E1 is the worst one drop if we're looking to cut? HH necessary? Do you want to see 2+ AC? 18 land is the absolute fewest lands?
Yes to all. E1 is sometimes awkward in multiples and a terrible topdeck, though shaving too many can't be right because green one drops are at a premium with HH. HH is the reason this deck is even worth considering again because of how often you get to put 3 or 4 or 5 creatures into play on turn 2. Atarka's Command is the best card in the deck and you want to see it every game. Less than 18 lands with no cantrips makes it really hard to hit 2 lands on time. With 18 lands, you already have a 31.4% chance of having zero or one lands in your opener, and those 1 landers are only a little better than 50% to find a land in two drawsteps.
For completeness, the attached table is the probability of having EXACTLY X lands in your opening 7 with 18 lands in the deck.
The only placing list with AER has 0 HH; that's why I ask.
What % of 1-landers do you typically keep? 2? 3?
I keep a lot more than 2-3% of landers, but mostly those that are fairly strong if they hit the 2nd land. For example: Fetch, Ape, Nacatl, X1, HH, BTE is an easy keep no matter what the last card is, especially on the draw. I don't really know what the % is, but somewhere between 20-50% doesn't sound outrageous. 1 land hands that don't have a BTE or HH I'm suspcious of, though I don't want to outright say I wouldn't keep any of them. Still learning the deck.
I mean, what % of 1-landers do you keep? What % of 2-landers? What % of 3-landers?
This isn't something I've recorded in my testing, but it's something that I could go back and do this evening. However, it would take a lot of time and not be very productive I think. The probabilities for keepable hands can be worked out analytically anyway.
I would keep any 2 or 3 lander with enough creatures (too few creatures is very unlikely when you play 32 of them). I'd mulligan pretty much all 4 landers. I'd keep 1 landers with multiple 1 drop creatures, since it allows me to build up a board presence on 1 land.
Suppose you want to maximize the probability of keeping your opening hand with the following assumptions: 0- and 4-or-more-land hands are unkeepable, 1-landers are keepable 50% of the time, 2-landers are keepable 100% of the time, and 3-landers are keepable 75% of the time. Then, you want 18 land, yup! Can you estimate these probabilities from your memory of your games?
Suppose you want to maximize the probability of keeping your opening hand with the following assumptions: 0- and 4-or-more-land hands are unkeepable, 1-landers are keepable 50% of the time, 2-landers are keepable 100% of the time, and 3-landers are keepable 75% of the time. Then, you want 18 land, yup! Can you estimate these probabilities from your memory of your games?
No, I can't. I would literally have to watch cockatrice replays of 200+ games to do it and I don't think that would be productive because it can be done analytically.
This is just math with the hypergeometric distribution and it could be worked out analytically. You don't need to arbitrarily say "1-landers are keepable 50% of the time" because you can write down the criteria for a keepable 1 lander and then calculate the probability of drawing a keepable 7 with 1 land.
Of course an analytical solution exists for a given set of criteria, but the devil is in the details. What are the criteria for you to keep a 1/2/3 lander?
Sometimes analytical solutions are less convenient than experimental ones.
Of course an analytical solution exists for a given set of criteria, but the devil is in the details. What are the criteria for you to keep a 1/2/3 lander?
Sometimes analytical solutions are less convenient than experimental ones.
This isn't something I've recorded in my testing, but it's something that I could go back and do this evening. However, it would take a lot of time and not be very productive I think. The probabilities for keepable hands can be worked out analytically anyway.
I would keep any 2 or 3 lander with enough creatures (too few creatures is very unlikely when you play 32 of them). I'd mulligan pretty much all 4 landers. I'd keep 1 landers with multiple 1 drop creatures, since it allows me to build up a board presence on 1 land.
An analytical solution is not less convenient here. This isn't some crazy theory that has to be tested. There is an exact solution to the question you're asking. Any experimentally determined answer will have error bars, but the analytical solution is exact.
This literally amounts to writing down the probability that a given 7 card hand has 2 or more 1-drops and then calculating that probability and that is equal to the probability that a 1 land hand is keepable.
Any 2-3 lander is keepable provided that it's not just non-creature spells. This is also a probability that you can write down analytically.
1 land 6 ones is keepable? 4 lands, 2 Command, Experiment One is keepable? There are tons of non-keepable hands that meet your requirements. That's why analytical solutions are hard. Because, yes, comuting what you said is easy, but it's not actually realistic until we have tons more requirements. Hmmm. Just trying to optimistic the deck.
1 land 6 ones is keepable? 4 lands, 2 Command, Experiment One is keepable? There are tons of non-keepable hands that meet your requirements. That's why analytical solutions are hard. Because, yes, comuting what you said is easy, but it's not actually realistic until we have tons more requirements. Hmmm. Just trying to optimistic the deck.
What's your current list? Anything spicy?
You have to come up with a set of requirements either way. The best you're actually going to do is an approximation and it won't be sufficiently different from the hypergeometric function for the number of lands in your opener that spooly already posted. 2s and 3s are pretty much always fine. 4s aren't. 1s are if they have enough 1 drops and that can be calculated. Frankly, the exercise of thinking about what's keepable and what isn't is a more valuable exercise than writing down a probability.
I don't keep 4s. I definitely wouldn't keep one with a 1/1 and 2 commands because that hand is complete garbage.
6 one drops and one land? Sure, because I'll be putting power on the board for ages with it, which buys time to draw lands.
All of this land talk and reflecting on some hands that I've had to keep that have gone bad has made me want to add a land. Even burn is playing 20 lands these days. I think the shell is powerful enough that we just want to make sure we can consistently hit 2 lands by turn 2. I'm probably shaving a Kird Ape for another fetch, and might shave something else for another fetchable.
Caleb Durward posted a new iteration of Landfall Rallier on CFB, but still with 24 lands. I think that's too much for a deck that tops on 3CMC. 22 should be fine for it. But this list seems a little better than his first one.
The Choke in the SB could be lots of things, including: 2nd Hooligan, Finks, Declaration in Stone, Gut Shot, Thalia, etc. I'm not really sure what I want there. But the other 74 I'm pretty confident in.
All of this land talk and reflecting on some hands that I've had to keep that have gone bad has made me want to add a land. Even burn is playing 20 lands these days. I think the shell is powerful enough that we just want to make sure we can consistently hit 2 lands by turn 2. I'm probably shaving a Kird Ape for another fetch, and might shave something else for another fetchable.
Burn doesn't always use 20. The average based on mtgtop8 data is about 19.6, and I've personally been on 19 for a long time in Burn.
I don't think that this deck should play 20 lands, because this deck has a lot of pseudo-free spells. By going to 19, you decrease the probability of 1-landers by about 2%, decrease 2-landers by about 0.6%, and increase 3 landers by about 2%. Also, by going to 19, you increase the probability that you draw exactly 1 more land by T2 on the draw given that you started with 1 land from 44.4% to 45.7%. On the play, the probability that you hit land 2 on T2 goes from 32% to 33.9%.
What's all of this mean? The probability of an N card 7 for 18 lands is (in percent):
0 6.99
1 24.45
2 33.70
3 23.65
4 9.10
5 1.91
6 0.20
7 0.01
The probability that a 1 lander on the play turns into 2 lands by T2 on the play is 32% and on the draw is 44.4% (about 10% to become 3), which means that of those 24 games on the play, 7.8 pan out by T2 and 16.6 don't. On the draw, 13.2 games pan out (since hitting lands on T1 and T2 means that it panned out) and 11.2 don't.
For 19 lands:
0 5.82
1 22.12
2 33.18
3 25.41
4 10.7
5 2.47
6 0.29
7 0.01
The probability that a 1 lander on the play turns into 2 by T2 on the play is 33.9% and on the draw is 45.7% (about 11% to become 3), of 22 games on the play, 7.5 pan out by T2 on the draw and 14.6 don't. On the draw, 12.6 pan out and about 9.5 don't.
The total effect is that in 200 games (100 on the play, 100 on the draw), 27.8 games are 1 landers that don't pan out with 18 lands and and 24.2 are 1 landers that don't pan out with 19 lands. Simultaneously, you've changed the probability of 4+ and 0 landers.
For 20, this goes to 18.9 games per 200 pan out, 20.8 do not (as far as 1 landers).
Finally, let's define all 2s and 3s as keepable, all 0s and 4+s as unkeepable, and let's choose to keep all 1s (since the probability of a given 1 lander being keepable is approximately identical for each configuration).
For 18 lands, 67.9% of 7 card hands are keepable or 1 landers that pan out and 32.1% are unkeepable of 1 landers that don't pan out.
For 19 lands, 68.6% and 31.4%. For 20 lands, 68.8% and 31.2%.
I recognize that this ignores the consequences of a mulligan in each case, since 18 lands will produce fewer keepable 6 card hands. It also ignores that an 18 that does pan out is less likely to flood and therefore more likely to be very explosive. I suspect that this effect is large enough that the 0.2% you gain by going from 19 to 20 is outweighed by it and therefore 20 is too many lands. The difference is still only 0.7% for going from 18 to 19. In comparison to Burn, which I support 19 as a minimum for, this build is loaded with psuedo free spells which I think provides justification to play 18 but no fewer. Burn can run fine on 2 lands but is pretty happy to see a 3rd. I feel like we want 2 lands and find less value in a 3rd. I don't think 19 is wrong, but I do think 20 is too many. I will stick to 18, though.
19 seems right to me. I looked at the probabilities before coming to this conclusion, but a big piece of my reasoning for <20 is that burn plays 4 Searing Blaze, so the extra land late is pretty useful for them many times, whereas our "landfall" stuff can be turned out by trading in combat (and turning on HH's revolt isn't so important late when our hand is empty).
I think you're wrong that we don't want a 3rd land though. We have to trigger our Bushwhackers somehow, and on two land there are only 5-6 ways to trigger him.
19 seems right to me. I looked at the probabilities before coming to this conclusion, but a big piece of my reasoning for <20 is that burn plays 4 Searing Blaze, so the extra land late is pretty useful for them many times, whereas our "landfall" stuff can be turned out by trading in combat (and turning on HH's revolt isn't so important late when our hand is empty).
I think you're wrong that we don't want a 3rd land though. We have to trigger our Bushwhackers somehow, and on two land there are only 5-6 ways to trigger him.
I don't mean that we don't want land 3, just that land 3 is less valuable than land 2. With 2, you're still putting a lot of power on the board and swinging with it and you can still cast Atarka's Command on T3 with 2 lands. I don't think 19 is wrong, but I do think 20 is wrong. I'd drop an Experiment One if I was going to 19.
I already dropped one. Dropping 2 doesn't seem right with 4 HH, hence shaving the Ape.
Right, I forgot you dropped one already for a Ghor-clan. I'll think about going to 19, but I think there's enough free stuff to stay at 18. I may do it just to get another Stomping Ground.
Yes to all. E1 is sometimes awkward in multiples and a terrible topdeck, though shaving too many can't be right because green one drops are at a premium with HH. HH is the reason this deck is even worth considering again because of how often you get to put 3 or 4 or 5 creatures into play on turn 2. Atarka's Command is the best card in the deck and you want to see it every game. Less than 18 lands with no cantrips makes it really hard to hit 2 lands on time. With 18 lands, you already have a 31.4% chance of having zero or one lands in your opener, and those 1 landers are only a little better than 50% to find a land in two drawsteps.
For completeness, the attached table is the probability of having EXACTLY X lands in your opening 7 with 18 lands in the deck.
The only placing list with AER has 0 HH; that's why I ask.
What % of 1-landers do you typically keep? 2? 3?
I keep a lot more than 2-3% of landers, but mostly those that are fairly strong if they hit the 2nd land. For example: Fetch, Ape, Nacatl, X1, HH, BTE is an easy keep no matter what the last card is, especially on the draw. I don't really know what the % is, but somewhere between 20-50% doesn't sound outrageous. 1 land hands that don't have a BTE or HH I'm suspcious of, though I don't want to outright say I wouldn't keep any of them. Still learning the deck.
This isn't something I've recorded in my testing, but it's something that I could go back and do this evening. However, it would take a lot of time and not be very productive I think. The probabilities for keepable hands can be worked out analytically anyway.
I would keep any 2 or 3 lander with enough creatures (too few creatures is very unlikely when you play 32 of them). I'd mulligan pretty much all 4 landers. I'd keep 1 landers with multiple 1 drop creatures, since it allows me to build up a board presence on 1 land.
No, I can't. I would literally have to watch cockatrice replays of 200+ games to do it and I don't think that would be productive because it can be done analytically.
This is just math with the hypergeometric distribution and it could be worked out analytically. You don't need to arbitrarily say "1-landers are keepable 50% of the time" because you can write down the criteria for a keepable 1 lander and then calculate the probability of drawing a keepable 7 with 1 land.
Sometimes analytical solutions are less convenient than experimental ones.
I already answered that.
An analytical solution is not less convenient here. This isn't some crazy theory that has to be tested. There is an exact solution to the question you're asking. Any experimentally determined answer will have error bars, but the analytical solution is exact.
This literally amounts to writing down the probability that a given 7 card hand has 2 or more 1-drops and then calculating that probability and that is equal to the probability that a 1 land hand is keepable.
Any 2-3 lander is keepable provided that it's not just non-creature spells. This is also a probability that you can write down analytically.
What's your current list? Anything spicy?
You have to come up with a set of requirements either way. The best you're actually going to do is an approximation and it won't be sufficiently different from the hypergeometric function for the number of lands in your opener that spooly already posted. 2s and 3s are pretty much always fine. 4s aren't. 1s are if they have enough 1 drops and that can be calculated. Frankly, the exercise of thinking about what's keepable and what isn't is a more valuable exercise than writing down a probability.
I don't keep 4s. I definitely wouldn't keep one with a 1/1 and 2 commands because that hand is complete garbage.
6 one drops and one land? Sure, because I'll be putting power on the board for ages with it, which buys time to draw lands.
My list is in my signature.
Caleb Durward posted a new iteration of Landfall Rallier on CFB, but still with 24 lands. I think that's too much for a deck that tops on 3CMC. 22 should be fine for it. But this list seems a little better than his first one.
WGUBR 5c Humans
GWR Naya Zoo
Legacy:
GW GW Maverick
R Goblins
4 Arid Mesa
4 Wooded Foothills
1 Bloodstained Mire
1 Verdant Catacombs
1 Stomping Ground
1 Sacred Foundry
1 Temple Garden
1 Mountain
1 Forest
4 Goblin Guide
3 Kird Ape
3 Experiment One
4 Narnam Renegade
4 Wild Nacatl
4 Burning-Tree Emissary
4 Reckless Bushwhacker
4 Atarka's Command
4 Lightning Bolt
2 Ghor-Clan Rampager
1 Mutagenic Growth
2 Destructive Revelry
1 Tin Street Hooligan
2 Lightning Helix
1 Phyrexian Unlife
4 Path to Exile
2 Domri Rade
2 Grafdigger's Cage
1 Choke
The Choke in the SB could be lots of things, including: 2nd Hooligan, Finks, Declaration in Stone, Gut Shot, Thalia, etc. I'm not really sure what I want there. But the other 74 I'm pretty confident in.
Burn doesn't always use 20. The average based on mtgtop8 data is about 19.6, and I've personally been on 19 for a long time in Burn.
I don't think that this deck should play 20 lands, because this deck has a lot of pseudo-free spells. By going to 19, you decrease the probability of 1-landers by about 2%, decrease 2-landers by about 0.6%, and increase 3 landers by about 2%. Also, by going to 19, you increase the probability that you draw exactly 1 more land by T2 on the draw given that you started with 1 land from 44.4% to 45.7%. On the play, the probability that you hit land 2 on T2 goes from 32% to 33.9%.
What's all of this mean? The probability of an N card 7 for 18 lands is (in percent):
For 19 lands:
The total effect is that in 200 games (100 on the play, 100 on the draw), 27.8 games are 1 landers that don't pan out with 18 lands and and 24.2 are 1 landers that don't pan out with 19 lands. Simultaneously, you've changed the probability of 4+ and 0 landers.
For 20, this goes to 18.9 games per 200 pan out, 20.8 do not (as far as 1 landers).
Finally, let's define all 2s and 3s as keepable, all 0s and 4+s as unkeepable, and let's choose to keep all 1s (since the probability of a given 1 lander being keepable is approximately identical for each configuration).
For 18 lands, 67.9% of 7 card hands are keepable or 1 landers that pan out and 32.1% are unkeepable of 1 landers that don't pan out.
For 19 lands, 68.6% and 31.4%. For 20 lands, 68.8% and 31.2%.
I recognize that this ignores the consequences of a mulligan in each case, since 18 lands will produce fewer keepable 6 card hands. It also ignores that an 18 that does pan out is less likely to flood and therefore more likely to be very explosive. I suspect that this effect is large enough that the 0.2% you gain by going from 19 to 20 is outweighed by it and therefore 20 is too many lands. The difference is still only 0.7% for going from 18 to 19. In comparison to Burn, which I support 19 as a minimum for, this build is loaded with psuedo free spells which I think provides justification to play 18 but no fewer. Burn can run fine on 2 lands but is pretty happy to see a 3rd. I feel like we want 2 lands and find less value in a 3rd. I don't think 19 is wrong, but I do think 20 is too many. I will stick to 18, though.
I think you're wrong that we don't want a 3rd land though. We have to trigger our Bushwhackers somehow, and on two land there are only 5-6 ways to trigger him.
I don't mean that we don't want land 3, just that land 3 is less valuable than land 2. With 2, you're still putting a lot of power on the board and swinging with it and you can still cast Atarka's Command on T3 with 2 lands. I don't think 19 is wrong, but I do think 20 is wrong. I'd drop an Experiment One if I was going to 19.
I already dropped one. Dropping 2 doesn't seem right with 4 HH, hence shaving the Ape.
Right, I forgot you dropped one already for a Ghor-clan. I'll think about going to 19, but I think there's enough free stuff to stay at 18. I may do it just to get another Stomping Ground.