme Cole
Jordan F wrote:One other interesting question that I thought of just a minute ago.
Has anyone ever done what I'll refer to as a reverse Peter Lee/Philip Jarvis? In other words, has anyone ever been on the show in an initial appearance where they won at least one game, and then came back in another series and lost in their first(and only) appearance? For fairly obvious reasons I'm ruling out anyone who lost their first game in the Supreme Championship or 30th Birthday Championship.
Only
Bruce Lambert has won fewer games on a later run than on an earlier one.
The query I ran also gave me
Duncan Dale-Emberton and
Helen Wrigglesworth, because it looks to the database like they had two runs, but actually they both had a single run which happened not to be continuous. So they don't count.
Re: Ask Graeme?
Posted: Tue Nov 12, 2013 3:37 pm
by Graeme Cole
Andy Platt wrote:What are the mode and mean margins of victory?
And would it be possible to compare that to apterous?
Thanks
As with Charlie's apterous stats, if a game was decided by a tiebreak the margin is zero.
Mean margin in 9 rounders
Prelims: 16.97
Finals: 16.15
All: 16.79
Mean margin in old 15 rounders
Prelims: 29.40
Finals: 27.11
All: 28.95
Mean margin in new 15 rounders (up to the end of series 68)
Prelims: 42.20
Finals: 38.71
All: 41.77
Mode
I've not divided these into prelims and finals, because with only the finals there'd be so little data the mode won't mean much.
9 rounder: 10 (146 times), followed by 4 (131 times) and 7 (109 times)
Old 15 rounder: 5 (71 times), followed by 16 (65 times) and 3 (62 times)
New 15 rounder up to end of S68: 39 (3 times), followed by loads of different margins twice.
Re: Ask Graeme?
Posted: Tue Nov 12, 2013 3:38 pm
by Graeme Cole
Gavin Chipper wrote:What about declarations under 100?
There haven't been any besides the two I mentioned.
Re: Ask Graeme?
Posted: Tue Nov 12, 2013 3:41 pm
by Gavin Chipper
Graeme Cole wrote:Gavin Chipper wrote:What about declarations under 100?
There haven't been any besides the two I mentioned.
Oh yeah. For some reason I didn't read that bit.
Re: Ask Graeme?
Posted: Wed Nov 13, 2013 12:30 am
by sean d
Thanks Graeme. That's a big and increasing average margin of victory. A couple of factors, I suppose, are the increase in the number of really good top end players and possibly a drop off in the quality at the other end.... certainly they seem to be finding it harder to find the 200+ contestants required per year.
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 4:08 pm
by Martin Sinclair
How about a table which shows the top 10 players who have the highest % of max from their time on the show in a series? They can have played however many games. The equation for this should be: = (Total Points scored / Total max available) x 100. An example, let's look at Jack Hurst's run. He had 1276 points out of 1377. If I've done the calculation correctly, that should get 92.6652142338. Surely that's in the top 10?!
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 4:33 pm
by Fred Mumford
Graeme has already done a similar table, but it's buried in the spoilers thread for 10th September 2013, not coincidentally the day a great octorun ended.
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 4:35 pm
by Jon Corby
Fred Mumford wrote:Graeme has already done a similar table, but it's buried in the spoilers thread for 10th September 2013, not coincidentally the day a great octorun ended.
Are you sure that's the precise equation he used though?
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 4:37 pm
by Gavin Chipper
If they can play any number of games there will be loads of people on 100%.
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 4:40 pm
by Fred Mumford
Jon Corby wrote:Fred Mumford wrote:Are you sure that's the precise equation he used though?
No, but I thought it easier to use the word "similar" than dig out the thread to check......
If not, it will at least give a good pointer until Graeme replies though.
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 5:06 pm
by Martin Sinclair
How, Gavin?
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 6:23 pm
by Jon Corby
Are you an American Indian Martin?
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 7:31 pm
by Gavin Chipper
Martin Sinclair wrote:How, Gavin?
People who lost their first game but had easy numbers.
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 8:44 pm
by Martin Sinclair
It's impossible to get 100% max percentage if you lose, though. And no, I'm not, Jon.
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 9:20 pm
by Gavin Chipper
Martin Sinclair wrote:It's impossible to get 100% max percentage if you lose, though. And no, I'm not, Jon.
For some reason I thought you were talking about numbers rounds.
Re: Ask Graeme?
Posted: Thu Nov 14, 2013 9:40 pm
by Martin Sinclair
No, I didn't mention "numbers" once in it
Re: Ask Graeme?
Posted: Fri Nov 15, 2013 12:49 am
by Dave Preece
Not quite what you asked, but...
Graeme Cole wrote:Dave Preece wrote:Highest ever % of max???
I meant to look into this when I read this post a few days ago, but only got round to it now.
In a word, yes.
Jen got 952 out of 1024, which is 92.97%. This beats Dylan Taylor (974/1050 = 92.76%) who in turn beat Craig Beevers, who had held the record for years (907/987 = 91.89%).
However, there is the question of the new format. There are more numbers rounds now and you might say these are usually easier to max than letters rounds. So let's just look at letters...
Top ten 15-round octochamps ordered by percentage of max on letters in their heats. Jen's still top.
Code: Select all
LETTERS SCORE MAX %
1 Jen Steadman 597 627 95.22
2 Dylan Taylor 613 653 93.87
3 Julian Fell 650 701 92.72
4 Jack Hurst 655 717 91.35
5 Edward McCullagh 602 659 91.35
6 Craig Beevers 609 667 91.30
7 Giles Hutchings 609 669 91.03
8 Andy Platt 620 682 90.91
9 Kirk Bevins 644 711 90.58
10 Chris Davies 626 693 90.33
It's a similar story if we also include conundrums. But we can do better than that. If we adapt new 15-rounder scores to the old 15-round format, by multiplying letters scores by 11/10 and multiplying numbers scores by 3/4, then we can include letters, numbers and conundrums for everyone. And the table of octotals ordered by percentage of max looks like this...
Code: Select all
OCTOTAL MAX %
1 Jen Steadman 937.95 1007.45 93.10
2 Dylan Taylor 957.55 1036.05 92.42
3 Craig Beevers 907 987 91.89
4 Jack Hurst 946 1031 91.76
5 Edward McCullagh 896 979 91.52
6 Julian Fell 924 1015 91.03
7 Giles Hutchings 954.40 1049.15 90.97
8 Kirk Bevins 925 1031 89.72
9 David O'Donnell 880 995 88.44
10 Chris Davies 892 1013 88.06
Best octorun ever.
Re: Ask Graeme?
Posted: Fri Nov 15, 2013 12:50 am
by Dave Preece
Jen has the best total % of max if all games are considered, this of course may go down...
Re: Ask Graeme?
Posted: Sat Nov 30, 2013 9:31 pm
by Martin Sinclair
I've just done some research on something which I'm sure many will have briefly thought about doing (apologies if it already has and I'm an idiot). It's the calculation for what your score would've been if your opponent maxed every single round throughout. So, the way to do this is by taking all the rounds which you did score on (but didn't max on), and subtract the score which got in those rounds from your overall total. You can ignore any rounds which you failed to score on altogether. Also, I feel all the points which you scored on in the conundrum rounds need to be subtracted from the total, as your opponent could've hansforded you each time. Harsh, but true. I've looked up a few impressive octoruns, and here's how it currently stands.
1) Dylan Taylor, 823 (151 points lost). Octorun maximum: 1050
2) Jen Steadman, 786 (166 points lost). Octorun maximum: 1024
3) Giles Hutchings, 748 (217 points lost). Octorun maximum: 1060
4) Edward McCullagh, 744 (152 points lost). Octorun maximum: 979
5) Julian Fell, 737 (187 point lost). Octorun maximum:
However, some octoruns obviously have higher scores available in theirs than others, so perhaps it would be best to judge it by the lowest amount of points lost. As you can see, Dylan's still top. Another good way would be to measure the percentage achieved from your maximum with your new "impossible to get lower" score. With this, here's the new leaderboard:
1) Dylan Taylor: 78.380952381%
2) Jen Steadman: 76.7578125%
3) Edward McCullagh: 75.9959141982%
4) Jack Hurst: 71.4839961203%
5) Julian Fell: 71.4285714286%
Re: Ask Graeme?
Posted: Sat Nov 30, 2013 9:33 pm
by Martin Sinclair
which you got in those rounds*, "impossible to get lower than"*
Re: Ask Graeme?
Posted: Sun Dec 01, 2013 12:30 am
by Gavin Chipper
Martin Sinclair wrote:I've just done some research on something which I'm sure many will have briefly thought about doing (apologies if it already has and I'm an idiot). It's the calculation for what your score would've been if your opponent maxed every single round throughout. So, the way to do this is by taking all the rounds which you did score on (but didn't max on), and subtract the score which got in those rounds from your overall total. You can ignore any rounds which you failed to score on altogether. Also, I feel all the points which you scored on in the conundrum rounds need to be subtracted from the total, as your opponent could've hansforded you each time. Harsh, but true. I've looked up a few impressive octoruns, and here's how it currently stands.
1) Dylan Taylor, 823 (151 points lost). Octorun maximum: 1050
2) Jen Steadman, 786 (166 points lost). Octorun maximum: 1024
3) Giles Hutchings, 748 (217 points lost). Octorun maximum: 1060
4) Edward McCullagh, 744 (152 points lost). Octorun maximum: 979
5) Julian Fell, 737 (187 point lost). Octorun maximum:
However, some octoruns obviously have higher scores available in theirs than others, so perhaps it would be best to judge it by the lowest amount of points lost. As you can see, Dylan's still top. Another good way would be to measure the percentage achieved from your maximum with your new "impossible to get lower" score. With this, here's the new leaderboard:
1) Dylan Taylor: 78.380952381%
2) Jen Steadman: 76.7578125%
3) Edward McCullagh: 75.9959141982%
4) Jack Hurst: 71.4839961203%
5) Julian Fell: 71.4285714286%
This is all getting a bit complicated. Number of maxes is simpler and probably just as good. And because of your system, conundrums are totally ignored.
It would be interesting to see the like of Jen and Dylan added to the number of maxes list. Thanks Graeme!
Re: Ask Graeme?
Posted: Sun Dec 01, 2013 11:39 am
by Jennifer Steadman
Dylan got 101 maxes and I got 99.
Any form of ranking people by overall game performance should include conundrums. The fact that you can be beaten on them by anyone Hansfording quick enough doesn't make them irrelevant...
Re: Ask Graeme?
Posted: Sun Dec 01, 2013 12:07 pm
by Martin Sinclair
Yeah, the conundrum aspect is definitely debatable. If we were to include the points got from those, it would look like this:
1) Dylan Taylor - 873
2) Jennifer Steadman - 846
3) Giles Hutchings - 818
4) Edward McCullagh - 804
5) Jack Hurst - 797
But, this goes back to my point of "impossible to get lower than". The above aren't the correct figures. For example, in Dylan's run, he got a conundrum in 28 seconds, so he easily could have not got it.
Not sure what the best way of looking at this is
Maybe only do the calculations for letters and numbers?
Re: Ask Graeme?
Posted: Sun Dec 01, 2013 4:21 pm
by Martin Sinclair
If we are to do that, then the points will remain the same, but the percentages will probably increase, as there's a lower maximum available. The equation for this is: ("impossible to get lower than" score) / (octorun maximum - 80 points) x 100 = percentage. Here's the top 5:
1) Dylan Taylor: 84.8453608247%
2) Jen Steadman: 83.2627118644%
3) Edward McCullagh: 82.7586206897%
4) Julian Fell: 77.5401069519%
5) Jack Hurst: 77.4973711882%
Re: Ask Graeme?
Posted: Sun Dec 01, 2013 7:38 pm
by Dave Preece
Surely pure and simple points scored / available points x 100 is the best and only way to 'rank' players?
Re: Ask Graeme?
Posted: Sun Dec 01, 2013 7:42 pm
by Jon O'Neill
Dave Preece wrote:Surely pure and simple points scored / available points x 100 is the best and only way to 'rank' players?
I think you should multiply by 7.12349867123409781293 actually.
Re: Ask Graeme?
Posted: Sun Dec 01, 2013 8:06 pm
by Gavin Chipper
Jon O'Neill wrote:Dave Preece wrote:Surely pure and simple points scored / available points x 100 is the best and only way to 'rank' players?
I think you should multiply by 7.12349867123409781293 actually.
Well that's at least two methods already then.
Re: Ask Graeme?
Posted: Mon Dec 02, 2013 12:11 am
by Dave Preece
Jon O'Neill wrote:Dave Preece wrote:Surely pure and simple points scored / available points x 100 is the best and only way to 'rank' players?
I think you should multiply by 7.12349867123409781293 actually.
Inaccurate.
Re: Ask Graeme?
Posted: Mon Dec 02, 2013 12:27 am
by Gavin Chipper
At the moment, number of maxes seems to be considered a good way to judge things, but it ignores anything less than the max, so if the max is 8, then 5, 6, and 7 are all the same (worthless). So a way round this would be to if y points for the xth best word. So if there's 3 words that have 8 letters, 4 with 7 and 10 with 6, the words would be ranked at 1 (for 8 letters), 4 (for 7 letters), 8 (for 6 letters) and 18 (for 5 letters). The rank, x, then has to be converted into a score, y. You could have y=1/x, which I think would be quite neat. Another way of doing the rankings would be to take the middle rather than the top position for each rank. So if there's 3 words with 8 letters, they're all ranked at 2, so you'd get 1/2 points rather than 1. Or you could even go to the other extreme and give 1/3 points for a max.
Unfortunately recaps don't record the number of words at each length below the max, but we do have the number at the max, so we could use some formula based on statistical likelihood. It might be if there's 3 8s and that's the max, then on average there's 6 7s. So you could go by that.
For numbers, you could see how many solutions there are according to
here and base a system on that. But obviously how many solutions there are isn't an open and shut case.
I've written stuff about this before actually. See
here and
here.
Just make sure you multiply by 100 at the end.
Re: Ask Graeme?
Posted: Mon Dec 02, 2013 5:21 pm
by Dave Preece
Amazingly, not one of these five contestants returned for a second go, for whatever reason?
Re: Ask Graeme?
Posted: Tue Dec 03, 2013 2:10 pm
by Martin Sinclair
Yeah, as someone mentioned at the end of Jen's run, your amount of maximums is definitely a better way of measuring than % of overall maximum achieved. This is because, say, in a certain round, you could lose 18 points with that way of measuring. The collection of maximums is better as whenever you fail to get the max, you only lose 1 point, every time. The leaderboard for it could do with updating:
1) Dylan Taylor - 101/120 (84.1666666667%)
2) Jen Steadman - 99/120 (82.5%)
3) Edward McCullagh - 95/120 (79.1666666667%)
....
It's also good that I included the %, as I feel that can be more representative. For example, in Zarte's run, he only played 6 games, but got 61/90 maximums. That's 67.7777777778%, which is higher than anyone who got 81 or lower maximums from an octorun.
Jen and Dylan's octoruns are the two best, undoubtedly.
Re: Ask Graeme?
Posted: Tue Dec 03, 2013 5:24 pm
by Clive Brooker
This game, recently made famous, illustrates one of the problems with counting the number of maxes, at least whilst the data remains less than 100% perfect. The recap denies Tim Morrissey the max on round 5, but I don't believe SPREITEN was valid at the time. A system which awarded Tim 6/7 for that round as opposed to 0/1 would appear to be introducing less systematic bias against earlier players.
Re: Ask Graeme?
Posted: Tue Dec 03, 2013 5:33 pm
by Martin Sinclair
Yeah, this is an issue. I wonder if anyone's thought about going through someone's entire octorun with the ODO, making changes to valid/invalid words. Dylan's run would change, as he'd have OWLERIES allowed, giving him 982 and the highest % of maximum achieved. I'm sure plenty others would be affected too...
Re: Ask Graeme?
Posted: Tue Dec 03, 2013 5:37 pm
by Gavin Chipper
Clive Brooker wrote:This game, recently made famous, illustrates one of the problems with counting the number of maxes, at least whilst the data remains less than 100% perfect. The recap denies Tim Morrissey the max on round 5, but I don't believe SPREITEN was valid at the time. A system which awarded Tim 6/7 for that round as opposed to 0/1 would appear to be introducing less systematic bias against earlier players.
Yes, but that's just an error in the recap, and I thought people had gone through the earlier games with the old dictionary.
But in any case, I don't think errors from earlier games should entirely dictate the system we use. I think 6/7 gives it too much. I think someone who gets the max of 7 6 times in a row and blobs one round has done a better job than someone who gets 6 (when the max is 7) 7 times in a row.
I also think maxes is a poor measure for players who aren't that great. Sometimes a player will thrash another player but it might still be something like 3 all in maxes or 4-3 because all the non-maxed rounds where there is a difference between the two is ignored.
Re: Ask Graeme?
Posted: Tue Dec 03, 2013 6:46 pm
by Innis Carson
Martin Sinclair wrote:Yeah, this is an issue. I wonder if anyone's thought about going through someone's entire octorun with the ODO, making changes to valid/invalid words. Dylan's run would change, as he'd have OWLERIES allowed, giving him 982 and the highest % of maximum achieved. I'm sure plenty others would be affected too...
We don't yet know all the words valid in ODO that aren't in previous dictionaries, or vice versa, so it wouldn't be possible to calculate the % of maximum at present. And besides, it wouldn't be a very meaningful statistic anyway since nobody was trying to play by the ODO.
Re: Ask Graeme?
Posted: Wed Dec 04, 2013 12:57 am
by Martin Sinclair
If we were to do the letters and numbers thing, then the top 3 would look like this:
1) Dylan Taylor: 96/112 (85.7142857143%)
2) Jen Steadman: 93/112 (83.0357142857)
3) Edward McCullagh: 89/112 (79.4642857143%)
... This means that it's a guarantee that the 3 above would have score in at least the amount of rounds shown above from the letters and numbers.
Re: Ask Graeme?
Posted: Wed Dec 04, 2013 12:13 pm
by Clive Brooker
Gavin Chipper wrote:Yes, but that's just an error in the recap, and I thought people had gone through the earlier games with the old dictionary.
But in any case, I don't think errors from earlier games should entirely dictate the system we use. I think 6/7 gives it too much. I think someone who gets the max of 7 6 times in a row and blobs one round has done a better job than someone who gets 6 (when the max is 7) 7 times in a row.
I'm not sure if the old recaps have been checked against the then current dictionary in a systematic way. Also I don't think any which have been so checked (some recappers did this as a matter of course) have been marked accordingly.
I did of course mean 7/8, not that this affects the argument.
Blobbing a round is often caused by a player taking a risk to try and match an opponent's declaration. This is a problem to any system which uses actual scores, but with the counting maxes system the player is not additionally penalized for what may well have been good game-play.
Re: Ask Graeme?
Posted: Wed Dec 04, 2013 11:44 pm
by Jon O'Neill
Graeme Cole wrote:Here's the first bit - 15 round octochamps ordered by the number of maxes out of 120 in heat games. A maxed round is where the player got the most points available. Tiebreaks aren't counted.
The 9-rounder one is going to need a bit more thought to exclude people who had more than one run, so I'll come back to that one another day.
Code: Select all
1. Edward McCullagh 95
2. Julian Fell 94
3. Jack Hurst 91
4. Craig Beevers 89
4. Jonathan Rawlinson 89
6. Stewart Holden 86
7. Kirk Bevins 85
8. Chris Davies 84
9. David O'Donnell 82
10. Eoin Monaghan 81
11. Conor Travers 80
12. Chris Wills 79
12. Graeme Cole 79
14. Adam Gillard 78
14. George Greenhough 78
14. John Mayhew 78
14. Oliver Garner 78
14. Paul Gallen 78
19. Jack Welsby 77
20. Andrew Hulme 76
20. Innis Carson 76
20. Paul Howe 76
23. Peter Lee 75
24. Daniel Pati 74
25. Marcus Hares 73
26. Jimmy Gough 72
26. Martin Bishop 72
28. Chris Cummins 71
28. Ryan Taylor 71
28. Tom Barnes 71
31. Aaron Webber 70
31. David Barnard 70
31. Grace Page 70
34. Jon Corby 69
34. Richard Heald 69
36. Lee Hartley 68
36. Matthew Shore 68
38. Charlie Reams 67
38. Mark Deeks 67
38. Mark Tournoff 67
38. Neil Zussman 67
38. Paul James 67
38. Steven Briers 67
44. Jack Worsley 66
45. Andy McGurn 65
45. James Hurrell 65
45. Jeffrey Hansford 65
45. John Brackstone 65
45. Tom Rowell 65
50. Stuart Earl 64
50. Tom Hargreaves 64
52. John Hunt 63
52. Kevin Thurlow 63
52. Scott Gillies 63
55. John Davies 62
55. Junaid Mubeen 62
55. Stuart Solomons 62
55. Wendy Roe 62
59. Jean Webby 61
59. Jon O'Neill 61
59. Martin Gardner 61
62. Richard Brittain 60
62. Shane Roberts 60
64. Cate Henderson 59
64. John Gray 59
64. Jonathan Coles 59
64. Michael Bowden 59
64. Ross Allatt 59
64. Steven Moir 59
70. Brian Selway 58
70. David Edwards 58
70. Mike Pullin 58
70. Stu Horsey 58
74. Danny Hamilton 57
74. Jim Bentley 57
74. Keith Maynard 57
74. Liam Shaw 57
74. Rose Boyle 57
79. Nik Von Uexkull 56
80. Richard Pay 55
80. Rupert Stokoe 55
82. David Von Geyer 54
82. Jeffrey Burgin 54
82. Sweyn Kirkness 54
82. Tim Reypert 54
86. James Roberts 53
86. Kai Laddiman 53
86. Nick Wainwright 53
86. Paul Keane 53
90. Amey Deshpande 52
90. Jayne Wisniewski 52
90. Ned Pendleton 52
93. Carl Williams 51
93. Dave Taylor 51
93. Heather Styles 51
93. Michael Macdonald-Cooper 51
97. Gary Male 50
97. Judith Young 50
97. Julia Wilkinson 50
100. Danny Pledger 49
101. Tony Warren 48
102. James Doohan 46
103. Joe Zubaidi 44
104. Brenda Jolley 43
104. Tia Corkish 43
106. Chris Marshall 42
106. David Thirlwall 42
108. Chris McHenry 37
108. Steve Wood 37
108. Suzi Purcell 37
Edited to include Rose Boyle, David Barnard and Heather Styles from series 67, and Kevin Thurlow and Richard Pay who weren't in the list due to muppetry on my part.
Sorry for asking for an update, but I'd love to see this one up-to-date. Also if possible with what series they were from or what date they debuted
Re: Ask Graeme?
Posted: Thu Dec 05, 2013 1:04 pm
by JackHurst
Which contestant who played all of their games in the old format would have the biggest difference between their total score then, and their total score converted into new money (ie their numbers score multiplied by 4/3 and their letters score multiplied by 10/11)?
Re: Ask Graeme?
Posted: Thu Dec 05, 2013 1:59 pm
by Zarte Siempre
JackHurst wrote:Which contestant who played all of their games in the old format would have the biggest difference between their total score then, and their total score converted into new money (ie their numbers score multiplied by 4/3 and their letters score multiplied by 10/11)?
Now that's a fucking question!
Re: Ask Graeme?
Posted: Thu Dec 05, 2013 2:00 pm
by Gavin Chipper
JackHurst wrote:Which contestant who played all of their games in the old format would have the biggest difference between their total score then, and their total score converted into new money (ie their numbers score multiplied by 4/3 and their letters score multiplied by 10/11)?
What's your score in new money? (It might be on here somewhere but I can't be bothered to look.)