Theory
How Many Lands Should You Include in Your Cube?
One decision that is universal to almost every cube is the question of how many lands to include. In this article, we’ll examine a number of benchmarks that might inform what percentage of your cube should be dedicated to mana fixing and utility lands and help you choose a manabase that aligns with your cube design goals.
Retail Limited
“Manabases in retail draft decks are terrible.”
Many designers, consciously or not, have rooted their environment in their experience drafting retail sets. It’s the only non-cube precedent we have for what a draft ought to feel like. Given that this is one of our few guiding lights, it’s important to recognize that manabases in retail draft decks are terrible. It’s a fact of Limited that you will win and lose more games to drawing too few, too many, or the wrong kind of lands than in Constructed.
We can attempt to quantify just how bad the manabases are in retail Limited by looking at the ASFAN for mana fixing lands in a given set. ASFAN is a term used by Magic R&D to describe how much of a certain type of card shows up in a booster pack on average, and when multiplied by the three packs we open in a typical draft, we can estimate how much of that effect each player will have access to in an average draft pool.
Fixing Land ASFAN | Avg. Number of Fixing Lands in Draft Pool | Equivalent Fixing Lands in 360 Card Cube | |
---|---|---|---|
Guilds of Ravnica | 1.1 | 3.3 | 26.4 |
Ravnica Allegiance | 1.1 | 3.3 | 26.4 |
Theros Beyond Death | 0.16 | 0.5 | 4 |
Modern Horizons | 0.14 | 0.4 | 3.2 |
War of the Spark | 0.1 | 0.3 | 2.4 |
Dominaria | 0.08 | 0.25 | 2 |
Throne of Eldraine | 0.05 | 0.16 | 1.28 |
Zendikar Rising | 0.07 | 0.2 | 1.6 |
Kamigawa: Neon Dynasty | 0.98 | 2.94 | 23.52 |
Streets of New Capenna | 1.02 | 3.05 | 24.4 |
In many sets, we are not likely to even have a single fixing land in our pool — players are almost always building a manabase from basics alone. Expansions with explicit multicolor themes like Guilds of Ravnica and Streets of New Capenna include markedly more fixing lands, but still max out around an ASFAN of 1, which is comparable to a cube containing approximately 7% fixing lands. Assuming all of the non-basics we draft are on-color, 3 fixing lands out of the 16-17 we’ll be running in our deck is still a pitiful density by constructed standards.
It’s important to remember that lands are not the only source of mana fixing in most sets. Mana fixing in some capacity is a core part of green’s identity, with most expansions featuring a mana dork or Rampant Growth effect in addition to some form of mana rock or bauble that any deck can play. While these non-land sources of fixing make it harder to capture relative availability of mana fixing from set to set, cubes often include similar effects, so comparing these values is still informative.
Wizards has a lot of complex, often conflicting priorities when designing a set. They have a limited number of slots in each expansion, and with them must print cards for Constructed players of all stripes, tell a story about the plane, highlight the set’s core mechanics in sufficient density, and excite players of wildly different experience with and investment in the game, all in addition to creating an engaging draft format. The amount of fixing players have when drafting a retail set is the result of all of these factors. As cube designers, unencumbered by this complicated array of priorities, we are free to ignore this inherited benchmark and make our mana bases suit our draft and gameplay goals.
Constructed
In sharp contrast with drafted manabases, Constructed players get to build decks with essentially however many fixing lands they want. Each format has its own pool of available fixing of different power levels (meaning, the costs of the fixing lands relative to basics are of different magnitudes) but deckbuilders get to run what they believe to be the optimal number. These decks aim to reduce, as much as possible, the number of games decided by mana issues, which constitutes an alternative benchmark for manabase quality:
Fixing Lands | Fixing Land Ratio | Equivalent Fixing Lands in 40 Card Deck | Equivalent Fixing Lands in 360 Card Cube | |
---|---|---|---|---|
Historic Uro Midrange | 21 | 0.35 | 14 | 126 |
Modern Uro Omnath | 18 | 0.3 | 12 | 108 |
Modern RW Burn | 16 | 0.27 | 11 | 96 |
Legacy Temur Delver | 15 | 0.25 | 10 | 90 |
Modern UR Blitz | 15 | 0.25 | 10 | 90 |
Standard Esper Doom Foretold | 20 | 0.25 | 10 | 90 |
Legacy UB Ninjas | 14 | 0.23 | 9 | 84 |
Legacy Snowko | 13 | 0.21 | 9 | 78 |
Modern RG Midrange | 13 | 0.21 | 9 | 78 |
Standard UB Flash | 12 | 0.2 | 8 | 72 |
Legacy BG Depths | 11 | 0.18 | 7 | 66 |
Standard RG Aggro | 10 | 0.17 | 7 | 60 |
Pauper RW Monarch | 9 | 0.15 | 6 | 54 |
Standard UR Tempo | 8 | 0.13 | 5 | 48 |
Pauper UB Delver | 8 | 0.13 | 5 | 48 |
Pauper UR Faeries | 7 | 0.11 | 5 | 42 |
This sampling of decklists is by no means comprehensive, and obviously decks come and go in constructed metas, but nevertheless we can see that the ratios of constructed decklists dedicated to fixing lands varies dramatically. There are many contributing factors. Three and four color decks require more fixing than two color decks. Some decks have access to powerful non-land fixing which takes pressure off of the fixing lands themselves. Some Legacy decks succeed at using Brainstorm in conjunction with fetches to fix their mana and are disincentivized to play an overabundance of non-basic lands because of the existence of cards like Wasteland and Blood Moon. Different formats run at different speeds and therefore punish stumbling on mana to different degrees. Pauper decks don’t have access to untapped dual lands but can’t afford to fill up on enters-the-battlefield-tapped lands and play off curve all game.
There is one more important factor to account for here: the above “equivalent fixing lands in 360 card cube” only works as a comparison point if every single non-basic land is maindecked by our drafters, which is not realistic. How many non-basics are likely to end up in sideboards, casualties of a draft changed direction part-way through or just the luck of the packs? This is an incredibly complicated question to answer, as it’s subject to a variety of confounding factors, not the least of which is subjective bias on the part of human drafters. So, instead of trying to model this question, let’s look at actual data.
“On average, players maindeck roughly ⅔ of their fixing lands, with the remaining ⅓ being off color.”
A survey of over 75,000 deck lists drafted from more than 5,600 unique cubes on Cube Cobra can shine some light on how many fixing lands end up maindecked and how many languish in sideboards.1 On average, players maindeck roughly ⅔ of their fixing lands, with the remaining ⅓ being off color. The following table is broken down by the density of fixing lands available in the cube, and as we can see, cubes with a higher percentage of fixing lands also end up with a higher percentage of their fixing lands maindecked.
Fixing Land Density in Cube | Number of Decks | Average Maindeck Fixing Lands | Average Sideboard Fixing Lands | Maindecked Fixing Land Percentage |
---|---|---|---|---|
2%-3% | 433 | 0.75 | 0.52 | 59.23% |
3%-4% | 1467 | 1.06 | 0.96 | 52.49% |
4%-5% | 1578 | 1.23 | 0.91 | 57.33% |
5%-6% | 3426 | 1.44 | 0.85 | 62.84% |
6%-7% | 5613 | 1.89 | 1.20 | 61.15% |
7%-8% | 5549 | 2.05 | 1.33 | 60.66% |
8%-9% | 10224 | 2.37 | 1.50 | 61.19% |
9%-10% | 12543 | 3.04 | 1.48 | 67.27% |
10%-11% | 11183 | 3.43 | 1.51 | 69.36% |
11%-12% | 10836 | 3.82 | 1.68 | 69.44% |
12%-13% | 9285 | 4.54 | 1.36 | 76.96% |
13%-14% | 2037 | 4.02 | 1.50 | 72.86% |
14%-15% | 1207 | 3.87 | 1.74 | 69.01% |
15%-16% | 280 | 4.83 | 1.49 | 76.48% |
16%-17% | 88 | 4.16 | 1.25 | 76.89% |
17%-18% | 68 | 3.49 | 1.19 | 74.53% |
18%-19% | 24 | 3.92 | 1.08 | 78.33% |
If we’re aiming for draft decks with manabases similar to constructed decks, we need to account for the inherent variance in draft by packing more fixing lands than we want to end up in our players’ maindecks. The following table looks at the same constructed decklists from above, comparing how many fixing lands we would need to include in a cube assuming an average of 75% of lands make the maindeck. (According to our data from Cube Cobra, 67% of fixing lands are maindecked on average, but we observed in the prior table that as the density of fixing goes up, the ratio of maindecked fixing does as well.)
Equivalent Fixing Lands in 360 Card Cube (100% Maindecked) | Equivalent Fixing Lands in 360 Card Cube (75% Maindecked) | Equivalent Fixing Lands in 360 Card Cube (67% Maindecked) | |
---|---|---|---|
Historic Uro Midrange | 126 | 168 | 189 |
Modern Uro Omnath | 108 | 144 | 162 |
Modern RW Burn | 96 | 128 | 144 |
Legacy Temur Delver | 90 | 120 | 135 |
Modern UR Blitz | 90 | 120 | 135 |
Standard Esper Doom Foretold | 90 | 120 | 135 |
Legacy UB Ninjas | 84 | 112 | 126 |
Legacy Snowko | 78 | 104 | 117 |
Modern RG Midrange | 78 | 104 | 117 |
Standard UB Flash | 72 | 96 | 108 |
Legacy BG Depths | 66 | 88 | 99 |
Standard RG Aggro | 60 | 80 | 90 |
Pauper RW Monarch | 54 | 72 | 81 |
Standard UR Tempo | 48 | 64 | 72 |
Pauper UB Delver | 48 | 64 | 72 |
Pauper UR Faeries | 42 | 56 | 63 |
These numbers are still very rough approximations, but we can draw some safe-enough conclusions about land counts that would, on average, result in constructed-level mana fixing:
- To emulate two-color Pauper or Standard decks, 60 to 100 fixing lands should be included in 360-card cubes (17%-28%)
- To emulate two-color Legacy and Modern decks or 3+ color Standard decks, 100 to 140 fixing lands should be included in 360-card cubes (28%-39%)
- To emulate 3+ color decks in Legacy, Modern, or Historic, 120 to 180 fixing lands should be included in 360-card cubes (33%-50%)
That a ton of lands — many times more than the 8%-12% included in most cubes on Cube Cobra. Before we investigate whether or not this actually desirable, or even possible, let’s check our math.
Frank Karsten’s Manabase Math
Rather than relying on precedent from Limited or Constructed decks, we can derive mathematical approximations from first principles to determine what our manabases ought to look like. Calculating the optimal number of lands to reliably cast spells in 40-card decks is no easy task, but luckily Frank Karsten has done all the heavy lifting already. In his seminal How Many Colored Mana Sources Do You Need to Consistently Cast Your Spells?, Frank answers this exact question:
Colored Mana Sources Required to Reliably Cast Spells
Mana cost | Example spell | Number of Colored Mana Sources Required |
---|---|---|
Primordial Mist | 6: | |
Restoration Angel | 7: | |
Goblin Rabblemaster | 8: | |
Kari Zev, Skyship Raider | 9: | |
Grave Titan | 9: | |
Llanowar Elves | 10: | |
Mystic Confluence | 10: | |
Wrath of God | 11: | |
Liliana of the Veil | 12: | |
Vorapede | 13: | |
Eidolon of the Great Revel | 14: | |
Cryptic Command | 14: | |
Benalish Marshal | 16: |
As we can see from Frank’s work, the answer for how many duals lands you need for a two-color deck is not cut-and-dry. A mostly green deck can splash Primordial Mist somewhat easily, but you’re only going to be able to reliably curve Eidolon into Marshal if every single one of your lands is a dual land. Let’s look at a couple of more common examples of cards we might want to run alongside one-another:
Most Demanding Color A Mana Cost | Most Demanding Color B Mana Cost | Fixing Lands Required (17 Land Deck) | Fixing Lands Required (16 Land Deck) | Fixing Lands Required (15 Land Deck) |
---|---|---|---|---|
1 | 2 | 3 | ||
2 | 3 | 4 | ||
3 | 4 | 5 | ||
4 | 5 | 6 | ||
4 | 5 | 6 | ||
5 | 6 | 7 | ||
6 | 7 | 8 | ||
8 | 9 | 10 |
Keep in mind that the “most demanding spell” means that no other spells in the same color can be harder to cast. So, if we consider our first row — Isamaru and Rabblemaster — these numbers only hold true if all of the spells in our hypothetical Boros aggro deck cost ,
,
,
, or
. That means no
,
spells, or
spells. While this specific pairing doesn’t require too much fixing, in many environments it’s unlikely we’ll be able to draft a deck that fits those parameters.
It’s important to note that Frank is not describing the ideal number of sources for a given set of spells, he’s describing the minimum, by his definition. From the perspective of an individual player, more fixing is always better. For a blue-white control deck to play Counterspell alongside Wrath of God an optimistic minimum of 8 fixing lands is required to meet Frank’s standard for making both reliably castable. If we assume that the average two-color, 40-card draft deck will want 5-8 fixing lands, we can extrapolate to find the cube totals:
Fixing Lands per Deck | Equivalent Fixing Lands in 360 Card Cube (100% Maindecked) | Equivalent Fixing Lands in 360 Card Cube (75% Maindecked) | Equivalent Fixing Lands in 360 Card Cube (67% Maindecked) |
---|---|---|---|
5 | 40 | 53 | 60 |
6 | 48 | 64 | 72 |
7 | 56 | 75 | 84 |
8 | 64 | 85 | 96 |
This, like our example Pauper and Standard decklists, suggests that between 60-100 fixing lands would be a good minimum at 360 to ensure that two-color decks are able to reliably cast their spells (according to Frank Karsten’s definition of “reliably”). Cubes that want to emulate 3-color Standard decks or higher-quality manabases should, of course, aim for a higher quantity.
How many lands is too many lands?
If precedent from Constructed and Frank Karsten’s math suggests that some cube designers may want to run a great deal more lands than most lists on Cube Cobra, this begs the question: how many lands can we actually squeeze into a 360 card cube? As an upper limit, assume that eight drafters will need 184-200 maindecked, nonland cards (23-25 spells apiece), theoretically leaving the remaining 176 to 160 cube slots open for mana-fixing. This logic has a hole, though, since drafting involves too much variance and risk for any one drafter to have zero off-color or otherwise unplayable cards, much less all eight drafters.
In order to calculate a more realistic upper limit for mana-fixing density, we need to know the average number of uncastable cards drafters end up with. Once again, our Cube Cobra data comes in handy!
Fixing Land Density in Cube | Number of Decks | Average Uncastable Cards in Pool |
---|---|---|
2%-3% | 433 | 7.79 |
3%-4% | 1467 | 7.91 |
4%-5% | 1578 | 7.67 |
5%-6% | 3426 | 7.65 |
6%-7% | 5613 | 8.01 |
7%-8% | 5549 | 8.23 |
8%-9% | 10224 | 8.00 |
9%-10% | 12543 | 7.60 |
10%-11% | 11183 | 7.33 |
11%-12% | 10836 | 7.03 |
12%-13% | 9285 | 6.45 |
13%-14% | 2037 | 7.07 |
14%-15% | 1207 | 7.13 |
15%-16% | 280 | 6.62 |
16%-17% | 88 | 7.28 |
17%-18% | 68 | 7.94 |
18%-19% | 24 | 8.33 |
“96-120 lands is the most you can include at 360 before players start ending up short on playable, non-land cards.”
As the above table shows, the average number of unplayable cards per deck actually doesn’t appear to change much based on the amount of fixing in the cube — it basically hovers around 7 or 8. Eight drafters will therefore need 240-264 spells to fill out their draft pool (23-25 maindeck and 7-8 uncastable), leaving 96-120 slots for lands. Therefore, in most environments, 96-120 lands is the most you can include before players start ending up short on playable, non-land cards. Note that unlike our previous calculations, this applies to all lands, not just fixing lands. It may sound risky to fly so close to the sun, but remember that players have agency over their draft and will change their priorities as the draft develops to try and arrive at a functional deck.
Competition for Land Slots
Ultimately, how many lands you include in your cube depends on your design goals. We’ve established that the most lands you can reasonably include before players, on average, start coming up short on playables is between 96 and 120. But what are the ramifications of including such a high density of lands? In a given draft pool, if you set aside the maindecked spells and incidental, off-color draft picks, the remaining cards represent a number of important features defining your draft environment.
These cards belong to a couple of categories:
- All lands, including fixing lands and utility lands
- Cards that are conditional or matchup dependent, such as sideboard cards
- On-color cards that are unplayable in some pools, such as dedicated build arounds and combo pieces
- On-color, but difficult-to-cast cards you may not have the requisite fixing for
- On-color cards that don’t belong to one of the former groups; just your least powerful or appealing options
This proportion of each deck’s pool, roughly 1/3rd, can be extrapolated to the cube as a whole. With each draft, 1/3rd of the on-color cards your players consciously drafted cannot be maindecked as spells — there is simply no room. Though it’s a common design choice to fill cubes with slightly below-rate cards meant as redundancy for the premier options, many of these cards fall in that 1/3rd margin of the cube. Because their rate is relatively weak, the below-rate redundant spells tend to ride sideboards the majority of the time instead of seeing meaningful play.
One way to activate this space is to fill it with lands, which allows drafters to play a larger proportion of their drafted cards. If you do choose to run a lower density of fixing, your players will have a relatively large pool of spells they can’t fit in their maindeck. Activate those slots by filling the margins of your cube with narrow build-arounds, sideboard cards, and other committal picks that have a big impact when they are maindecked. To me, it’s a good sign when I look through my players’ sideboards after a draft and see cards that could, if the draft had broken differently, have been exciting and powerful cards in different decks, rather than just the least powerful cards in the cube.
Going Even Higher
All of the numbers outlined in this article assume a typical 360 card draft, with each player ending up with a 45 card draft pool, but one of the most beautiful aspects of Cube is that none of these numbers are set in stone. If you’d like to experiment with even better mana fixing in draft, you can simply increase your pack size, so each player gets more picks and you can afford to dedicate more slots to non-basic lands. Or decrease your minimum deck size, so players need fewer spells to fill out their deck. 16 card packs, resulting in a 384 card pool for an 8 player draft, have been increasingly popular amongst some Cube designers for largely this reason.
Conversely, if you decide to draft only a portion of your cube (e.g. a 6-person draft of 360, or a 8-person draft of a 540-card cube), then you can use hypergeometric calculation to adjust the benchmarks given throughout the article to account for the variability of a partially drafted cube. For example, to yield an 80% chance of at least 90 fixing lands in a 360-card draft of a 540-card cube, you’d need about 140 fixing lands out of 540.
What About Five Color Good Stuff?
There is a common misconception that an abundance of mana fixing causes a cube to devolve into “five-color good stuff” decks, which many designers abhor. The theory is that with enough mana-fixing lands, players can just take any and all non-basics and every powerful card they’re passed, regardless of color, and wind up with a competitive deck. In reality, the effectiveness of this strategy is a confluence of many factors, with availability of mana-fixing being one of the least impactful.
“Good mana-fixing makes all of the decks in your cube better, including the aggressive ones.”
The amount of mana-fixing available does not change the relative risk of playing additional colors in a given deck. Whether there is an abundance of dual lands in your cube or none, players will have to choose between playing a two-color manabase or a relatively less reliable three or more color manabase. Playing more colors always makes your mana more precarious. We can see this fact played out in Constructed formats. If playing three, four, or five colors came at no cost, the Constructed decks would use their access to effectively limitless fixing and would all be many-colored. Yet, a survey of any Constructed meta at almost any time reveals plenty of one- and two-colored decks. Conversely, playing three or more colors is often correct in retail Limited decks, especially in Sealed, which very rarely have access to any fixing lands, much less the kind of manabase we’ve recommended here. Clearly the availability of fixing lands is not the primary factor in the viability of many-colored decks.
Instead, the successful decks in a given meta are a result of the complete pool from which they’re constructed. Whether a three or more color deck is viable usually has a lot more to do with individual power outliers than the available non-basic lands. Omnath, Locus of Creation was to blame for the greedy piles that dominated Standard in the fall of 2020, and we saw the metagame settle on fewer-colored decks in the wake of its ban. Arcum's Astrolabe proved to be a problem just about everywhere by allowing decks to play many colors regardless of what their manabase looked like.
From a cube design perspective, availability of mana fixing is unique in that it affects all of the decks in your environment, excepting mono-colored ones, not just decks in three or more colors. If limiting the success of greedier decks is a goal of yours, there are many ways to address the issue that will not also hamstring two-color decks:
Limit the payoffs for playing many colors. This means making sure that power outliers, especially gold ones, don’t play well alongside one another. Kolaghan's Command is generically good and playable in almost any deck, but a card like Judith, the Scourge Diva requires a more specific context to shine, one that does not lend itself to just playing good cards from all the colors together.
Buff aggressive strategies. Fast, proactive decks should be able to punish their many-colored opponents, which are inherently slower and less consistent.
Narrow the delta in power level between the best and worst cards in your cube. If some cards are much better than others, the reward for playing powerful cards from many colors may outweigh the risk of a sketchy mana base.
Provide strong motivations to play fewer colors. Powerful spells with many single-color pips and competitive archetypes that are wholly contained within one or two colors offers clear lanes for drafting successful one- and two-color decks.
Boost the power of synergy. Greedier decks tend to thrive when individual card quality outshines decks with a synergistic plan greater than the sum of its parts.
Ultimately, while the availability of fixing does have some impact on the viability of greedier decks, cutting down on fixing lands is one of the least effective ways to limit their success.
Conclusions
The number of fixing lands you ought to include in your cube is entirely a matter of your design goals. However, having looked at the data from Cube Cobra and given feedback on countless new cubes over the years, my personal opinion is that most cubes should include more lands. The community has been trending this direction, and I expect it to continue.
I believe the average 8%-12% fixing density that can be observed on Cube Cobra at time of publishing is a result of many years of inherited rules-of-thumb, unsubstantiated by theory, that can ultimately be traced back to a cards-first, power-driven cube design ethos patterned from retail Limited formats. Designers focused on maximizing power have historically begun by putting the most individually powerful cards together and then pursuing the emergent themes, rather than starting with a cohesive design goal. By this approach, one would include only the most powerful lands (maintaining singleton, of course), regardless of their density, rather than setting a desired benchmark of fixing density based on gameplay goals.
In short, mana-fixing trends low because most cubes are singleton and people don’t want to put Sungrass Prairie in their cube. Or, more likely, most people base their initial list off a existing cube, and the most visible ones have historically been power-driven, and their designers didn’t want to put Sungrass Prairie in their cubes.
My goal is to enable designers with a robust set of tools, that they might make an informed decision for themselves. In summary:
- If most are your decks are two colors, and you want your players to be able to cast their spells reliably — according to Frank Karsten’s definition — you should include no fewer than 60 fixing lands in a 360 card cube (~17% of the cube).
- Including more than 96-120 lands in a 360 card cube is risky, as players may struggle to find enough non-land cards in their colors to construct a playable deck.
- Given that players only maindeck around half of the non-land cards drafted in a typical pod, non-basic lands should be weighed against sideboard cards, narrower build-arounds, cards with restrictive casting costs, and otherwise conditional cards. If your sideboards often include on-color cards that are just less appealing than other options in your pool rather than belonging to one of these categories you have some untapped design space to explore.
- Dominance of five-color “goodstuff” decks is more than likely due to under-performance of fast, proactive decks, multicolored power outliers, or big deltas in individual card power level than it is to abundance of fixing.
- If you’re trying to emulate the feel of retail Limited, you can calculate the ASFAN of fixing lands in that set and extrapolate to your given cube size. For many sets, this will result in you not including any fixing lands, as retail Limited decks have terrible mana.
- Given that all the draft data on Cube Cobra is collected drafting against bots, it’s likely that there are some differences between the patterns observed in these decks compared to decks drafted against other humans. This is the best data we have access to, and I believe it still provides a good baseline for analysis. If anyone from Wizards of the Coast is reading this and wants to give us a dataset of decklists from MTGO or Arena cubes, we would be very appreciative!↩