TTC  Modern Draft Statistics
TTC  Modern Draft Statistics
The crew shares some fascinating statistics on the winningest (and crappiest) cards in Modern Masters draft.
 Lurkon
 Posts: 1027
 Joined: 09 Mar 2013, 09:50
 First Video: Man Cooking: Meatshroom
 Location: South Carolina, USA
Re: TTC  Modern Draft Statistics
It is really fun listening to you guys dance around trying not to tell us that running a card will win us games.
This was a lot of fun to listen to, I'd love to see these kinds of statistics for sets I've actually played with.
This was a lot of fun to listen to, I'd love to see these kinds of statistics for sets I've actually played with.
Do a Pivot Table!
"Cover me in lube and call me an athlete!" ~ Kathleen De Vere
"Cover me in lube and call me an athlete!" ~ Kathleen De Vere
Re: TTC  Modern Draft Statistics
Well, we didn't want to intentionally misinterpret the data.
It's very easy to see how Tromp has a high win percentage, because if you can't win with it, you don't cast it. Still a powerful card.
It's very easy to see how Tromp has a high win percentage, because if you can't win with it, you don't cast it. Still a powerful card.
Re: TTC  Modern Draft Statistics
Just saw the video, and wanted to pop in and say thanks for the compliments. I enjoyed the statistics, and glad you put them to good use =)
Daniel
Daniel
Creator of MTGBot, a twitch IRC bot designed to help broadcasters streaming Magic: the Gathering.
 tamaness
 Posts: 2673
 Joined: 17 Oct 2008, 03:44
 First Video: LRReview: Desert Bus
 Location: Stuck between a rock and a hard place
 Contact:
Re: TTC  Modern Draft Statistics
I'm pretty sure that's Crovax on Volrath's Motion Sensor.
Formerly madAlric
All of the Crapshots in one place
All of the Crapshots in one place
 Lord Hosk
 Posts: 6587
 Joined: 07 Dec 2011, 08:30
 First Video: Checkpoint: Into the breach
 Location: Half and inch below the knuckle of the ring finger. MI
Re: TTC  Modern Draft Statistics
I would love to see if they could get stats of "winner" of a drafts rating and "average draft rating" my sample was Very small compared to theirs I assume it was massively automated.
Beware Bering Crystal Bears, Bearing Crystals. (Especially if the crystals they are bearing are, themselves, Bering Crystal Bears.) Old, Stupid Proverb
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
Re: TTC  Modern Draft Statistics
Yay, more TTC!
Loved the episode, the stats and the stories. Great as always.
Also: the whacky thing for Organ Harvest is that your teammates can sacrifice creatures to get the effect, which is pretty awkward as they are usually in completely different games...
Loved the episode, the stats and the stories. Great as always.
Also: the whacky thing for Organ Harvest is that your teammates can sacrifice creatures to get the effect, which is pretty awkward as they are usually in completely different games...
"if it ain't shiny, rub it on your hiney"

 Posts: 10
 Joined: 12 Jul 2013, 00:05
 First Video: TTC
Re: TTC  Modern Draft Statistics
Lord Hosk wrote:I would love to see if they could get stats of "winner" of a drafts rating and "average draft rating" my sample was Very small compared to theirs I assume it was massively automated.
Here you go.
A graph of user count v rating over the sample supplied to Graham.
http://imgur.com/WFWe1B4
bunched to the left as added <1500 together
Re: TTC  Modern Draft Statistics
I can confirm that ^^ mtgratingtester ^^ is the source for these delicious stats
 Lord Hosk
 Posts: 6587
 Joined: 07 Dec 2011, 08:30
 First Video: Checkpoint: Into the breach
 Location: Half and inch below the knuckle of the ring finger. MI
Re: TTC  Modern Draft Statistics
What I mean is that I went in and manually looked at all the RTR drafts 4 times over 2 days, and typed !rating (name) of everyone in the draft then averaged it to get this.
Average rating of:
Swiss
Top two: 1642
Whole Field: 1630
4322
Top two: 1680
Whole Feild: 1642
84
Top Two: 1737
Whole Field: 1653
Could you set it up to check ALL the data like you did here for the rating of the Winner of the draft, the top two of the draft and ALL participants of the draft? Or something like that? My data was only from the limited number of drafts I could do by opening the draft, typing the name, writing down the number moving on... I would love to see if it holds up over the massiveness you are able to grab.
Average rating of:
Swiss
Top two: 1642
Whole Field: 1630
4322
Top two: 1680
Whole Feild: 1642
84
Top Two: 1737
Whole Field: 1653
Could you set it up to check ALL the data like you did here for the rating of the Winner of the draft, the top two of the draft and ALL participants of the draft? Or something like that? My data was only from the limited number of drafts I could do by opening the draft, typing the name, writing down the number moving on... I would love to see if it holds up over the massiveness you are able to grab.
Beware Bering Crystal Bears, Bearing Crystals. (Especially if the crystals they are bearing are, themselves, Bering Crystal Bears.) Old, Stupid Proverb
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
 Utilitarian
 Posts: 1211
 Joined: 05 Nov 2012, 00:16
 First Video: Checkpoint 01
 Location: Burnaby, BC
Re: TTC  Modern Draft Statistics
I object to Tromp being called the most ridiculous Overrun effect
That status goes to Overwhelming Stampede, which I can't believe was ever printed.
It's in DotP 2014 ^_^
That status goes to Overwhelming Stampede, which I can't believe was ever printed.
It's in DotP 2014 ^_^
Look at the cards. LOOK AT THE CARDS!
The Saga of Ballchannels: My Dwarf Fortress Let's Play
The Saga of Ballchannels: My Dwarf Fortress Let's Play

 Posts: 10
 Joined: 12 Jul 2013, 00:05
 First Video: TTC
Re: TTC  Modern Draft Statistics
Lord Hosk wrote:What I mean is that I went in and manually looked at all the RTR drafts 4 times over 2 days, and typed !rating (name) of everyone in the draft then averaged it to get this.
Average rating of:
Swiss
Top two: 1642
Whole Field: 1630
4322
Top two: 1680
Whole Feild: 1642
84
Top Two: 1737
Whole Field: 1653
Could you set it up to check ALL the data like you did here for the rating of the Winner of the draft, the top two of the draft and ALL participants of the draft? Or something like that? My data was only from the limited number of drafts I could do by opening the draft, typing the name, writing down the number moving on... I would love to see if it holds up over the massiveness you are able to grab.
You were very close for most of them. This is over the 20k games:
84
Avg 1731
Top2 1740
STDEV 76
4322
Avg 1683
Top2 1701
STDEV 80.5
Swiss
Avg 1671
Top2 1698
STDEV 86.2
 Lurkon
 Posts: 1027
 Joined: 09 Mar 2013, 09:50
 First Video: Man Cooking: Meatshroom
 Location: South Carolina, USA
Re: TTC  Modern Draft Statistics
Interesting. As one would expect, there's an increase in player rating when you jump up to 84, but there isn't much of a difference between the ratings of players of Swiss and 4322.
Do a Pivot Table!
"Cover me in lube and call me an athlete!" ~ Kathleen De Vere
"Cover me in lube and call me an athlete!" ~ Kathleen De Vere
Re: TTC  Modern Draft Statistics
The same folks play Swiss and 4322 because they don't have the confidence to face the 84.
But if you're winning Swiss matches, it's much better value to try the 84 queues.
But if you're winning Swiss matches, it's much better value to try the 84 queues.
Re: TTC  Modern Draft Statistics
I made a script that estimates some of the statistics using Bayesian probability theory.
Results:
'90% confidence interval' means that the model is 90% certain that the true value is in that range.
Measure of being a good card: appearing in winning decks.
Results:
Code: Select all
Cloudgoat Ranger: 1169 wins, 604 losses
Probability of being in a winning deck: 65.915%
90% confidence interval: [64.055%; 67.755%]
Probability of being a good card: 100.000%
Progenitus: 4 wins, 3 losses
Probability of being in a winning deck: 55.556%
90% confidence interval: [28.924%; 80.710%]
Probability of being a good card: 63.672%
Bridge from Below: 0 wins, 6 losses
Probability of being in a winning deck: 12.500%
90% confidence interval: [0.730%; 34.816%]
Probability of being a good card: 0.781%
Blood Moon: 0 wins, 3 losses
Probability of being in a winning deck: 20.000%
90% confidence interval: [1.274%; 52.713%]
Probability of being a good card: 6.250%
Bonesplitter: 2650 wins, 2371 losses
Probability of being in a winning deck: 52.777%
90% confidence interval: [51.618%; 53.935%]
Probability of being a good card: 99.996%
'90% confidence interval' means that the model is 90% certain that the true value is in that range.
Measure of being a good card: appearing in winning decks.
Code: Select all
< qrpth> !explain Desert Bus 2
<@lrrbot> James Turner

 Posts: 10
 Joined: 12 Jul 2013, 00:05
 First Video: TTC
Re: TTC  Modern Draft Statistics
qrpth wrote:I made a script that estimates some of the statistics using Bayesian probability theory.
Results:Code: Select all
Cloudgoat Ranger: 1169 wins, 604 losses
Probability of being in a winning deck: 65.915%
90% confidence interval: [64.055%; 67.755%]
Probability of being a good card: 100.000%
Progenitus: 4 wins, 3 losses
Probability of being in a winning deck: 55.556%
90% confidence interval: [28.924%; 80.710%]
Probability of being a good card: 63.672%
Bridge from Below: 0 wins, 6 losses
Probability of being in a winning deck: 12.500%
90% confidence interval: [0.730%; 34.816%]
Probability of being a good card: 0.781%
Blood Moon: 0 wins, 3 losses
Probability of being in a winning deck: 20.000%
90% confidence interval: [1.274%; 52.713%]
Probability of being a good card: 6.250%
Bonesplitter: 2650 wins, 2371 losses
Probability of being in a winning deck: 52.777%
90% confidence interval: [51.618%; 53.935%]
Probability of being a good card: 99.996%
'90% confidence interval' means that the model is 90% certain that the true value is in that range.
Measure of being a good card: appearing in winning decks.
not exactly sure how you calculate Probablility of being a good card?
Full data available here
http://ge.tt/5Rdvfgl/v/0
Re: TTC  Modern Draft Statistics
mtgratingtester wrote:not exactly sure how you calculate Probablility of being a good card?
Integral of the posterior probability density function from 0.5 to 1.0.
mtgratingtester wrote:Full data available here
http://ge.tt/5Rdvfgl/v/0
I ran my script on the full dataset and here are the top 30 cards:
Code: Select all
1: Vedalken Shackles (rating: 100.00000000000000%; win prob: 72.836%; rarity: M)
2: Vedalken Dismisser (rating: 100.00000000000000%; win prob: 62.509%; rarity: C)
3: Tromp the Domains (rating: 100.00000000000000%; win prob: 85.366%; rarity: U)
4: Tidehollow Sculler (rating: 100.00000000000000%; win prob: 62.365%; rarity: U)
5: Skeletal Vampire (rating: 100.00000000000000%; win prob: 69.407%; rarity: R)
6: Sanctum Gargoyle (rating: 100.00000000000000%; win prob: 57.291%; rarity: C)
7: Rude Awakening (rating: 100.00000000000000%; win prob: 78.500%; rarity: R)
8: Riftwing Cloudskate (rating: 100.00000000000000%; win prob: 60.647%; rarity: U)
9: Pestermite (rating: 100.00000000000000%; win prob: 57.560%; rarity: C)
10: Murderous Redcap (rating: 100.00000000000000%; win prob: 59.760%; rarity: U)
11: Mulldrifter (rating: 100.00000000000000%; win prob: 59.636%; rarity: U)
12: Marsh Flitter (rating: 100.00000000000000%; win prob: 63.798%; rarity: U)
13: Latchkey Faerie (rating: 100.00000000000000%; win prob: 58.289%; rarity: C)
14: Imperiosaur (rating: 100.00000000000000%; win prob: 58.588%; rarity: C)
15: Faerie Mechanist (rating: 100.00000000000000%; win prob: 57.407%; rarity: C)
16: Errant Ephemeron (rating: 100.00000000000000%; win prob: 59.480%; rarity: C)
17: Cloudgoat Ranger (rating: 100.00000000000000%; win prob: 65.915%; rarity: U)
18: Amrou Scout (rating: 100.00000000000000%; win prob: 58.692%; rarity: C)
19: Æthersnipe (rating: 99.99999999999999%; win prob: 57.915%; rarity: C)
20: Thundercloud Shaman (rating: 99.99999999999979%; win prob: 61.179%; rarity: U)
21: Oona, Queen of the Fae (rating: 99.99999999999964%; win prob: 65.979%; rarity: R)
22: Sporesower Thallid (rating: 99.99999999999946%; win prob: 58.369%; rarity: U)
23: Stir the Pride (rating: 99.99999999999913%; win prob: 65.854%; rarity: U)
24: Sword of Light and Shadow (rating: 99.99999999999429%; win prob: 68.039%; rarity: M)
25: Incremental Growth (rating: 99.99999999999137%; win prob: 63.525%; rarity: U)
26: Electrolyze (rating: 99.99999999999069%; win prob: 59.060%; rarity: U)
27: Flickerwisp (rating: 99.99999999998536%; win prob: 58.796%; rarity: U)
28: Sporoloth Ancient (rating: 99.99999999983821%; win prob: 56.418%; rarity: C)
29: Meloku the Clouded Mirror (rating: 99.99999999976612%; win prob: 63.551%; rarity: R)
30: Sword of Fire and Ice (rating: 99.99999999976504%; win prob: 66.828%; rarity: M)
Code: Select all
188: Crush Underfoot (rating: 2.24002136580020%; win prob: 47.043%; rarity: C)
189: Arcbound Wanderer (rating: 1.98343091713762%; win prob: 43.986%; rarity: C)
190: Fury Charm (rating: 1.52030239592511%; win prob: 45.032%; rarity: C)
191: Festering Goblin (rating: 1.43924037588680%; win prob: 47.658%; rarity: C)
192: Bridge from Below (rating: 0.78125000000000%; win prob: 12.500%; rarity: R)
193: Greater Gargadon (rating: 0.74163412990679%; win prob: 42.807%; rarity: R)
194: Extirpate (rating: 0.63629150390625%; win prob: 22.222%; rarity: R)
195: Summoner's Pact (rating: 0.50316666327932%; win prob: 37.500%; rarity: R)
196: Reach Through Mists (rating: 0.28426579246221%; win prob: 47.061%; rarity: C)
197: Greater Mossdog (rating: 0.26329227955340%; win prob: 45.822%; rarity: C)
198: Sylvan Bounty (rating: 0.20124226759130%; win prob: 40.891%; rarity: C)
199: Angel's Grace (rating: 0.04561170935631%; win prob: 20.690%; rarity: R)
200: Perilous Research (rating: 0.04241730067842%; win prob: 44.763%; rarity: C)
201: Lava Spike (rating: 0.03768397384116%; win prob: 45.574%; rarity: C)
202: Æther Vial (rating: 0.02145202542113%; win prob: 41.996%; rarity: R)
203: Manamorphose (rating: 0.01284628212620%; win prob: 44.989%; rarity: U)
204: Paradise Mantle (rating: 0.01156059400594%; win prob: 44.634%; rarity: U)
205: Arcbound Stinger (rating: 0.00632045415149%; win prob: 46.586%; rarity: C)
206: Relic of Progenitus (rating: 0.00572821815057%; win prob: 39.016%; rarity: U)
207: Stingscourger (rating: 0.00172395455853%; win prob: 46.124%; rarity: C)
208: Narcomoeba (rating: 0.00171781182065%; win prob: 36.596%; rarity: U)
209: Stinkweed Imp (rating: 0.00067637976677%; win prob: 46.458%; rarity: C)
210: Jhoira of the Ghitu (rating: 0.00013366427432%; win prob: 28.814%; rarity: R)
211: Rift Elemental (rating: 0.00008952249253%; win prob: 41.898%; rarity: C)
212: Syphon Life (rating: 0.00006509801909%; win prob: 42.128%; rarity: C)
213: Squee, Goblin Nabob (rating: 0.00003111605357%; win prob: 39.114%; rarity: R)
214: Hammerheim Deadeye (rating: 0.00000415762164%; win prob: 42.230%; rarity: C)
215: Hana Kami (rating: 0.00000133099942%; win prob: 37.576%; rarity: C)
216: Runed Stalactite (rating: 0.00000006253135%; win prob: 41.698%; rarity: C)
217: Stinkdrinker Daredevil (rating: 0.00000000005429%; win prob: 42.377%; rarity: C)
Code: Select all
< qrpth> !explain Desert Bus 2
<@lrrbot> James Turner

 Posts: 10
 Joined: 12 Jul 2013, 00:05
 First Video: TTC
Re: TTC  Modern Draft Statistics
more data
http://ge.tt/1MWbjhl/v/0
http://ge.tt/1MWbjhl/v/0
Code: Select all
Total Games 20928
Avg Turns 9.0
Avg Lands 6.2
Win % on Play 49.9%
Win % on Draw 50.1%
Mulligan to 6 5 4
Win% Average 36.4% 21.6% 14.8%
Win% on play 35.9% 19.0% 16.7%
Win% on draw 36.9% 24.6% 13.3%
Name Played % of Field Win%
WUB (Esper) 908 2.5% 52.6%
UGR 870 2.4% 51.1%
BGR (Jund) 533 1.5% 48.2%
WGR 906 2.5% 46.9%
WUG 817 2.2% 54.1%
UBG 671 1.8% 50.4%
WBR 473 1.3% 51.8%
WR (Boros) 4173 11.5% 51.0%
WUR 1134 3.1% 49.6%
UBR (Grixis) 827 2.3% 52.5%
BR (Rakdos) 2641 7.3% 48.5%
BG (Golgari) 1589 4.4% 48.6%
WBG 666 1.8% 48.3%
UG (Simic) 2031 5.6% 50.8%
WG (Selesnya) 3099 8.5% 49.8%
WU (Azorios) 3596 9.9% 55.2%
GR (Gruul) 1950 5.4% 45.0%
UR (Izzet) 2843 7.8% 48.8%
WB (Orzhov) 2902 8.0% 55.3%
UB (Dimir) 3702 10.2% 52.1%
 Lord Hosk
 Posts: 6587
 Joined: 07 Dec 2011, 08:30
 First Video: Checkpoint: Into the breach
 Location: Half and inch below the knuckle of the ring finger. MI
Re: TTC  Modern Draft Statistics
Graham wrote:The same folks play Swiss and 4322 because they don't have the confidence to face the 84.
But if you're winning Swiss matches, it's much better value to try the 84 queues.
When I ran all my games I had much better return on the 4322. in 16 drafts of each I got back 46 packs in 4322. 39 packs in swiss, but only 32 in 84.
There is a definite jump in skill level to the 84s which explains why I got back almost a pack less on average.
Beware Bering Crystal Bears, Bearing Crystals. (Especially if the crystals they are bearing are, themselves, Bering Crystal Bears.) Old, Stupid Proverb
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
 Lord Hosk
 Posts: 6587
 Joined: 07 Dec 2011, 08:30
 First Video: Checkpoint: Into the breach
 Location: Half and inch below the knuckle of the ring finger. MI
Re: TTC  Modern Draft Statistics
mtgratingtester wrote:
You were very close for most of them. This is over the 20k games:
84
Avg 1731
Top2 1740
STDEV 76
4322
Avg 1683
Top2 1701
STDEV 80.5
Swiss
Avg 1671
Top2 1698
STDEV 86.2
Thank you very much! is that RTR, MM or all drafts? I dont know if it would matter im just curious about stats.
Beware Bering Crystal Bears, Bearing Crystals. (Especially if the crystals they are bearing are, themselves, Bering Crystal Bears.) Old, Stupid Proverb
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
 phlip
 Posts: 1789
 Joined: 24 Apr 2010, 17:48
 First Video: Eternal Sonata (Unskippable)
 Location: Australia
Re: TTC  Modern Draft Statistics
Lord Hosk wrote:Thank you very much! is that RTR, MM or all drafts? I dont know if it would matter im just curious about stats.
This is what I was just about to ask, too... If you have the data for a "normal" (nonMMA) draft format, like RTR block, I'd love to run some analysis on it re: the 4322 EV debate... (I could do the same analysis for MMA figures, but it would be less likely to be widely applicable, since I suspect the format drew more better players.)
While no one overhear you quickly tell me not cow cow.
but how about watch phone?
[he/him/his]
but how about watch phone?
[he/him/his]
 Lord Hosk
 Posts: 6587
 Joined: 07 Dec 2011, 08:30
 First Video: Checkpoint: Into the breach
 Location: Half and inch below the knuckle of the ring finger. MI
Re: TTC  Modern Draft Statistics
That was exactly what I was thinking, Modern Masters had a different group of players than "standard" "current" stuff.
My big complaint about the bad EV arguments for 4322 because "You are just throwing away a pack because they pay less" There is a clear skill difference and skill matters in magic the gathering, meaning that slightly less skilled players have a better chance of winning and a much better chance of getting packs back in a 4322 than a 84.
It is extremely unlikely that you will be able to go infinite on 4322's no matter how good you are, and next to impossible to come out ahead. If you are very good either are possible in 84's.
My experience would show that you can slow you bleeding and get many more games in for the investment though if you are below the "play forever" skill level.
My big complaint about the bad EV arguments for 4322 because "You are just throwing away a pack because they pay less" There is a clear skill difference and skill matters in magic the gathering, meaning that slightly less skilled players have a better chance of winning and a much better chance of getting packs back in a 4322 than a 84.
It is extremely unlikely that you will be able to go infinite on 4322's no matter how good you are, and next to impossible to come out ahead. If you are very good either are possible in 84's.
My experience would show that you can slow you bleeding and get many more games in for the investment though if you are below the "play forever" skill level.
Beware Bering Crystal Bears, Bearing Crystals. (Especially if the crystals they are bearing are, themselves, Bering Crystal Bears.) Old, Stupid Proverb
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!

 Posts: 10
 Joined: 12 Jul 2013, 00:05
 First Video: TTC
Re: TTC  Modern Draft Statistics
That depends on whether the LRR guys want to do a show on the DGR data first, but happy to publish it after.
 Lord Hosk
 Posts: 6587
 Joined: 07 Dec 2011, 08:30
 First Video: Checkpoint: Into the breach
 Location: Half and inch below the knuckle of the ring finger. MI
Re: TTC  Modern Draft Statistics
So those ratings numbers are for Modern Masters?
Beware Bering Crystal Bears, Bearing Crystals. (Especially if the crystals they are bearing are, themselves, Bering Crystal Bears.) Old, Stupid Proverb
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
[–]Graham_LRR
You hear that Khoo? We're almost better than the comic!
Re: TTC  Modern Draft Statistics
These data are pretty cool. I'm not going to take the time to dig them out, but there was 1 common that was never cast, 6 uncommons, 5 rares and no mythics.
In 19,243 drafts, each draft saw about 16750 (+240) copies of each common (fudging for foil lands), 7700 (+56) of each uncommon, 2540 (+22) of each rare and 1280 (+9) of each mythic. Recall even though mythics are 1 out of 8 packs, there are only 15 mythics compared to 53 rares which accounts for the counterintuitive result.
It may be possible to back out an estimate on the proportion of times each card was included in a deck, but it would be a pretty wild estimate making some pretty strong assumptions.
Other fun statistics:
11 games were won by a player having played only 1 land
56 with 2 lands
402 with 3 lands
1638 with 4 lands
3917 with 5 lands
5269 with 6 lands
...
21 games with 15 or more lands
4 games with 19 or more lands
And Zempre won a game on turn 25 having played 20 lands.
@qrpth if those are Bayesian estimates, they're credible intervals, not confidence intervals. What does "Measure of being a good card: appearing in winning decks." mean, anyway? What was your prior distribution that gave you credible intervals extending well above 0 when the empirical evidence was Bridge From Below was 0for6?
In 19,243 drafts, each draft saw about 16750 (+240) copies of each common (fudging for foil lands), 7700 (+56) of each uncommon, 2540 (+22) of each rare and 1280 (+9) of each mythic. Recall even though mythics are 1 out of 8 packs, there are only 15 mythics compared to 53 rares which accounts for the counterintuitive result.
It may be possible to back out an estimate on the proportion of times each card was included in a deck, but it would be a pretty wild estimate making some pretty strong assumptions.
Other fun statistics:
11 games were won by a player having played only 1 land
56 with 2 lands
402 with 3 lands
1638 with 4 lands
3917 with 5 lands
5269 with 6 lands
...
21 games with 15 or more lands
4 games with 19 or more lands
And Zempre won a game on turn 25 having played 20 lands.
@qrpth if those are Bayesian estimates, they're credible intervals, not confidence intervals. What does "Measure of being a good card: appearing in winning decks." mean, anyway? What was your prior distribution that gave you credible intervals extending well above 0 when the empirical evidence was Bridge From Below was 0for6?
ecocd  echoseedee (We apologize for the inconvenience)
Who is online
Users browsing this forum: No registered users and 1 guest