Deckbuilder's Toolkit - The Atlas

Greetings, Sorcerers! In the first installment of this series, I want to discuss my methodology for creating a consistent and reliable mana/threshold base for Sorcery decks. There are few things more frustrating than having an awesome spell in your hand and enough mana to cast it... only to realize that you don’t meet the proper threshold requirement!

If you just want to see the numbers with very brief explanations you can find them here: Deckbuilder's Toolkit: Calculating Threshold Requirements

Read on for a more in depth explanation. All card names have links to a full card image.

During the deck-building process everyone gets to choose what ratio of risk to reward they are willing to accept. My personal goal when brewing is to try to have around a 90% chance or higher of hitting threshold requirements on curve. This means that if I have Colicky Dragonettes in my deck (and I really want to cast it on turn 3), I aim to have 2 fire threshold by turn 3 at least 90% of the time.

Additional things that I consider are: how important it is for my deck to cast Colicky Dragonettes on turn 3 if I have it in my hand and how likely am I to have Colicky Dragonettes by turn 3? How many other 3-mana spells do I have in my deck that also require 2 fire threshold? How many 3-mana spells do I have in my deck that could be cast if I don’t have 2 or more fire threshold by turn 3?

If I have 3 Colicky Dragonettes in my Spellbook and they are my only double fire cards at 3 mana, I’m going to be less worried about hitting double fire by turn 3. If most or all of my 3-mana spells require double fire, then I’ll try to aim for even higher than 90% consistency.

While a well-constructed Atlas is a crucial element of any deck, I like to start the process of deck-building by focusing mainly on the Spellbook. Once I have a rough draft of the spells I want to include, I can take a more informed approach to managing how many elemental sites I'll need to include in the Atlas. Even though the primary initial focus is typically on drafting the Spellbook, I will also set aside sites with abilities that I want to consider. If my deck is playing a defensive strategy, I might set aside things like Free City, Troll Bridge, or Bottomless Pit. While some number of these utility sites may need to be excluded to accommodate threshold requirements in the end, it can be worthwhile to have them in mind during early drafts.

Once you have a general idea of how your Spellbook will look, it's time to start seeing if you can realistically cast those spells!

Before we get into the finer details, I want to outline some basic assumptions that I've considered in my calculations for the different tables below. These assumptions form an average baseline from which to operate but are not hard and fast rules for how to build or play your decks. After explaining these assumptions—and their ramifications for Atlas construction—I'll cover how to handle special cases.

Assumption 1: The average deck will mulligan up to 2 sites in an attempt to draw necessary elemental threshold. There are, of course, plenty of situations where you will mulligan a different number of sites. When considering deviations from this assumption in your own calculations, it is important to think about how your deck is likely to handle mulligans in most cases. Are you playing a hyper-aggressive, 2-element deck that has a very low curve in your Spellbook? You're probably going to mulligan fewer spells since it is unlikely that you will have many opening hands with cards that cost 4-9 mana. You're probably also unlikely to want to draw lots of extra sites during your draw steps to try to fix your threshold since you don't need a bunch of mana to operate. This means that you are much better off trying to get your threshold requirements worked out as early in the game as possible, and therefore, you will often mulligan more sites to dig for the requisite threshold.

If you're running a mono-element deck with a higher range of mana costs, you will likely mulligan spells more often for curve purposes, but also spend more of your draw steps on sites to reach higher mana numbers.

Assumption 2: After mulligans, the average aggro deck will draw a site on the first draw step in an attempt to hit the threshold requirements that are in the player's hand (you've likely got a turn 2 or turn 3 play in hand and now just need to get the threshold to play it.) After the first draw step, the average situation calls for drawing a site during 33% of the draw steps until relevant elemental threshold and mana requirements are met.

The average mid-range deck will follow a similar mulligan strategy but will draw a site 50% of the time until required threshold needs are met.

The average Pathfinder deck will activate to see a new site on each of the first 6 turns. Turn 7 will be used to reposition and then turn 8, 9, and 10 will be used to activate and see another site.

Assumption 3: You are on the play. This is only for the purpose of calculating the number of natural draw steps you have had on any given turn. When you are on the draw, it's going to be easier to meet threshold requirements because you see an extra card, but you (probably) don't want to build a deck that is going to struggle more on the play.

With these assumptions in mind, we gather the following:

Aggro:

Mulligans: 5 sites seen

Turn 1: (on the play) - 5 sites seen

Turn 2, 3 and 4: 6 sites seen (You will most likely draw another site on the second turn, so the table uses this as the baseline, but that won't always be the case so calculate accordingly if you deviate from this assumption.)

Turn 5: 7 sites seen

Turn 6 and 7: 8 sites seen

Mid-range:

Mulligans: 6 sites seen

Turn 1: (on the play) - 5 sites seen

Turn 2: 6 sites seen

Turn 3 and 4: 7 sites seen

Turn 5 and 6: 8 sites seen

Turn 7 and 8: 9 sites seen

Turn 9: 10 sites seen

Pathfinder:

Turn 1: 1 site seen

Turn 2: 2 sites seen

Turn 3: 3 sites seen

Turn 4: 4 sites seen

Turn 5: 5 sites seen

Turn 6 and 7: 6 sites seen

Turn 8: 7 sites seen

Turn 9: 8 sites seen

Turn 10: 9 sites seen

With these numbers in mind, let's take a look at the following tables:

Using these tables as a reference, one can quickly find my suggested number of threshold sources for a given turn/threshold requirement; e.g., if you are running a mid-range deck and you want to be able to cast a 5-mana, double-threshold card on turn 5—like Giant Shark—you will want 12 water sources in your deck to give you a 92.7% chance of success.

If you're running an aggro deck and have tons of 3-mana, double-air spells, it's going to be important for you to have 2 air threshold on turn 3, and so you'll want at least 15 air sources in your deck.

For cards like Browse or Divine Healing which have a mana cost of 1 but require triple threshold, the earliest you are likely to be able to cast these cards is turn 3. In the case of Divine Healing, however, you are unlikely to want to cast it until turn 5 or later. Because of this, I recommend calculating threshold requirements for Divine Healing in the Turn 5 & 6 or even 7 & 8 column leading to a 14-16 earth threshold source recommendation for mid-range decks.

While these tables are useful as a quick guide to help you figure out a general idea of how many elemental sources you should include in your deck, there are a lot of other factors to consider. What do you do if you want to make sure you have double fire by turn 3 but also have double water by turn 5? What if you are running multi-element spells like Black Knight or Sir Gawain? How do variables like Seer's ability, Land Surveyor, or Annual Fair change the equation?

First though, let's talk about the power behind the process. It's not magic, it's... hypergeometric math! It might sound intimidating, but with access to online calculators it can be easy to implement for deck-building. I personally like to use this calculator as it is very user-friendly. This tool allows you to input the following:

Population Size: total number of cards that are in your deck. In my examples, I’m using the new standard 30-card Atlas and 50-card Spellbook.

Sample Size: how many cards you will draw or have access to in the time frame you wish to calculate. Your opening hand allows you to look at 3 sites and 3 spells; the mulligan then allows you to look at another 0-3 cards in a combination of your choice of sites and/or spells. Each additional card you draw within the parameters of the equation will add to your Sample Size; e.g., if you look at your opening hand and decide that you want to mulligan all of your sites, you will have a sample size of 6 cards from the Atlas on turn 0. If you want to calculate the odds of drawing specific threshold by turn 3 and you are willing to mulligan all sites from your opening hand and draw from your Atlas during all of your first 3 draw steps, you can have a Sample Size of up to 8 cards on the play and 9 cards on the draw by the third turn.

Successes in Population: how many cards in the deck meet the requirements for the calculation. How many fire-threshold sites do you have total in your Atlas?

Successes in Sample: the number of those specific cards that you want to draw. In the case of Colicky Dragonettes on turn 3, this number would be 2 since we want 2 or more fire threshold sites by turn 3.

In the example above we can see that with a "Population Size" of 30, a "Sample Size" of 8 (full mulligan plus 2 draw steps), 15 "Successes in Population" (half of our Atlas is fire sites), and 2 "Successes in Sample," we have a 98.2% chance to see 2 or more fire sites by turn 3.

A 98% chance of success if fantastic, but in Sorcery, there is always more to consider. The first issue is whether it’s prudent for your deck to spend valuable mulligan slots on just your Atlas and draw from the Atlas for the first several turns rather than having access to more spells. Another issue is that this is an imperfect formula since there may be cards in your Spellbook like Ruby Core that provide threshold.

How do we account for these issues? The fact is, there isn’t a one-size-fits-all solution. That doesn’t mean that we shouldn’t try to use all good tools to our advantage, though! My suggestion is to think really hard about how your deck will utilize mulligans and how many sites you’ll want to draw during your first several draw steps; this will allow you to come up with Sample Size numbers that match your deck’s goals.

How do you calculate threshold requirements for multi-element spells? The math for this becomes extremely complex if we want something super accurate; instead, let's settle for something more simple, albeit less exact.

Let's say we want to be able to reliably cast Yellow Knight on turn 3 in a mid-range deck. If we look at the table for turn 3 and a single threshold requirement, we can see that we need at least 8 sources that provide that threshold. Yellow Knight needs both earth and fire threshold, though.

If we have 8 sources each of earth and fire, we can take our probability of drawing either and multiply them together to find the odds of drawing both.

            0.916 x 0.916 = 0.839

Using this formula, we know we have a roughly 84% chance to hit both earth and fire by turn 3. The problem with this solution is that the variables are not independent of each other. Following the example set by Dr. Frank Karsten in his definitive article about colored mana sources for Magic: The Gathering decks, I'll suggest something similar to his solution to this problem: split the threshold requirements, then increase each threshold source in your deck by 1.

For our example with Yellow Knight, we would increase our elemental source count (or Successes in Population) to 9 each for earth and fire if we wanted to reliably cast it on the third turn.

Unlike MTG where lands are typically adding mana and color (threshold) in increments of one each, some sites in Sorcery add mana and multiple threshold. This is incredibly helpful when trying to meet multiple threshold requirements and means that multi-element sites lend a higher weight towards your goals when compared to the lands of MTG. Because of how multi-element sites increase our chances of reliably casting multi-element spells on curve, you can adjust your calculations a bit depending on the number of these sites you have in your atlas.

Let's keep using Yellow Knight for our example. Say you have 3 copies of Steppe in your deck along with Colour out of Space and Glastonbury Tor for a total of 5 sites that produce both earth and fire threshold. If you see up to 7 total sites by turn 3, you have a 76.4% chance to hit one of these multi-element sites, which means that in many of your games you will be able to cast Yellow Knight without having to dig for additional sites after the first multi-element site is drawn.

In fact, in just under half of your games (using the mid-range model) you will see one of these multi-sites within your opening hand before mulligans. For Sorcery, I posit that the more multi-element sites you include that satisfy all threshold requirements for a given card, the less likely you are to need the 1 additional source of each element in question.

Let's look at another more complex example of multi-element threshold requirements to illustrate the concept further, and let's use an aggro deck for this one. Let's say we have 8 individual 3-mana, double-water cards in our Spellbook that we really want to be able to cast by turn 3. On top of that, we are running Pact with the Devil and two copies of Panorama Maniticore which we want to be able to cast sometime around turn 5-6 if we draw them (roughly 40% chance to see one of these spells in the first 5-6 turns). Additionally, we have Lady of the Lake and Cerberus in Chains in the 3-mana slot and they have triple-water and double-fire requirements respectively. We're running 3 Oasis, Colour out of Space, and Joyous Garde in our Atlas. What does the rest of our Atlas need to look like in order to cast these spells reliably?

Taking a look at the table, we see that we need at least 15 water sources for the 3-mana spells with double-water. We can see in the 5-mana column that we would need at least 13 sources of fire to cast something like Pact with the Devil or Panorama Manticore on turn 5 (or 12 sources to cast on turn 6). In the case of Lady of the Lake, we would need a whopping 20 water sources to cast her on turn 3. Since there is only one copy of her in the deck, we are only going to draw her by turn 3 about 12% of the time and similarly for Cerberus in Chains, we will only need the double fire by turn 3 (15 recommended earth sources) 12% of the time. If these are the only two cards that are putting significant stress on our numbers like this, I would meet in the middle for sources (18 water and 14 fire). Since we're already including the 3 Oasis, Colour, and Garde, we need 13 additional water sources and 9 additional fire sources.

That's a good recommendation for if we want to be able to cast at least one of these spells on curve. But what if we want to have a good shot of being able to cast our 3-mana, double-water spells on turn three AND then go on to reliably cast Pact with the Devil or Panorama Manticore around turn 5? First we take the % chance to hit our 3-drops (aiming for 91.6% with 15 water sources). Then our % to hit the double fire by turn 5 (91.1% at 13 fire sources). Next we multiply these together to see our odds of getting both in the same game and come out with 83.4%. If we REALLY want to make sure we can cast everything on the turns we want we should add an additional source each of water and fire.

Remember when I previously suggested that for multi-element requirements you can ignore the 1 source bump for every 5 dual/multi sites you have with the relevant elements? That works well for single-element pips. When you start working with more than one pip of the same element on spells, it really starts to strain your chances of success. I suggest adding the additional water and fire source in this example even if you have 5 or more multi-sites like Oasis, etc. in the Atlas

If you're working with a deck that is interested in drawing a significantly different ratio of sites to spells, I recommend using the hypergeometric calculator to determine the number of required sources.

In general, decks that will look at fewer sites over the course of the critical turns will need to run significantly more elemental sources compared to decks that are willing and able to have more looks at the Atlas.

How do abilities like those found on Seer or Land Surveyor affect the target numbers?

Both of these abilities count as a "look" at the Atlas for the purpose of adjusting the Sample Size of your hypergeometric input. Seer's ability is always an option for you each turn, so if you are planning to use the ability to scry your Atlas 50% of the time over the course of 4 turns, that would count as 2 looks.

Land Surveyor, on the other hand, is not a guarantee since you won't draw it every game. If you have 4 copies in a 50-card Spellbook, you will see it about 50% of the time after looking at 8 spells.

If your deck contains 1 such card. It is negligible for threshold calculations.

2 such cards counts as 20% of a "look".

3 such cards counts as 33% of a look.

4-7 such cards counts as 50% of a look.

8-10 such cards count as 75% of a look.

These numbers may seem low but they are weighted in a way to adjust for how relevant the threshold is likely to be when you are likely to draw one of these cards. This is erring on the side of caution since most players would rather have the deck be more consistent than necessary rather than not enough. One thing of note is that if you are using Land Surveyor in a Deathspeaker deck, the odds of you drawing a Surveyor don't go up but each one drawn will count as 2 potential looks since you can recast it from the Cemetery.

How do we consider cards like the Cores or Blacksmith Family which are elemental sources but are in our Spellbook rather than the Atlas? If you have a single Core in your deck, you're only going to see it 16% of the time after 8 looks at the Spellbook. This isn't a lot, but it is something. My suggestion is to count a single Core as follows:

1 such card - 15% of a source on average

2 - 20% average

3 - 33% average

4/5 - 50% average

6/7/8 - 75% average

The more spells you see during the time-frame you want to count these as a source, the higher you can rate them. If you have enough fire sources in your deck to accommodate everything except Meteor Shower at 9 mana, you can consider Ruby Core as a higher % of a source for calculating Meteor Shower's requirements.

If you need to add fire sources for early plays, Ruby Core becomes much less statistically relevant.

While this isn't enough to change how many whole sources you will put into your Atlas on it's own, it can be combined with other partial sources to make up a full source, or can be used as a tie-breaker if you're torn between wanting a slight boost on sources, but not willing to commit to a full additional space in your Atlas.

Cards like Mirror Realm and Valley of Delight count as 1 full elemental source in many cases but have some caveats.

Mirror Realm can never be your first threshold source for a given element unless you're playing it nearby an opponent's site that matches the element you need. That isn't going to happen consistently enough to count it as any % of a source.

Once you have at least one element established though, Mirror Realm is good at doubling up for costs requiring 2-4 threshold. If you have cards with more than one element pip in your Spellbook Mirror Realm is worth the following % of a source:

• 1 element in your deck has spells with more than one threshold pip of the same type - Mirror Realm counts as a full source for the sake of casting those spells.

• 2 elements in your deck have spells with more than one threshold pip of the same type - Mirror Realm counts as 50% of a source towards both elements.

• 3 or 4 elements in your deck have spells with more than one threshold pip of the same type - Mirror Realm counts as 25% of a source towards all elements.

For example, if you are playing a mid-range water deck and you have multiple 5-mana water spells with two water threshold pips, count Mirror Realm as a full water source for the relevant section of the table reference. 5-mana, double-threshold spells require 12 water sources. You can have 11 true water sources and count Mirror Realm as the 12th.

If instead you have a mid-range water/fire deck that runs 5-mana water spells with 2 water pips and fire spells costing 5 mana and 2 fire pips, count Mirror Realm as half of a source for each of water and fire. This would mean you could run 11 water sources 11 fire sources and Mirror Realm to bring you to a total of 11.5 sources each. You would need to find a way to fit in an additional half a source for each element to meet the table's recommended numbers.

Valley of Delight is very flexible, but once you commit to a certain element, it can no longer be counted as any other elemental source. The more diverse your deck's threshold requirements are, the lower you should count each Valley of Delight. For two-color decks with mostly single element pips, I would count each Valley as a whole source. For decks that run 2 elements but have many cards with more than one element pip, I would count Valley as half of a source. For decks that run 3-4 elements and have mostly single element pips, I would count them as a half of a source for each element. For decks with 3-4 elements and spells with more than one pip, I would count them as 25% of a source.

Floodplain is able to provide double water assuming that you have a non-water site to flood with it. This means that while it doesn't count differently for single threshold requirement numbers, it does give you a little boost for requirements over 1 threshold. For calculations where you are estimating required sources for 2 or more threshold pips on a spell, I suggest counting Floodplain as 1.33 sources on average. This one is a bit tricky since if you draw it and can utilize the flood, it's a guaranteed 2 sources but there will be plenty of times where you won't draw it and there will also be plenty of decks that are heavy enough into water that you won't always have a nearby non-water site to flood. Depending on how many water sites you are running in your Atlas, you can adjust up or down from 1.33. If you are running very few or no additional water sites besides Floodplain, you can count it as the full 2 water sources for calculating needs on water spells with 2 or more threshold pips.

Annual Fair counts as a full source for all elements but cannot be counted towards requirements with respect to strict curve adherence. If you want to cast a 4-mana, double-fire spell on turn 4, this can't be counted on as a fire source in your calculations because it requires a mana to produce the threshold. If you are okay with casting your 4-drop for 5 mana, you can count Annual Fair as a full source.

In Sorcery, we have a sweet Avatar that comes pre-loaded with elemental affinity!

Ostensibly, we could just say that Elementalist adds a full source of each element to your affinity and leave it at that. In reality, though, always having access to a single threshold for each element has powerful implications for source numbers in your deck.

Say, for example, you have a deck that has 4 copies of Cloud Spirit, a Grim Reaper, and a Highland Princess. It's going to be pretty important for you to hit double-air by turn 2. If we refer to the chart for both aggro and mid-range decks, we see that the recommended number of air sources is at least 15. With Elementalist always providing an air source, you actually only need to hit one air source from your deck by turn 2 to unlock these spells. Looking back at the tables, we can see only 9 air sources are needed to hit a single threshold by the second turn. That is a huge drop in the required sources!

If your deck has spells/abilities that require more than a single elemental pip of one type, you can still use the tables by simply moving a column to the left for each recommendation within the same turn/mana range. If your deck has only spells that require a single elemental pip of any given type, you're free! You can run any sites you want without worrying about threshold at all!

Now that we have covered all of this information, let's take a look at a full Spellbook and go over how it all ties together with building the Atlas.

This is a fast deck that looks to top out around 6 sites total, so I will refer to the aggro table. With 9 total double-fire 3-drops, the deck really wants to have 2 fire threshold on the third turn. I also need to have double-earth starting around the time I have 5 sites in play. Jihad isn't essential to play on turn 2, so we will count it as 4+ cost with two fire pips. There are 2 triple-fire cards in the deck and I would be happy to cast either of them starting on turn 4. If I can get Morgana le Fay down on the third turn that's great, but she's still going to be good on later turns and there are plenty of other 3-drops in the deck so it's unlikely that I won't have other options on turn 3. In fact, if I want to maximize the potential value of Morgana, I should wait cast her once I have 5 or 6 mana so that I have a good chance of immediately casting one of the spells that she draws before the opponent has a chance to kill her.

Being a Sorcerer deck, I will typically play one site per turn for the first 4 turns of the game but sometimes the 5th and 6th site will not be played as soon as possible. Depending on the texture of my hand and how it lines up with what the opponent is doing, it is not uncommon to activate Sorcerer to draw a second spell on the 5th and/or 6th turn rather than playing a site. In some rare cases—especially with a Core or Imperial Road in hand and/or if I desperately need to play a 2 or 3-mana spell and don't have one yet—I may draw an extra spell as early as the third turn.

With this in mind, we will start with the table's suggestions and then see if there is room in the Atlas to bump the threshold sources a little bit for extra consistency. In a game-plan where it is essential to spit out as many spells as possible each turn, stumbling on threshold even for a turn can result in losing the game.

Looking at the table, I see that I will want at least 15 fire sources to hit double-fire by turn 3. To cast Boudicca on turn 4, I would need 20 sources. Considering that she is a Unique and will also be a powerful play on later turns, I may need to accept that casting her on the 4th turn is unlikely. Instead, I'll try to be able to cast Boudicca and Morgana around turn 5 with 18 fire sources.

For double-earth by turn 5, I need at least 13 sources and since there is a small chance that I will draw (and want to play) Gigantism or Sir Priamus on turn 4, I should consider running closer to 15, perhaps splitting the difference and running 14. Shield Maidens being a 3-mana, double-earth spell makes me feel like I should probably be looking at 15 earth sources a little more closely.

In all, I'd likely want to include 18 fire sources and 14-15 earth sources. 3 Steppes are an automatic inclusion; Colour out of Space, Glastonbury Tor, and Caerleon-Upon-Usk are all worth considering. My inclination is to choose Colour out of Space for sure as it is the more versatile option in terms of placement.

This deck typically places its first 4 sites in a forward-facing "T" pattern, so unless one of the Castles is in my opening hand to play first, I'll need to weigh the strategic implications of placing it as a second site on my back row early or wait until after turn 4 to play it.

Let's start with 3 Steppe, Colour, and Glastonbury Tor. I now need 13 more fire sources and 9-10 more earth sources.

Since this is a 2-element deck with multiple element pips on lots of cards, I can include Mirror Realm at about half of a source for each of fire and earth. Similarly I can include 2 copies of Valley of Delight and count them as one full source for the pair.

I also have Ruby Core and Onyx Core which each count for around 20% of a source. Neither of these are really counting for much, but maybe the Onyx Core can be the deciding factor for me on whether to include the extra earth source. Let's say that it does and that I now need 11.5 more fire sources and 8.5 more earth sources.

Smokestacks of Gnaak, River of Flame, and 3 Poisoned Well all seem like fantastic inclusions. That leaves me still needing 6.5 more fire sources; the pre-Arthurian Legends version of this deck ran 5 Deserts, so I'll add 7 Deserts now that I am moving up to a 30-card Atlas. This puts me half a fire source over the needed 18, which is good since it will help with Boudicca and Morgana.

Pillar of Zeiros and 2 Rift Valley are easy picks for earth which brings me down to 5.5 more earth needed. Holy Ground and Gnome Hollows can be good in the aggro mirrors; Gnome Hollows can get in the way of my own minions which might be a problem. Since I'm running Torshammar Trinket and Boudicca, perhaps I should consider some number of Villages.

I could also run Quagmire to pin down opposing minions until I get to untap and deal with them on my own terms.

Impenetrable Copse could be nice if I expect to play against Archimago, Battlemage (Grapple Shot), Flamecaller, or decks with Pudge Butcher. It would potentially get in the way of being able to finish the game with my own Firebolts, but since I could always just step forward into the Copse to shoot the Firebolts, maybe this isn't a real concern. It would be a bummer though if I needed to spend the Firebolts earlier to kill an enemy minion that was being protected by the Copse...

When there are a lot of good options like this I tend to choose a variety of them since in best-of-1 tournaments, I can mulligan the unnecessary ones that I draw in the opener and know that I never have to see them again.

Lets go with a mix and say 1 Holy Ground, 1 Village, 1 Gnome Hollows, and 1 Impenetrable Copse. This brings me down to needing 1.5 more earth sources and I have a total of 3 slots left in the Atlas to get to 30 cards.

I don't know if I'm quite ready to abandon copies of Imperial Road yet. In addition to being part of the deck's namesake, it has proven time and again to be incredibly powerful in Sorcerer aggro even though it can sometimes get in the way of optimal threshold (like it is threatening to do now.) Additionally, I really want to try out Fields of Camlann which is fire and I don't like the idea of cutting any of the fire sites that are already chosen.

Let's take a closer look at my earth threshold requirements. I have 1 Shield Maidens, 1 Gigantism, 1 Sir Priamus, and 2 Earthquakes. Shield Maidens is only about a 12% chance to draw by turn 3, so perhaps I should just ignore it as a potential 3-drop. It's also not particularly good on turn 3 in many situations, so let's slide it over to the 5-drop, double-earth category which suggests 13 total earth sources.

Gigantism is good on turn 4, especially if I played Quarrelsome Kobolds on turn 3. With only one in the deck, the odds of hitting it by turn 3 are very low, so maybe I should be calculating this as a later play, like Shield Maidens; maybe turn 5 at the earliest (13 suggested sources) with more likelihood of turns 6-7 (12 suggested sources.) Sir Priamus is another 4-drop with 2 earth pips, so he and Gigantism means that there is a 22.8% chance to hit one or both by turn 4. This is still low probability, so I think it is safe to keep them both in the turn 5-7 slots.

Earthquake is a card that I will sometimes cast on turn 5, but in most cases at this stage of the game I am able to keep my opponent's minions at bay using a combination of small damage sources like Vile Imp and Deserts. After turn 5, though, some decks can begin presenting two threats per turn or much larger threats that survive the smaller pings. If I want to cast Earthquake on turn 5, I need 13 earth sources. If I can hold out until turn 6, that brings it down to 12 required sources. With 5 total earth cards that have double threshold pips in the deck, there is a 55% chance to hit one of them by turn 5. This means that I definitely don't want to go below 13 sources for earth but that 13 is sufficient and I don't feel the need to go all the way up to 14.

There we go! I talked my way out of having to get rid of the Imperial Roads and still get to include Fields of Camlann. After more testing it's very possible that I will still end up wanting to add an additional earth source or simply opt to cut the Shield Maidens. Depending on how Glastonbury Tor feels, that may also need to be cut, in which case I would definitely need to either replace it with an earth source or adjust the Spellbook a bit so that things are more consistently castable with fewer sources.

Here is the final decklist from the example. If you want to check for the most recent updates to the deck, you can do so here.