Art: David Palumbo
twitch.tv/sierkovitz
18 May 2024
Support 17Lands.com
Magic Numbers #119
Finding deck plans using data
Preamble
We are all prone to conspiracy theories
It is how you respond while challenged
That makes the difference
Figuring the plans
There are two easy data points that tell us:
What’s the plan for a given color pair
How good are those color pairs in implementing the plan
Win rate for games ending at turn X
% of games ending on turn X
Figuring the plans
The two metrics tell them about two aspects of the color pair
Win rate tells us what is the plan and how good is the deck once the plan is implemented
Distribution of game ends tells us how reliable is the deck if things go according to plan
Figuring the plans
Both need to work
Deck that wins 100% long games but gets to a long game 10% of the time might not be great
Deck that wants to play aggro, but only has 52% win rate in games that end early is not going to be great either
Figuring the plans
Knowing which of the two fails is a key to solving formats
Because you can address the right problem in draft and deckbuilding
Win Rate at Turn X
This metric shows us several things:
When in a game does a deck win most?
Great WR early - low in late game (usually aggro)
Decent WR any time, preference to mid-game (midrange)
Terrible WR early, good WR late (control)
Great WR only in mid-game (mid-term)
Win Rate at Turn X
Turn cycle game ended
Control
Midrange
Aggro
Aggro
Control
Mid-term
Aggro
Midrange
Aggro
I don’t even
Win Rate at Turn X
Wants to play long game but even if games go long doesn’t win much - power or build problem
Win Rate at Turn X
Looks OK at every stage of the game - really solid midrange numbers
Win Rate at Turn X
Decent, but early win rate not amazing
Win Rate at Turn X
Can demolish early, and has some staying power
Win Rate at Turn X
When plan of going to late game works - it works well
Win Rate at Turn X
No plan established,
Doesn’t win
Win Rate at Turn X
Hard to say what it wants to achieve
Win Rate at Turn X
Wants to be aggro but doesn’t win much, looks like a power issue
Win Rate at Turn X
Decent in early-mid game but has problems going into the late game
Win Rate at Turn X
Great early, but drops off drastically in the late game. Needs to win more long games to be a real contender
Percent of games ending on turn X
This metric tells us: is the deck capable of enforcing the plan consistently?
Percent of games ending on turn X
WU is actually good in getting till late game. So if you can fix the problem of power / winning late games - you are golden
Percent of games ending on turn X
WB can make games last and wins well enough in those games and spans a wide range
Percent of games ending on turn X
WR is also good in implementing its plan - improving the builds to bump win rate is key
Percent of games ending on turn X
WG struggles a bit with ending the games early
Probably needs to focus on early game a bit more to be at its best
Percent of games ending on turn X
UB is great at ending games in turns 11-14. But there is a drop-off in win rate if games go longer than that. Think of ways to end games in UB.
Percent of games ending on turn X
UR does poorly. But - it can end games as a mid-term deck. So won’t kill early, but doesn’t want a long game. You should reinvent it with that in mind.
Percent of games ending on turn X
BR doesn’t win enough in early games, but also doesn’t end games early often enough. This might be an unsolvable issue.
Percent of games ending on turn X
BG decks go very late, but are not that great in long games. Think of killing opponent before they draw into stuff you can’t deal with.
Percent of games ending on turn X
RG decks are capable of ending games early better than others. And rarely go long. Focus on making sure games end rather than improving late game.
From data to solutions - UW
First: cards that help with the plan
On average - 35% of games go past turn 10
Which cards help the deck go longer?
Percent of games ending on turn X
WU is actually good in getting till late game. So if you can fix the problem of power / winning late games - you are golden
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
Seconds - the problem with the color pair was that even if it manages to go long - the win rate is still not spectacular
Cards that are in the decks that win a lot of long games should be a hint for what are we looking for
Win rate in games that last more than 10 turns
Win Rate at Turn X
Wants to play long game but even if games go long doesn’t win much - power or build problem
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - UW
From data to solutions - WR
First: cards that help with the plan
On average - 35% of games go past turn 10
Which cards help the deck go longer?
Win Rate at Turn X
Decent, but early win rate not amazing
Percent of games ending on turn X
WR is also good in implementing its plan - improving the builds to bump win rate is key
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - WR
From data to solutions - BG
You can go about it in two ways:
1. Making sure you kill opponents early enough
2. Make sure your long game improves
Win Rate at Turn X
Decent in early-mid game but has problems going into the late game
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
Win Rate at Turn X
Decent in early-mid game but has problems going into the late game
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
From data to solutions - BG
Acknowledgements
17Lands Team:
Viralmisnomer: @17lands
ztm: @ale_ballini
JohnnyDeeWSC
GrantWu
Art: @fakejakebrowne