[Deck] Belcher

Quote:

---

Originally Posted by Vacrix

Nice results cb4!
Do you have a standard of comparison though? It looks really slow to me because I play SI. Does anyone have results w/out Grim Monolith to compare? Also, its a little unclear your results. Whats the exact number of times you won on turn 1, 2 ,etc. I can't tell from looking at your info because some say (6 gobs, 8 belch) so I'd imagine thats 6 EtW but when do they go all the way T2 or 3, and does the Belcher connect or misfire?

Also, how does Welder look post-board? It allows you to make use of your Monolith even after it got used for T: Add 3.

---

I can't see any situation I would want welder to pop up on a draw. You have to get him out, keep him alive a turn then have a the right artifacts in play. Whatever the situation he is just slowing things down. If you looking at boarding him in vs counter spell deck then are you not bringing in Xantid Swarm?

I all for make this deck more resilient but I don't see how welder can do that. Your better off playing with Guttural Response, Blasts or going back to R/G/B and using duress.

You can take advantage of setting up andwinning through multiple countermagic. Gutteral Response is marginal and only answers so many countermagic: one. Also, Welder allows you to rely less on mana during the combo turn because you can get the 'slow play inevitability', or bait a counter. If they don't counter it, you can slowly play manasources and play around each counterspell after the other. Rarely do I see the Welder get countered, especially if you just RFG SSG and then play your Welder, your opponent doesn't know if you can then easily play out a Belcher with the remaining 5 cards and just sit on basically the same strategy, or go into EtW, missing an opportunity to counter the key ritual. Also, a 7 card hand with Welder is much stronger than a winning hand because you don't need to mulligan for a fast win or EtW. It guarentees you multiple tries if they opponent doesn't have removal.

First you play Belcher, and it likely eats a counter. Then just bring it into play via Welder and force mana sources until you get one off and Belch. Mana sources are much more plentiful in the deck. Think of it like running 1-4 more Xantid Swarms, except it protects you getting Belcher into play instead of protecting EtW chains. I'm not saying replace the Swarms with Welders. Rather, cut down on the Wish board and you make room to help against your worst matchup: U.dec.

In my testing, belcher only failed to win once when activated. I noted it in the "on the play" data. The question of when goblins can go the distance is a good one. Playing 10+ on turn one should usually win. Turn 2+ it depends if your opponent can get an answer (echoing truth, EE, etc.) or if the can put down enough blockers. I have more detailed data from the testing on paper. How many goblins would you need to make on turn 2 or 3 to consider it lethal? It will always be theorhetical because someone can always answer a million goblins for 2 mana. Maybe making goblins after turn 2 is not a win. I would be open to opinions on that.

I will do the same study today with wraiths. I will still only keep a 6 or 7 card hand with a win condition because I would never want to need to draw one. With 11 win conditions in the deck (3 ETW, 4 belch, 4 wish) and just over 50 cards in the deck, the probablity of drawing one is about 20%. Using that estimate your odds of drawing one or more win conditions goes roughly like this...

In 1 draw 20%
In 2 draws 36%
In 3 draws 49%
In 4 draws 59%
In 5 draws 68%

That means you have about a one in three chance of not drawing any win condions in five draws. Your chance of drawing a win condition isn't even around 50/50 unless you have 3 draws to do it.

You make good points about welder. I don't know what I would cut from the board though. It would proably have to be some blasts and I don't know if I want to part with them.

I will post my testing results with wraiths later today.

Just finished testing with wraith. I ran the same list I posted last page with wraiths in for monoliths.


ON THE PLAY (25 games)

T1- 10 games (7 gobs, 3 belch)
T2- 7 games (5 gobs, 2 belch)
T3- 6 games (2 gobs, 4 belch) (one of these belches only went for 18 dam on turn 2 but was able to belch for lethal again T3)
T4- 0 games
T5- 1 game (belch)
1 game the deck failed go off in 5 turns

Mulled to 6 twice
Mulled to 5 twice


ON THE DRAW (25 games)

T1- 17 games (9 gobs, 8 belch)
T2- 2 games (2 belch)
T3- 5 games (1 gobs, 4 belch) (one of these belches was a turn 1 for 6 damage and rebelched turn 3 for lethal)
T4- 1 game (belch) (this was a turn 3 belch that failed and rebelched turn 4 for the win)
T5- 0 games

Mulled to 6 twice


OVERALL (50 games)

T1- 54%
T2- 18%
T3- 22%
T4- 2%
T5- 2%
Fail- 2%

Mull to 6 - 8%
Mull to 5- 4%

According to this study the deck is faster and mulls less with wraith over monolith. I do have to say that the set of 25 games on the play with monolith got alot worse draws than any other set of 25 games and this is a small sample size. The set of 25 games I did with wraith on the draw only mulled twice (both to 6 cards) in 25 games. This seems to be a very good set of draws. I will definitely have to test more with both. I was ready to dismiss wraith, but I may have to reconsider that and test more.

I was about to begin this testing, but it seems you have been doing good work. Could you also test with Monolith in the Manamorphose slot, just for completeness? I completely understand this isn't something that would normally be considered, but I was just curious!

Why would you want to test in the Manamorphose slot? What would you play instead? Street Wraith? Wraith is just a bad version of Manamorphose. It doesn't count as a spell. What would you play instead?

cb4.. would you be down to test a spoils list? I'm curious.. EtW for much on the 1st turn doesn't really impress me to tell the truth. It might get there against a few decks, but people pull answers out of their ass for it. TES can do the exact same thing, maybe a turn slower on avg. but with protection and a backup plan. Clearly, Monolith improves classic Belcher but I wonder if Monolith improves the viability of Spoils Belcher as well.

lol, I did clearly state that I did know that that request was out of the ordinary, I want to know for the sake of completeness of testing, as sometimes the expected result is not the correct result. Is that a bad thing now?

Its not. I was just asking what you would run instead. Actually, I accidentally said it twice. You would have 4 open slot if you moved Grim Monolith to the Manamorphose slot.

Vacrix, I would be down to test it. I ruptured my achilles and am laid up all summer, so I have time. Post a list for me.

I think all the tests of how fast the deck will go off are redundant What kills Belcher is the long hall not if it goes off on turn 1 or turn 2. Fishbowl testing has been done it needs to be play tested against real decks with disruption. Everyone though 2 land belcher was old tech and it top 8s in the St Louis SSG event. Sideboard has 3 duress and 4 swarm.

Build need to start looking at endurance not speed. This game is against another player not a goldfish.

Quote:

---

Originally Posted by jtwilkins

I think all the tests of how fast the deck will go off are redundant What kills Belcher is the long hall not if it goes off on turn 1 or turn 2. Fishbowl testing has been done it needs to be play tested against real decks with disruption. Everyone though 2 land belcher was old tech and it top 8s in the St Louis SSG event. Sideboard has 3 duress and 4 swarm.

Build need to start looking at endurance not speed. This game is against another player not a goldfish.

---

Well it's only against other players if they are playing blue spells usually. If not, then it's usually goldfishing.
Just kidding.

But I see where your coming from. 4 Xantids are all I run in the board, and I really need more. I run the R/G version, because I don't like black in the deck. I am thinking of running ReB's in the board as well.

The deck is great fun to play.

I have recently dropped 4 Blood Moon from the board, as they have been a little hit and miss lately, but have changed up to 2 Magus of the Moon, as people usually board out their removal and 2 Pyroblast. This seems to have improved my blue matchup slightly, as I can now protect the first turn Swarm.

Funny thing about that 2 land Belcher from the SCG St Louis 5K event was that he was only running the Bayou for SB cards, he could have quite easily run the 1 land version with the 4th Manamorphose and had the Bayou in the SB!

Hey guys,

maybe you are interested in this:
We did some statistic on university that can be applied to MtG.
I calculated the chances of winning with Belcher if activated.
I sumulated 50 activations. X= frequency

I assumed that there are 50 cards left in deck (one of wich is Taiga) and Belcher is activated. Here are the results:

1-Land-Belcher with Taiga revealed

damage
0 XX
2 X
4 XX
6
7
8
9
10 X
12 XX
14 XX
16 X
18 X
>18 XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX (38x)

If your opponent is on 20 you have winning chance of 76%

Here the results with both of them in deck still 50 cards left:

2-Land-Belcher
damage
0 XXX
1
2 XX
3 X
4 X
5 XX
6 XX
7
8 XXXXX
9
10 XXXXX
11 X
12 XX
13
14
15
16
17
18 X
19 X
>19 XXXXXXXXXXXXXXXXXXXXXXXX (24)

If your opponent is on 20 you have winning chance of 48%
Which is half of the 1-Land-Belcher condition although you have 2 Lands in deck!


Maybe these data can help you.

GL

That is actually pretty interesting, did you have a program for that or is it all done by hand? If its a computer, any chance we can have a bigger sample? Say 500 on each?

Quote:

---

Originally Posted by yankeedave

That is actually pretty interesting, did you have a program for that or is it all done by hand? If its a computer, any chance we can have a bigger sample? Say 500 on each?

---

I did it by hand but I am sure there ARE programs that can do this. Actually I think I can figure out how it works with excel, I think there is a function.
Here is the way I did it, maybe someone has an idea or knows programs that can help:

I wanted to know how often Belcher is lethal if activated (and there are still land(s) in deck).
So I assume we drew 7 + 3 more cards until we can go off, leaves 50 cards in deck, including 1-2 lands.
Therefore the chance on revealing - let's say - Taiga is 1:50.
For Belcher to be deadly with Taiga in deck we have to reveal 10 cards BEFORE we reveal Taiga.

So here is how it works:
Nearly every calculator we use for university has a "random" function. There you get a random number between .00 and .99
Each number is even likeliy to occur because it's random.
I "coded" Taiga as .01 and .51. All other numbers between .00 and .99 (which are 98) are coded for any other card that is not Taiga.
Now I use the random-function as often as I wish until either:
.01 or .51 occurs
or
we have 10 numbers in a row that aren't .01 or .51
(depending on what comes first)

Example:
.28, .61, .95, .15, .78, .01 (We have 5 numbers that aren't Taiga and the last one is .01). So we got 5*2 damage.

I repeated it 50 times and always recording the damage.


For the 2-Land-condition I did the same thing:
Taiga = .01 and .51
Bayou = .02 and .52

Only that we have use the random function until
.01, .51, .02 or .52 occur
or
we have 20 number is a row that aren't .01, .51, .02 or .52 (since Bayxou needs 20 cards revealed to kill)


What we need is a program that does the simulation (the random-function) for us AND that counts the frequency of the numbers .01, .51, .02 and .52.
Maybe someone can help out with that? So we can easily simulate 500 or even 5000 samples.

By the way:
In statistics we call results "significant" if we have more than 40 samples. I wouldn't say that more testing brings up higher or lower kill-rates you only get more counts for the other 1-19 damages.

Wow you did it the hard way.

Let's start at the beginning. The chance of a specific sequence of events is equal to the chance of the multiples of it's component parts, assuming that each event is independent.

An example: the chance of flipping heads on a coin three times in a row = P(3) = 1/2*1/2*1/2 = .125 = 12.5%

So assuming 20 life, we know we need to flip 10 cards. Chance [with one land in the deck] = P(10) = 49/50*48/49*47/48*.....41/42*40/41 = .8 = 80% This is pretty close to your experimental value.

Now, with two lands in the deck P(10) = 48/50*47/49*46/48....40/42*39/41 = .637 = 63.7% This is significantly more than your statistical value, which suggests that you had a run of bad luck. Luck can easily skew statistics when sample sizes are quite small.

-Silent Requiem

Edit: Another problem with your model is that you are drawing from the same 50 cards each time, when in reality you are drawing from a smaller pool with each progressive card. Under your model, for example, it is possible (but improbable) to do 1,000 damage (or any other arbitrarily large number), whereas the above model correctly predicts a point at which you have a 100% chance of the next card being a land (ie, when it is the only card left).

Further Edit: Actually, you probably only need the first nine cards, because you really don't care what the 10th card is... it's guarenteed to be lethal. You can adjust the above math by ommiting the final fraction (although technically it simply becomes a "1").

Bah, kids these days and their crude statistical tools. Here's a more elegant method from a more civilised time:


Same assumption, 50 cards in the deck, one of them is a Taiga, opponent is at 20.

The chance that you don't flip the Taiga in the first reveal is 49/50
The chance that you don't flip the Taiga in the second reveal is 48/49
...
The chance that you don't flip the Taiga in the 10th reveal is 40/41

*

The product of these chances is our chance to kill, and this simplifies to 40/50 = 0.8 because 41 to 49 show up in both numerator and denominator.

*

For 2 Taigas, individual chances go from 48/50 to 39/41, so our chance to kill is 39*40/(49*50) = 0.637




EDIT: Ok, sorry for being very late. Silent Requiem, your numbers are correct. You don't trim the last fraction because Belcher deals damage equal to the revealed *nonland* cards. If the 10 revealed card is a Taiga, you only deal 18.

Silent Requiem's math is correct. Although, with one land sitting in 50 cards, you can find the probability of going off for 20 even easier. If the land is one of the top 10 cards, you lose. If it is one of the bottom 40, you win. 40/50 = .8. Same result. You can find the probability of going off for 20+ with 1 land belcher at any time by doing (# of cards left in deck -10) / number of cards left in deck. With 53 cards left you would have a 43/53 = .811 = 81.1% of 20+ damage. This way of thinking about it can be done during the game.

You could expand this into a formula for one land belcher...

D= damage you need to do
N= number of cards left in deck

FORMULA: (N- (D/2 rounded up)) / N

FORMULA IN WORDS: Take the number of cards left in your deck and substract half the damage you need to do (rounded up). Take this number and divide by the number of cards left in deck.

Example: I have 47 cards left in my deck and need to do 17 damage to win.

(47- (17/2)) / 47
47 - 9 / 47
38/47 = .808 = 80.8%

O.k. the way you did it is much easier, but I didnt know those formulas.
Nevertheless they dont bring resukts for the 2-land-one-of-which-is-bayou-condition.
Silent Requiem said I had bad luck but you ony reffered to a situation where 2 TAIGA are in deck and not 1 Taiga, 1 Bayou. This situation is different cause Bayou needs 20 cards to kill, and Taiga 10.
So I think this part of my calculation is right.

Land Grant invalidate those stats too, as there will be ain't no land anymore in the deck at the time you activate Belcher if you play the 1-land belcher, and only one if you play the regular 2-lands version.

I say that because some people could find that the percentage of win is pretty low (under 75%) when you activate Belcher. But it increases if you drew your land or cast Land Grant.

P-M