Post by: majofa on January 10, 2014, 12:12:32 am
Then each player will build 1 deck to face the decks of the other 4 players in their group.
The first deck has a maximum deck size of 40 cards; in addition, the deck will have 2x mark and 150hp.
The second deck: No upgraded cards - Maximum 4 copies of non-pillar/pend cards - Maximum deck size of 39 cards.
The win percentages will be added together, the lowest 2 players in each group will be eliminated.
PM your deck to majofa and Submachine.
GROUP A
mathman101
RootRanger
dawn to dusk
Calindu
Ungaros
GROUP B
Regyptic
rem4life
DANIEELA
Kakerlake
Zso_Zso
GROUP C
andretimpa
farscape
ji412jo
rob77dp
fabian771
mathman101
RootRanger
dawn to dusk
Calindu
Ungaros
GROUP B
Regyptic
rem4life
DANIEELA
Kakerlake
Zso_Zso
GROUP C
andretimpa
farscape
ji412jo
rob77dp
fabian771
Time Remaining:
Round Complete
Round Complete
Post by: andretimpa on January 10, 2014, 01:03:27 am
Post by: dawn to dusk on January 10, 2014, 02:40:28 am
Post by: majofa on January 10, 2014, 02:59:18 am
Like in the other rounds upgrades are allowed in the first deck?Correct.
Post by: RootRanger on January 10, 2014, 03:55:58 am
The first deck has a maximum deck size of 40 cards; in addition, the deck will have 2x mark and 150hp.Just to confirm, the first deck has a maximum of 40 and the second has a maximum of 39? Or did you intend for the first deck to have a minimum of 40?
The second deck: No upgraded cards - Maximum 4 copies of non-pillar/pend cards - Maximum deck size of 39 cards.
Post by: majofa on January 10, 2014, 04:30:48 am
Max for both.The first deck has a maximum deck size of 40 cards; in addition, the deck will have 2x mark and 150hp.Just to confirm, the first deck has a maximum of 40 and the second has a maximum of 39? Or did you intend for the first deck to have a minimum of 40?
The second deck: No upgraded cards - Maximum 4 copies of non-pillar/pend cards - Maximum deck size of 39 cards.
Post by: Kakerlake on January 10, 2014, 11:21:12 am
Post by: farscape on January 10, 2014, 03:11:24 pm
@Kakerlake: separate counter for each would be too easy, I think we need to build a single deck to fight against all, just like in previous rounds.
Post by: Kakerlake on January 10, 2014, 03:49:53 pm
In big bold and red letters. So i't basically against the rules to have the same rules again as last round.
- Each round will have different rules
Post by: rem4life on January 10, 2014, 05:35:08 pm
Max 40 cards restriction for 1st deck, max 39 cards for 2nd deck
It looks like different rules
Post by: andretimpa on January 10, 2014, 05:38:30 pm
I'm asking, because it's written in the rules:In big bold and red letters. So i't basically against the rules to have the same rules again as last round.
- Each round will have different rules
What if this round's rules are different because there is no difference between the rules? ?_?
Post by: majofa on January 10, 2014, 05:46:17 pm
Post by: majofa on January 13, 2014, 02:31:32 am
Post by: dawn to dusk on January 16, 2014, 09:49:01 pm
Post by: Submachine on January 17, 2014, 05:25:59 pm
fabian771 and Calindu haven't sent in any second decks yet, so I give them only some minutes more to send in something or else their percentages for all decks will be 0.
Question for majofa: I counted the players' percentages against their own decks too. Should I delete or keep them?
Post by: farscape on January 17, 2014, 05:48:58 pm
Question for majofa: I counted the players' percentages against their own decks too. Should I delete or keep them?
Well the OP said the decks need to fight the other players' decks only, so I never tested or designed mine to beat my own submission.
So if self-score also counts, that means the rules have changed after submission of decks... >:(
Each player will submit a deck. [No cards that the simulator won't use; No shards]
Then each player will build 1 deck to face the decks of the other 4 players in their group.
The first deck has a maximum deck size of 40 cards; in addition, the deck will have 2x mark and 150hp.
The second deck: No upgraded cards - Maximum 4 copies of non-pillar/pend cards - Maximum deck size of 39 cards.
The win percentages will be added together, the lowest 2 players in each group will be eliminated.
PM your deck to majofa and Submachine.
Post by: Submachine on January 17, 2014, 05:51:23 pm
Edit: ninjad you, rob. ;)
Post by: rob77dp on January 17, 2014, 05:52:38 pm
Each player will submit a deck. [No cards that the simulator won't use; No shards]
Then each player will build 1 deck to face the decks of the other 4 players in their group.
The first deck has a maximum deck size of 40 cards; in addition, the deck will have 2x mark and 150hp.
The second deck: No upgraded cards - Maximum 4 copies of non-pillar/pend cards - Maximum deck size of 39 cards.
The win percentages will be added together, the lowest 2 players in each group will be eliminated.
PM your deck to majofa and Submachine.
-snip-Time Remaining:
Round Complete
The yellow above indicates clearly not facing your own deck.[/s]
Ninja'd by the Submachine [gun]!
Post by: Submachine on January 17, 2014, 06:44:39 pm
| Group A | mathman101 | RootRanger | dawn to dusk | Calindu | Ungaros |
| mathman101 | 97 | 95 | 75 | 87 | |
| RootRanger | 0 | 21 | 10 | 9 | |
| dawn to dusk | 2 | 42 | 10 | 7 | |
| Calindu | 22 | 12 | 11 | 1 | |
| Ungaros | 12 | 19 | 13 | 7 | |
| TOTAL | 36 | 170 | 140 | 102 | 104 |
| Group B | Regyptic | rem4life | DANIEELA | Kakerlake | Zso_Zso |
| Regyptic | 45 | 17 | 21 | 18 | |
| rem4life | 17 | 17 | 20 | 35 | |
| DANIEELA | 29 | 8 | 30 | 8 | |
| Kakerlake | 23 | 51 | 37 | 50 | |
| Zso_Zso | 20 | 0 | 20 | 19 | |
| TOTAL | 89 | 104 | 91 | 90 | 111 |
| Group C | andretimpa | farscape | ji412jo | rob77dp | fabian771 |
| andretimpa | 22 | 6 | 13 | ||
| farscape | 1 | 4 | 8 | ||
| ji412jo | 82 | 97 | 15 | ||
| rob77dp | 21 | 48 | 26 | ||
| fabian771 | 63 | 49 | 22 | 41 | |
| TOTAL | 167 | 216 | 58 | 77 | 0 |
Post by: dawn to dusk on January 17, 2014, 07:32:21 pm
ggs all, you all made this comp hard enough for it to be a very enjoyable challenge. sadly, RL is making me rush my deck and i will get kicked because of it.
-_- disregard this please
Post by: Ungaros on January 17, 2014, 08:11:44 pm
Post by: Kakerlake on January 17, 2014, 09:40:23 pm
Nice work you folks who advanced. Especially farscape, how the bouncy ball of unseriousness did you manage to get 200% :O
Post by: Submachine on January 17, 2014, 09:42:16 pm
count how many of them were monoentropies ;D
Post by: dawn to dusk on January 18, 2014, 01:50:24 am
Post by: MyNameIsJoey on January 18, 2014, 02:15:54 am
Post by: majofa on January 18, 2014, 02:57:11 am
I tested the EXACT deck that farscape used and got much, much lower results, lower even than the deck i posted now. but oh well.Did you remember to do 2x mark and 150 hp?
Post by: MyNameIsJoey on January 18, 2014, 03:23:49 am
I tested the EXACT deck that farscape used and got much, much lower results, lower even than the deck i posted now. but oh well.Did you remember to do 2x mark and 150 hp?
1 did, and if i didnt i wouldve had better results dont you think?
Post by: majofa on January 18, 2014, 03:39:11 am
Ya, true.. but I ended up with the same %s as SubmachineI tested the EXACT deck that farscape used and got much, much lower results, lower even than the deck i posted now. but oh well.Did you remember to do 2x mark and 150 hp?
1 did, and if i didnt i wouldve had better results dont you think?
Post by: RootRanger on January 18, 2014, 03:58:49 am
Even then, the biggest difference was that Mono Entropy was a counter to your Crusader deck, but a very week choice against farscape's rainbow.
Post by: majofa on January 18, 2014, 04:07:49 am
Either you were using a browser other than Chrome, or you weren't using a similar deck. By "similar" I mean within 1-2 cards, because anything more than that can have a moderate effect on the results. But if you had the exact same testing conditions, the probability that you would get much lower results from sheer chance would be...almost zero.
Even then, the biggest difference was that Mono Entropy was a counter to your Crusader deck, but a very week choice against farscape's rainbow.
Oh, I misunderstood what he meant.
Post by: Odii Odsen on January 18, 2014, 10:25:45 am
I tested the EXACT deck that farscape used and got much, much lower results, lower even than the deck i posted now. but oh well.
Its actually very simple. Mono Entropy with 39 cards is a nice counter to your deck (almost 100%) but very weak against a 40 card bow (0% winratio). That mono entropy can't win against a bigger deck, cause its based on deck outs.
Post by: Submachine on January 18, 2014, 11:42:04 am
Post by: Odii Odsen on January 18, 2014, 12:34:14 pm
The percentages during the testing were very unstable, for example: a deck that I tested first showed 20% more winrate than second. It may be because it's all about luck whether a card or a combo gets drawn or not. And there were 4 tests, so all these ~20 differences might have added up. (20 is just a limitnumber, the differences were usually less than 20)
Not sure if I understand what you said, but I built (just for fun) a similar deck to farscape's one. And it had almost the same winratios. Even after more than 3x1000 runs. It varied between ~0,1 - 3 %
Post by: andretimpa on January 18, 2014, 12:53:16 pm
The percentages during the testing were very unstable, for example: a deck that I tested first showed 20% more winrate than second. It may be because it's all about luck whether a card or a combo gets drawn or not. And there were 4 tests, so all these ~20 differences might have added up. (20 is just a limitnumber, the differences were usually less than 20)
I made 5000 games for each matchup while I was testing. It seemed enough to remove any fluctuations, but that would a lot of work (and boring) to do it for all decks in all groups.
I tested the EXACT deck that farscape used and got much, much lower results, lower even than the deck i posted now. but oh well.
Its actually very simple. Mono Entropy with 39 cards is a nice counter to your deck (almost 100%) but very weak against a 40 card bow (0% winratio). That mono entropy can't win against a bigger deck, cause its based on deck outs.
While I was making preliminary tests I tested my deck with 6 antimatters instead of 4 and got 15% against farscape. This made no sense at all so I checked single games and it turns out the simulator was creating singularities from the supernovas, which didn't happened when I packed 4 antimatters. So, yeah, this simulator is kinda crazy.
Post by: Submachine on January 18, 2014, 01:25:17 pm
The percentages during the testing were very unstable, for example: a deck that I tested first showed 20% more winrate than second. It may be because it's all about luck whether a card or a combo gets drawn or not. And there were 4 tests, so all these ~20 differences might have added up. (20 is just a limitnumber, the differences were usually less than 20)
I made 5000 games for each matchup while I was testing. It seemed enough to remove any fluctuations, but that would a lot of work (and boring) to do it for all decks in all groups.
If majofa and the participiants agree, I'm willing to do this 5000 game for each for all groups to be more accurate. From now on, I'll use this method for the rest of the round.
Post by: Calindu on January 18, 2014, 01:26:31 pm
The percentages during the testing were very unstable, for example: a deck that I tested first showed 20% more winrate than second. It may be because it's all about luck whether a card or a combo gets drawn or not. And there were 4 tests, so all these ~20 differences might have added up. (20 is just a limitnumber, the differences were usually less than 20)
I made 5000 games for each matchup while I was testing. It seemed enough to remove any fluctuations, but that would a lot of work (and boring) to do it for all decks in all groups.
If majofa and the participiants agree, I'm willing to do this 5000 game for each for all groups to be more accurate.
I agree with this.
Post by: majofa on January 18, 2014, 07:19:49 pm
Post by: MyNameIsJoey on January 19, 2014, 05:57:32 am
And as root said, i realized there were 2-3 cards difference, hard to believe it made that much of a difference but looks like it did. My bad.
Post by: majofa on January 19, 2014, 08:24:08 am
| Calindu (http://dek.im/d/z56qqz46u3z16u5z1713745z177g7ah7do7gvz17k6z17n87q5z17t980a80g8pj) | dawn to dusk (http://dek.im/d/z474az374fzC7q0z27qcz47qd8pl) | mathman101 (http://dek.im/d/z35c5z57aiz27ajz27anzB7buz27k2z57k6z27n28pq) | RootRanger (http://dek.im/d/zA6qqz56u3z26u7z2713z177fz17h2z27k6z17n3z17q37tdz380a80g8pj) | Ungaros (http://dek.im/d/z26rdz17t4z17t5z57t6z17t9z17ta7tbz37tcz57tdz17um8pt) | TOTAL % | |
| Calindu (http://dek.im/d/zD4saz14vnz258vz25c3z35faz15iaz25liz35lmz36218pq) | 10.54% | 81.00% | 8.80% | 9.72% | 110.06% | |
| dawn to dusk (http://dek.im/d/z94vcz34voz3500z35l9z45rgz25ro8pj) | 7.92% | 96.34% | 22.00% | 10.54% | 136.80% | |
| mathman101 (http://dek.im/d/z45bsz35c5z35c7z35lmz35mqz15ohz35oiz15onz35oo8pr) | 29.12% | 5.36% | 0.02% | 11.32% | 45.82% | |
| RootRanger (http://dek.im/d/zD4sa4vpz152pz158sz15ia5ibz15liz35lm5roz161v8pq) | 19.46% | 48.30% | 95.14% | 22.68% | 185.58% | |
| Ungaros (http://dek.im/d/z35upz34suzC61oz361rz161uz3622z36258pt) | 1.42% | 5.32% | 83.80% | 6.68% | 97.22% |
This is the official simulation. (5000 games using Chrome)
Sorry that it's different than what Submachine posted.
Post by: majofa on January 19, 2014, 08:24:32 am
Post by: RootRanger on January 19, 2014, 06:47:16 pm
Just when I think Calindu had been taken care of...
Post by: rob77dp on January 20, 2014, 04:44:50 am
(Why is it all being re-simulated and why are the results - through Group A only so far - turning so very very different?)
Post by: RootRanger on January 20, 2014, 05:26:46 am
Post by: rob77dp on January 20, 2014, 05:35:17 am
Submachine used only 100 games, while majofa used 5000. Clearly the latter is more accurate.
Sub used only 100 (one hundred) sims? Yeah, that is too few for sure. I use a couple iterations of 1,500 in my testing before submission. (PC is not awesome enough to do 3,000 at a time) Even then I feel it isn't accurate enough for official standings and such.
[Is your "100" accurate or a typo that Sub really used "1,000" sims?]
Post by: majofa on January 20, 2014, 07:55:01 am
| Regyptic (http://dek.im/d/z16u1z46u3z26u5z46vez477gz377jz47t67t98pm) | rem4life (http://dek.im/d/z8710z2711z3712z3713z5718719z171az371bz371dz272i8pk) | DANIEELA (http://dek.im/d/z37t4z37t5z37t6z37t9z27taz17tbz37tcz17tiz37um8pt) | Kakerlake (http://dek.im/d/z57t4z37t5z57t67t8z27t9z17ta7tbz27tcz17toz57um8pt) | Zso_Zso (http://dek.im/d/z57dkz27dlz27dnz27e4zE7f2z17juz17k2z27k6z27kc8pq) | TOTAL % | |
| Regyptic (http://dek.im/d/z64vcz34vez34vfz34vlz34vnz650u8pj) | 18.04% | 23.16% | 23.04% | 17.66% | 81.90% | |
| rem4life (http://dek.im/d/z94saz34vj4vpz152l55v58sz15c2z25f6z15i8z15liz35lmz15onz15upz261q8pq) | 43.76% | 3.40% | 36.80% | 0.24% | 84.20% | |
| DANIEELA (http://dek.im/d/z74vcz34vez34vfz24vlz34vnz750u8pj) | 13.86% | 19.22% | 29.22% | 17.74% | 80.04% | |
| Kakerlake (http://dek.im/d/z34sjz54vcz34vez34vfz24vlz34vnz450u8pj) | 17.10% | 18.70% | 26.64% | 18.86% | 81.30% | |
| Zso_Zso (http://dek.im/d/z44vcz34vdz34vnz2500z450uz45l8z35lmz25lsz45mq8pj) | 25.98% | 34.06% | 7.60% | 46.56% | 114.20% |
Post by: Calindu on January 20, 2014, 07:09:23 pm
:O What a plot twist!
Just when I think Calindu had been taken care of...
Not going down that easy :p
Post by: dawn to dusk on January 20, 2014, 08:54:12 pm
Post by: majofa on January 20, 2014, 09:14:43 pm
So are we abiding by majo's test, or sub's tests?The latter ones since they were 5000 tests instead of 100.
Post by: andretimpa on January 20, 2014, 09:56:16 pm
So are we abiding by majo's test, or sub's tests?The latter ones since they were 5000 tests instead of 100.
Which (assuming things are gaussian) have 1/7 only of the fluctuations in Sub's
Post by: Regyptic on January 21, 2014, 10:40:02 am
Interesting that Kakerlake and I had almost exactly the same second deck but apparently the differences in his (-1Discord +1 quanta) were enough to get him an extra 5.34% in the 3 common matchups. Goes to show that every card counts. (Similar with DANIEELA's deck, only 2 cards different I think).
Post by: DANIEELA on January 21, 2014, 03:16:38 pm
Since we need to build deck #1,you should consider to incorporate winning rate of that deck together with deck #2.
That should be more fair to all.
(not for #1 round ,where we have got prebuild decks from organizer)
Post by: farscape on January 21, 2014, 03:23:24 pm
Suggestion for Computer Simulations II
Since we need to build deck #1,you should consider to incorporate winning rate of that deck together with deck #2.
That should be more fair to all.
(not for #1 round ,where we have got prebuild decks from organizer)
In an implicit way the #1 deck does contribute significantly, if you look at the tables, those who won also had #1 decks that were very hard to beat. Such deck gives you a better chance of winning by lowering the win-rates of your opponents.
Although, I see you would have overtaken Regyptic this round if the #1 win-rate was added, so it does make a difference.
Post by: rem4life on January 21, 2014, 11:02:28 pm
Post by: andretimpa on January 21, 2014, 11:12:59 pm
Any deadline for organizers to end the round and start another one?
Group C is still not published, so who knows?
Post by: majofa on January 21, 2014, 11:43:54 pm
Post by: majofa on January 22, 2014, 08:00:55 am
| andretimpa (http://dek.im/d/z66qqz56u3z36u5713z474f77g77j7ah7dm7dq7gm7jr7k27n67q57t980b8pj) | fabian771 (http://dek.im/d/z87gkz57gmz17grz47gtz17gu7h8z87i68pp) | farscape (http://dek.im/d/5ifz86qqz16ttz46u3z36u5z171371bz274fz177g77j7ah7dm7dq7juz27n87t9z180g8pj) | ji412jo (http://dek.im/d/zA7joz37jvz37k5z37k6z37n2z37t9z37tb8pt) | rob77dp (http://dek.im/d/z47t47t5z57t67t8z37t97ta7tbz27tcz37tf7thz27um8pt) | TOTAL % | |
| andretimpa (http://dek.im/d/z34sjz44vcz34vlz34vnz3500z650uz361tz361q8pu) | 63.16% | 0.76% | 83.64% | 24.38% | 171.94% | |
| fabian771 (http://) | 0.00% | 0.00% | 0.00% | 0.00% | 0.00% | |
| farscape (http://dek.im/d/zB4vcz34vdz24vgz34vnz3500zB50u8pj) | 21.74% | 42.48% | 94.96% | 51.50% | 210.68% | |
| ji412jo (http://dek.im/d/z94vcz34vfz34vgz34vlz34vnz3500z850u8pj) | 7.02% | 26.12% | 4.68% | 25.34% | 63.16% | |
| rob77dp (http://dek.im/d/z661oz361qz361rz361tz161uz2625z563a8pu) | 9.26% | 37.08% | 13.14% | 26.20% | 85.68% |
Post by: farscape on January 22, 2014, 03:59:51 pm
Post by: Submachine on January 22, 2014, 06:39:59 pm
Post by: farscape on January 22, 2014, 06:53:03 pm
So basically I didn't fail much with only 1000 simulations before we did the 5000-sim-tests.
Not as far as my position is concerned, but DANIEELA and Calindu may view things differently :P