Draft Development in ML 3.0
Discuss Upcoming Releases, Coding New Cards, Etc.
PLEASE DO NOT REPORT BUGS HERE!
PLEASE DO NOT REPORT BUGS HERE!
Moderators: BAgate, drool66, Aswan jaguar, gmzombie, stassy, CCGHQ Admins
Re: Draft Development in ML 3.0
by foolosopher » 08 Jan 2013, 13:39
It's included in the rar that I uploaded. I haven't changed something in any other file.
What else do you need?
What else do you need?
I long for the time where one man could make a difference...
- foolosopher
- Posts: 189
- Joined: 19 May 2010, 20:53
- Location: Greece
- Has thanked: 13 times
- Been thanked: 17 times
Re: Draft Development in ML 3.0
by Aswan jaguar » 13 Jan 2013, 19:27
I discovered why AI puts so many lands on its decks.A value on Manalink.csv -AI base value- is essential for AI to select the cards for it's decks and as the value for many new cards was left to 0 AI thought of them as unworthy to be put to the decks and so preferred to fill the decks with lands which have a value.
---
Trying to squash some bugs and playtesting.
Trying to squash some bugs and playtesting.
-
Aswan jaguar - Super Tester Elite
- Posts: 8131
- Joined: 13 May 2010, 12:17
- Has thanked: 748 times
- Been thanked: 479 times
Re: Draft Development in ML 3.0
by foolosopher » 18 Jan 2013, 22:45
Hi AJ,
You are on the spot, also there is an upper limit on the number of cards that are selected for a drafted deck, but not a lower limit. Then the rest of the cards up to 40 are filled with with lands and that's how you can get decks wit 20 or more lands. I 'll see what I can do. Can you tell me in which expansions this tends to happen?
You are on the spot, also there is an upper limit on the number of cards that are selected for a drafted deck, but not a lower limit. Then the rest of the cards up to 40 are filled with with lands and that's how you can get decks wit 20 or more lands. I 'll see what I can do. Can you tell me in which expansions this tends to happen?
I long for the time where one man could make a difference...
- foolosopher
- Posts: 189
- Joined: 19 May 2010, 20:53
- Location: Greece
- Has thanked: 13 times
- Been thanked: 17 times
Re: Draft Development in ML 3.0
by Gargaroz » 19 Jan 2013, 14:43
Whops, Foolosopher, I din't see the C file ! Everything is OK 

----
- Current / medium term task: adjusting the code for making Misdirection and such usable
- Long term task: inserting all the good stuff I left out from the "Golden Years" mod
- Current / medium term task: adjusting the code for making Misdirection and such usable
- Long term task: inserting all the good stuff I left out from the "Golden Years" mod
- Gargaroz
- Programmer
- Posts: 7097
- Joined: 06 Nov 2009, 11:11
- Has thanked: 82 times
- Been thanked: 595 times
Re: Draft Development in ML 3.0
by Aswan jaguar » 19 Jan 2013, 16:19
Foolosopher some sets don't have the value at all and there others sets which have some cards miss this value.Stassy already picked up the task and valued the cards for the last 3 coded sets(Ravnicas old,new,Time Spiral) I don't know if he will continue this work for all previous sets.Lets hope Stassy will finish it.
The evaluation is a big trouble if you really want to help AI evaluate the cards better since you not only have to:
1- consider the power of the card,but
2- how valuable and playable-understandable is by AI and
3- keep in mind that this evaluation concerns mostly Draft so cards that are powerless if their fellow combo cards miss are just junk.(that's why original developers had a 0 value there for some cards so AI wouldn't pick them at all also cards that are only good against certain decks or only against or for one colour like Karma,Tsunami,Blood Moon and I just saw that also basic lands have 0 value so AI correctly wouldn't draft them).About 80 cards out of the (~650 if I remember correctly)original game have a 0 value.

The evaluation is a big trouble if you really want to help AI evaluate the cards better since you not only have to:
1- consider the power of the card,but
2- how valuable and playable-understandable is by AI and
3- keep in mind that this evaluation concerns mostly Draft so cards that are powerless if their fellow combo cards miss are just junk.(that's why original developers had a 0 value there for some cards so AI wouldn't pick them at all also cards that are only good against certain decks or only against or for one colour like Karma,Tsunami,Blood Moon and I just saw that also basic lands have 0 value so AI correctly wouldn't draft them).About 80 cards out of the (~650 if I remember correctly)original game have a 0 value.
---
Trying to squash some bugs and playtesting.
Trying to squash some bugs and playtesting.
-
Aswan jaguar - Super Tester Elite
- Posts: 8131
- Joined: 13 May 2010, 12:17
- Has thanked: 748 times
- Been thanked: 479 times
Re: Draft Development in ML 3.0
by foolosopher » 19 Jan 2013, 22:31
AJ, like you say evaluation is a big trouble. Particularly in draft the logic is a bit different when building a deck, where one would go for some standard things like creatures, damage, removal, counters. Combos are mostly not considered and in order to achieve something like that, we would need to use something like a reference table.
Anyway I just saw that the power of some cards has been filled in, so maybe we check, how strong are the decks created for those sets.
I might try to add a few simple rules apart from the existing logic, like choosing all available creatures or card types that I mention above of chosen colours, in case the initial chosen cards are below 20.
We might also get some more feedback on what is lacking or wanted now that many expansions have been added in Draft.
Anyway I just saw that the power of some cards has been filled in, so maybe we check, how strong are the decks created for those sets.
I might try to add a few simple rules apart from the existing logic, like choosing all available creatures or card types that I mention above of chosen colours, in case the initial chosen cards are below 20.
We might also get some more feedback on what is lacking or wanted now that many expansions have been added in Draft.
I long for the time where one man could make a difference...
- foolosopher
- Posts: 189
- Joined: 19 May 2010, 20:53
- Location: Greece
- Has thanked: 13 times
- Been thanked: 17 times
Re: Draft Development in ML 3.0
by Aswan jaguar » 20 Jan 2013, 06:36
Now you have inserted all these sets I would certainly love to see again the feature for less players (6) to have more sets for singleton if it is easy to do of course and if it makes a difference.If the difference are only couple of sets more for 6 players than 8 players it doesn't worth the trouble.
What do you mean?make a table of available card combos and synergies so AI would try to pick those cards together (e.g if you draft card 3 try to get also card 17 and 56)?Have you got something like that in mind?Foolosopher wrote:
we would need to use something like a reference table
---
Trying to squash some bugs and playtesting.
Trying to squash some bugs and playtesting.
-
Aswan jaguar - Super Tester Elite
- Posts: 8131
- Joined: 13 May 2010, 12:17
- Has thanked: 748 times
- Been thanked: 479 times
Re: Draft Development in ML 3.0
by stassy » 20 Jan 2013, 14:33
I am currently working on the M13 set but I am trying a new sorting technique from Sonic and it might take more time to understand it (my method was more brute force like and thus generated more errors)
- stassy
- Moderator
- Posts: 5274
- Joined: 25 Feb 2009, 07:06
- Has thanked: 471 times
- Been thanked: 337 times
Re: Draft Development in ML 3.0
by foolosopher » 22 Jan 2013, 22:29
@AJ: Checked the latest csv and (if counted correctly) the difference between sets with cards coded < 120 and < 90 is
28 and 23, which means the 6 players option would allow 5 more sets for singleton, so unless a lot (say more than 10
) people ask for this, we 'll probably keep this option in reserve.
As for the reference table, you got it right, but it would put some more time (maybe significant) in the card selection process and we want to avoid that. (someone other than me would have to make it also
).
So we better wait and see Sonic and Stassy's sorting techniques and if the drafting is improved, but Stassy could you send or post a link of them to me to take a look? Thx.
28 and 23, which means the 6 players option would allow 5 more sets for singleton, so unless a lot (say more than 10

As for the reference table, you got it right, but it would put some more time (maybe significant) in the card selection process and we want to avoid that. (someone other than me would have to make it also

So we better wait and see Sonic and Stassy's sorting techniques and if the drafting is improved, but Stassy could you send or post a link of them to me to take a look? Thx.
I long for the time where one man could make a difference...
- foolosopher
- Posts: 189
- Joined: 19 May 2010, 20:53
- Location: Greece
- Has thanked: 13 times
- Been thanked: 17 times
Re: Draft Development in ML 3.0
by stassy » 23 Jan 2013, 03:59
I just sent to Sonic the M13 part, and it's just a separate list of cardname and their ranking so you might just ask him for the fully complete integrated up to date Manalink.xls
Also I am currently going in reverse starting from the latest coded expansion, however if you have a preferred expansion for draft that need to be done before you can say so.
What is done so far:
Also I am currently going in reverse starting from the latest coded expansion, however if you have a preferred expansion for draft that need to be done before you can say so.
What is done so far:
Next expansion in line:Return to Ravnica
Duel Decks: Izzet vs Golgari
From the Vaults: Realms
Magic 2013
Ravnica: City of Guilds
Innistrad block
Innistrad
Dark Ascension
Avacyn Restored
- stassy
- Moderator
- Posts: 5274
- Joined: 25 Feb 2009, 07:06
- Has thanked: 471 times
- Been thanked: 337 times
Re: Draft Development in ML 3.0
by foolosopher » 24 Jan 2013, 23:23
Thanks for the list Stassy, but what I wanted is the rules that you use to put a certain ranking to each card, unless I misunderstood what you meant...
I long for the time where one man could make a difference...
- foolosopher
- Posts: 189
- Joined: 19 May 2010, 20:53
- Location: Greece
- Has thanked: 13 times
- Been thanked: 17 times
Re: Draft Development in ML 3.0
by stassy » 25 Jan 2013, 03:58
It's in the AI base value thread in the Developmment section.
I just followed AJ advices on ranking, but bear in mind that someone already ranked some of the cards and used another ranking system (for example some cards have 200 while I limit myself to 100 for card considered as bomb card.)
I will not modify these and only replace non original cards 0 ranking with x ranking.
I just followed AJ advices on ranking, but bear in mind that someone already ranked some of the cards and used another ranking system (for example some cards have 200 while I limit myself to 100 for card considered as bomb card.)
I will not modify these and only replace non original cards 0 ranking with x ranking.
- stassy
- Moderator
- Posts: 5274
- Joined: 25 Feb 2009, 07:06
- Has thanked: 471 times
- Been thanked: 337 times
Re: Draft Development in ML 3.0
by Aswan jaguar » 25 Jan 2013, 05:32
Stassy I just want to say that even Microprose had originally cards with values with 100,200 like Ancestral Recall.
Last edited by Aswan jaguar on 08 Feb 2013, 17:53, edited 1 time in total.
---
Trying to squash some bugs and playtesting.
Trying to squash some bugs and playtesting.
-
Aswan jaguar - Super Tester Elite
- Posts: 8131
- Joined: 13 May 2010, 12:17
- Has thanked: 748 times
- Been thanked: 479 times
Re: Draft Development in ML 3.0
by stassy » 25 Jan 2013, 15:42
Yes I know, but the one who did the ranking before me put some 200+ on recently coded card...which is different from the ranking you suggested.
- stassy
- Moderator
- Posts: 5274
- Joined: 25 Feb 2009, 07:06
- Has thanked: 471 times
- Been thanked: 337 times
Re: Draft Development in ML 3.0
by Aswan jaguar » 26 Jan 2013, 12:14
Foolosopher can you do something for the mana ratio AI builds decks.If I am not wrong you have set AI to put 17 lands for each deck it builds (when cards have AI base value that is) and put 9 lands for the colour with most cards and 8 for the one with fewer cards (for 2 colours deck)and in most cases that is good as AI makes balanced 2 coloured decks but in a case I just come up against for 2 green cards 8 forests for 21 white 9 plains(and only 1 green needed
in it's casting cost) the chances of getting mana - screwed by the 8 forests is big.By far I am not a good MTG player but 5 forests in this case would be enough IMHO.


---
Trying to squash some bugs and playtesting.
Trying to squash some bugs and playtesting.
-
Aswan jaguar - Super Tester Elite
- Posts: 8131
- Joined: 13 May 2010, 12:17
- Has thanked: 748 times
- Been thanked: 479 times
Who is online
Users browsing this forum: No registered users and 9 guests