Tool Guide

Recent Decks

Every deck you open is automatically saved to your browser's local storage. The next time you visit the home page, your recent decks appear below the load form — click any entry to reopen it instantly without re-pasting the URL.

Farseek also saves your group layout. If you created custom groups, moved cards around, or reorganised the view, that state is restored exactly as you left it the next time you open the same deck.

• Up to 10 decks are kept in history, most recent first.
• A custom groups badge appears on any deck where you modified the default layout.
• Hover over an entry and click × to remove it from history.
• Clear all removes the full history at once.

Loading a Deck

You can load a deck in two ways:

• Paste a URL from Moxfield or Archidekt. The deck is fetched automatically.
• Paste a decklist in plain text — one card per line with quantity (e.g. 1 Sol Ring). Add Commander: Card Name to mark the commander. Pasted decklists are not saved to Recent Decks.

Deck update detection

When you reopen a deck that was previously saved, Farseek compares the fresh data from Archidekt or Moxfield against your saved version. If the deck list changed (cards added, removed, or quantity changed), an amber banner appears at the top offering to update.

Clicking Update deck applies the changes while preserving your custom groups, focused cards, and discounts as much as possible — cards that still exist in the new list keep their configuration; removed cards are silently dropped. Clicking Keep saved version dismisses the banner and continues with your saved state.

On Archidekt, any category marked as not included in deck is treated as a maybeboard — excluded from all counts and hidden by default.

Deck View & Card Groups

Cards are displayed in a masonry grid, automatically sorted into type groups: Commander, Creature, Planeswalker, Instant, Sorcery, Artifact, Enchantment, Land, and Other. Land always takes priority — a card like Darksteel Citadel (Artifact Land) or Dryad Arbor (Land Creature) goes into the Land group, not Artifact or Creature. For Modal Double-Faced Cards (MDFCs), the front face determines the group — so Agadeem's Awakening (Sorcery // Land) goes into Sorcery, not Land.

Selecting cards

Click any card row to select it — a subtle border appears on hover to indicate rows are clickable. Click multiple rows to build a selection; a toolbar appears at the top with available actions. Click a group header to select the whole group at once. Press Esc at any time to deselect everything and close any open overlay or dropdown.

Toolbar actions

Sideboard & maybeboard

These are hidden by default. Use the Show sideboard / maybeboard button (top right of the deck view) to toggle them. They are never counted in any totals or analysis.

Deck-level Tools

The permanent toolbar just below the navbar gives access to whole-deck analysis tools. Additional tools become available in the group toolbar when you select the Land group.

Draw Simulation

Available for any group (except the Land group, which has its own tool). Select a group and click Draw Simulation in the toolbar.

The chart answers: "If I draw N cards, what's the chance I'll have seen at least X copies from this group?" The X axis shows hand sizes from 7 (opening hand) to 12 (five turns in), and the Y axis shows the probability.

Use the Draw at least N input to change the threshold. Setting it to 1 asks "will I see at least one?" — useful for combo pieces. Setting it higher is useful for groups like lands where you want to reliably hit a certain count.

Multiple Draw Simulation panels can be open at once — open one for your creatures, another for your win conditions, and compare side by side.

Land Draw Simulation

Land-specific version of the Draw Simulation. Select the Land group and click Land Draw Simulation.

The chart shows two lines:

• Land (blue): probability based on lands in the Land group only.
• Land + MDFC (amber): same probability but counting Modal Double-Faced Cards (MDFCs) that can be played as lands.

Simulate more land

Click Simulate more land to reveal an input. Enter a positive number to simulate more lands, or negative for fewer. Dashed lines show the probability difference — deck size stays fixed, modelling a straight swap of land for non-land or vice versa.

Tap Analysis

Select the Land group and click Tap Analyze. This tool answers: how much does playing tap lands slow you down?

Land categories

The chart

Expected untapped mana by turn (T1–T5) for three scenarios:

• Current deck — your actual land base
• −1 tap land — replacing one always-tapped land with an untapped one
• 0 tap lands — theoretical ceiling: all always-tapped lands replaced

The footer shows mana lost to tap lands across T1–T4. A loss under 0.5 mana is negligible; over 2 mana suggests your tap lands are meaningfully hurting your early turns.

Missed Land Drops

Select the Land group and click Missed Land Drops. This tool answers a fundamental question: how often do you fail to play a land on a given turn, and when does it first happen? It runs 10,000 simulated games and tracks only land draws — rocks, dorks, and ramp are intentionally excluded so you see the raw land-base consistency.

Options

Reading the results

• Never miss (T1–TN)% — the headline number. The percentage of games where you hit every land drop through the tracked window. Comes with a tier badge: Very Consistent (≥90%), Consistent (≥75%), Shaky (≥55%), Screwed (<55%).
• Miss at least once — complement of the above.
• Avg first miss (when it happens) — among games that do miss, the average turn it first occurs. A late average (turn 6–7) is much less damaging than a miss on turn 2.
• First missed land drop by turn — bar chart showing which turn is most often the first miss. Red bars are early turns (high impact), fading to yellow for later ones.
• Cumulative chart — P(missed at least once by turn T), the probability that you have missed at some point by each turn. This grows monotonically and reaches "Miss at least once" at the final turn.

Manabase Analysis

Click Manabase Analysis in the deck toolbar. This tool distils your entire mana base into a single number and a human-readable rating — useful for quickly comparing configurations and identifying which colour is your bottleneck.

The score — average casting delay

The score is the flat-weighted average of the casting delay across every colored spell in your deck, measured in turns. A delay of +0.0 means you cast every spell exactly on its CMC turn, every game. A delay of +0.5 means on average you wait half a turn extra — missing curve on roughly one in three spells.

Every colored spell counts equally — a demanding {R}{R}{R} Kiki-Jiki matters as much as a simple {G} creature. This "flat weighting" ensures a thin color in your deck can't be hidden by having many easy-to-cast spells of another color.

Simulation options

All four options below are shared with Cast Probabilities — changing any one in either tool updates both simultaneously.

Per-color breakdown

Below the score, a row of coloured mana symbols shows the probability for each required colour — for example ⚫ B: 76% 🔴 R: 88% 🌿 G: 93%. This is the analytically computed probability of drawing enough sources of each colour in time for the spells that require it, averaged across all spells of that colour.

• 90%+ — that colour is well-supported.
• 80–89% — solid, with occasional gaps on demanding pips.
• Below 80% — this colour is a genuine bottleneck. Add duals that produce it.

The colour with the lowest percentage is always your weakest link. In the example above, B at 76% means that in roughly one in four games you are missing black mana when you need it. This is where to focus your land-base improvements.

Focus & Discount modes

Two card-selection modes let you tailor the analysis to your deck's actual game plan. Both modes are mutually exclusive and activated by dedicated buttons. Cards keep their colour when both panels are open — sky blue for focused, rose for discounted, purple for both.

Focus important cards

Click Focus important cards to enter focus mode. Clicking any card in the deck marks it as "important" — it glows sky blue. Press Esc or click Done to exit. Focused cards receive 5× weight in the score. Use this for key commanders, combo pieces, or payoffs you absolutely need on time. A warning banner appears if you combine focus with "Weight importance by spell cost" — both amplify high-CMC spells and can double-count their influence.

Discount cards

Click Discount cards to enter discount mode. Clicking any card adds it to the discount list with a default of half its CMC. A live panel shows each discounted card with [−] [amount] [+] steppers. Generic mana is removed first, then coloured pips. Setting a card to 0 means it's treated as free (always castable, zero delay). The same discount set is shared with Cast Probabilities.

Best-case extra lands

The slider adds N ideal lands (all five colours, always untapped) without changing deck size — modelling swapping N spells for perfect duals. If the score barely improves, your spells are simply demanding. If it jumps a tier, land count or fixing quality is the bottleneck.

How the score is calculated

For each colored spell the tool runs 400 simulated games, turn by turn:

1. Draw an opening hand via the Karsten mulligan model (keep 3–5 land hands; mull 0–2 and 6–7).
2. Each turn: draw a card, play a land with a realistic sequencing policy (untapped duals first, then basics — no oracle foreknowledge in early turns).
3. On the CMC turn: play the land that best covers the spell's missing colours.
4. Cast affordable rocks, dorks, or ramp whenever the mana is available.
5. Record the first turn the spell is castable (enough mana + right colour pips). The delay is that turn minus CMC.

The 400 delays are averaged per spell, then all spells are averaged into one deck-wide number. With "Weight importance by spell cost" on, each spell's delay is multiplied by its CMC before averaging so expensive spells count proportionally more.

Casting Probabilities

Click Cast Probabilities in the permanent deck toolbar. This tool runs a Monte Carlo simulation for every colored spell in your deck and shows a per-spell table of how consistently your mana base supports each one. Click Run Simulation to start — it takes a few seconds (1,500 games per spell). Spells with only generic costs are skipped.

Table columns

Every non-land spell with CMC > 0 appears in the table — both coloured and colourless spells. Free spells (CMC 0) are excluded since they're always castable.

Sorted by Delay descending by default (worst spells first). Click any column header to sort; click again to reverse; click a third time to reset.

Options

The first three options are shared with Manabase Analysis — changing any one in either tool updates both.

How it works

Spell-aware greedy simulation

Cast Probabilities simulates each spell as if it were the only spell you need to cast. The land selection policy is spell-aware: at every turn, the simulation picks whichever land in your hand best covers the missing colour pips for this specific spell. This models an idealised player who sequences their lands optimally for one goal, giving an upper bound on what's achievable per spell.

This is intentionally more optimistic than Manabase Analysis, which uses a more realistic land selection policy. Cast Probabilities answers "how consistently can I cast this spell if I'm focusing on it?" while Manabase Analysis answers "how does my mana base perform across the whole deck?"

Turn-by-turn play

Each game: draw opening hand (Karsten mulligan — keep 3–5 land hands, mull 0–2 and 6–7), draw one card per turn, play one land per turn. Affordable rocks and dorks are cast greedily each turn only if doing so doesn't prevent casting the target spell that turn — the simulation won't spend mana on a Signet if that mana was needed for the spell right now. The first turn the spell is castable (enough mana in play + right colour pips available) is recorded.

Tapped lands

Lands that always enter tapped (Temples, Triomes, bounce lands, etc.) are classified from oracle text and join the mana pool on the following turn — so a tapped land played on T1 is available from T2. Conditional lands (check, battle, fast, shock) are classified with realistic tap rates based on deck composition. This correctly penalises tap-heavy bases: a hand of three always-tapped lands means 0 mana on T1.

Mana rocks, dorks, and ramp

When non-land sources are enabled, the simulation greedily casts affordable sources each turn. Each type has a reliability discount to account for real game variance:

• Artifacts (rocks) — 75% reliability (~25% removal / late arrival rate).
• Creatures (dorks) — 50% reliability; they die more often. A dork cast on T1 also has summoning sickness — it can't tap for mana until T2.
• Ramp spells (Cultivate, Farseek, Nature's Lore…) — 100% reliability. The fetched land is subject to tap land rules: Rampant Growth enters tapped, Nature's Lore enters untapped.
• Colorless producers (Sol Ring, Thought Vessel) — counted for generic mana but don't satisfy coloured pip requirements.

Produced colors come from Scryfall's data. Fetch land colors are inferred from oracle text ("Search your library for a Forest or Plains" → produces G and W), so Windswept Heath and friends are correctly handled.

Color matching

Total mana and colour coverage are checked separately. For {2}{U}{U} (Mana Drain) you need 4 total mana and two distinct blue-producing sources in play — one Island cannot satisfy both {U} pips. This uses bipartite matching, so dual lands are handled correctly: Hallowed Fountain can cover one {W} pip or one {U} pip, but not both simultaneously. Hybrid pips count as a single coloured requirement. Phyrexian mana is ignored — the simulation assumes you always pay life.

Model Strengths & Limitations

Every mana analysis tool makes tradeoffs. Knowing what ours captures well — and where it falls short — helps you interpret the numbers correctly.

What we model well

• ✓
  Ramp acceleration. If you run Cultivate, Farseek, Birds of Paradise, or Sol Ring, the simulation correctly shows you have more mana available earlier. A ramp-heavy deck (Nature's Lore + Three Visits + Farseek) gets significant credit for that acceleration.
• ✓
  Tapland timing. Guildgates and Temples delay mana by exactly one turn. The simulation tracks this turn-by-turn — a hand with three tapped duals genuinely starts with 0 mana on T1, and you'll see it reflected in higher delays.
• ✓
  Multi-pip requirements. {R}{R}{R} (Kiki-Jiki) is genuinely much harder to cast consistently than {R} (Spellseeker), even at the same CMC. The simulation correctly penalises triple-pip spells more than triple-generic spells.
• ✓
  Per-color bottleneck identification. The per-color breakdown in Manabase Analysis shows you which colour is causing the most delays — so you know whether to add black sources or white sources, not just "more duals".
• ✓
  Fetch land colours. Windswept Heath, Polluted Delta, and other fetches are recognised from oracle text and their producible colours are inferred automatically (a fetch searching for "Forest or Plains" counts as both a G and W source).
• ✓
  Realistic early-game sequencing. Manabase Analysis uses a land selection policy that mirrors real play: play untapped fixing lands first, then basics — without assuming you magically know which spell you'll want to cast on turn 6.

Known limitations

• ✗
  Variable-cost spells are always evaluated at their printed cost. Blasphemous Act is always simulated as {8}{R}, even though in a typical Commander game with many creatures you'd cast it for {1}{R} or {2}{R}. X-spells (e.g. Stroke of Genius), convoke, delve, improvise, and emerge cards are similarly evaluated at their full printed cost.
  
  Workaround: use the Discount cards feature in either tool to reduce the simulated cost to what you'd actually pay. Both Manabase Analysis and Cast Probabilities (with "Apply card discounts" on) will then simulate the spell at that reduced cost.
• ✗
  Cost reduction effects are not modelled. Goblin Electromancer, Zirda the Dawnwaker, cost reducers, and similar cards do not make instants or sorceries cheaper in the simulation. Each spell is always evaluated at its full cost.
• ✗
  Alternative casting costs are ignored. Force of Will, Fierce Guardianship, Flare of Denial, and other pitch spells are evaluated as if you always hard-cast them. Similarly, cards you routinely cheat into play via Sneak Attack, reanimation, or Show and Tell get treated as needing their full mana cost — so their delay numbers may look alarming even though you'd rarely pay the full cost.
• ✗
  Each spell is simulated independently. The simulation does not model the competition between spells for the same mana. In a real game, you must choose whether to cast Cyclonic Rift or Counterspell with your blue mana this turn — the simulation assumes you always have enough for the one spell being evaluated. This makes both tools somewhat optimistic about what's achievable in a full game.
• ✗
  Phyrexian mana pips are ignored. A {1}{B/P} card (e.g. Phyrexian Obliterator has {B}{B}{B}{B}) — any Phyrexian pip is treated as if you always pay 2 life. Hybrid pips (like {B/G}) are handled as a single coloured requirement of the easier of the two colours to produce.