Deck statistics are most useful when you know which numbers are direct measurements and which are estimates. Manamancer calculates from the card records currently loaded for the deck. A change to a card, quantity, printing, commander, or partner can therefore change several panels at once.
Last reviewed: August 2, 2026.
Start with the counting boundary
The core totals count the main deck plus a commander and partner, when present. Sideboard and maybeboard cards are excluded. A companion is kept outside the deck proper and is also excluded from those totals. This matches the way the builder separates cards that belong to the playable deck from cards being considered or held in another zone.
Quantities matter. Four copies of a spell add four cards to the total, four entries to its curve bucket, and four times its colored mana symbols. A commander or partner contributes one card each.
Before diagnosing any surprising statistic, check that cards are in the intended zone and have the intended quantities. A perfect calculation over the wrong zone is still the wrong answer to your deck-building question.
Cards, lands, and average mana value
The headline card count is divided into lands and nonlands. Average mana value is calculated only across nonland cards:
sum of each nonland card's mana value × quantity ÷ number of nonland cards
Leaving lands out makes the average more useful for describing the spells you expect to cast. It also means adding a basic land will change the land and total counts without directly lowering the displayed average mana value.
For modal double-faced cards, Manamancer follows the front-face role used by the deck list. A Creature // Land card is treated as a creature rather than a land for this calculation. That is a deliberate simplification: the land face may still matter in games, but the same card should not jump between the list's land count and the statistics panel.
Reading the mana curve
Nonland cards are placed into buckets labeled 0 through 6 and 7+. A card's numeric mana value is rounded down to choose a bucket, and every copy contributes to the height. Lands do not enter the curve.
The graph answers “how much of the deck is concentrated at each printed mana value?” It does not answer all of these related questions:
- whether an alternate cost is normally paid;
- whether a spell is usually cast for X;
- whether cost reduction is reliable;
- whether a card is meant to be cycled, channeled, or played on its other face; or
- how many spells can actually be cast on curve with the current mana base.
Use the curve to find areas worth inspecting. A large 4-mana column might be intentional in a ramp deck and dangerous in a low-land tempo deck. Context supplies the conclusion.
Colored pips versus colored sources
The colored-symbol display parses W, U, B, R, and G symbols from nonland mana costs, multiplied by quantity. It describes demand: which colors appear in the spells you chose and how heavily. Hybrid or multi-pip costs can make that demand more complicated than a simple deck identity.
A pip is not a mana source. Ten blue symbols do not mean ten Islands are enough, and a land that produces several colors should not be counted as several physical lands. Use the source and curve-out analysis alongside the pip display.
The curve-out forecast estimates land-drop and colored-source availability by turn. Its source classifier accounts for common duals, fetch lands, and mana rocks, while tapped and conditionally tapped lands affect the untapped analysis. Those classifications are structured heuristics; unusual replacement effects, conditional mana, rituals, treasures created later, and table-specific play patterns can change real games.
Card types and custom categories
The type breakdown comes from each row's explicit deck category, plus the command zone. Categories such as Creatures, Instants, Sorceries, Artifacts, Enchantments, Planeswalkers, Battles, and Lands are organizational choices in the builder. If you manually place a card in a different category, the statistics follow that category rather than silently recategorizing it from the type line.
That behavior makes custom organization predictable, but it is another reason to check the deck list when a type count looks surprising.
Ramp, draw, removal, and wipe counts
The overview looks for common Oracle-text patterns to estimate several deck-building jobs:
- Ramp includes common mana abilities, land-to-battlefield searches, Treasure creation, and some effects that put lands into play. Lands themselves are not called ramp.
- Draw looks for ordinary “draw a card” and multi-card draw wording.
- Wipes look for common all-creatures or all-permanents destruction and exile wording, mass negative power/toughness, some damage-to-each wording, and broad sacrifice effects.
- Removal looks for common target destruction, exile, counters, damage, bounce, fight, and targeted sacrifice wording. Cards recognized as wipes are not also counted as spot removal by this summary.
These are deliberately described as heuristics. They can miss a card whose role depends on a keyword, a delayed trigger, a choice spread across faces, a type-changing interaction, or deck synergy. They can also count a modal card even when you expect to use its other mode most of the time. A tutor that finds a removal spell is not itself necessarily counted as removal.
Treat a low count as a prompt to inspect the list. Do not add a card only to make a tile reach a target, and do not assume a healthy-looking count means the deck has efficient or appropriately timed interaction.
Price and market totals
The deck value uses the selected price preference and the known printing allocations for each row. The commander and partner are included. When a chosen seller does not have a valid quote, Manamancer can show a fallback from another supported source; the interface marks that condition. A tilde indicates that the total has not yet been refreshed with live detail.
Prices are daily market estimates. Finish, condition, region, inventory, currency conversion, shipping, tax, and the moment a store updates its listing can all change what someone actually pays. The seller remains the final source for a transaction.
Probability Lab
The Probability Lab uses a hypergeometric calculation: drawing from a finite library without replacement. You choose how many matching copies are in the deck, how many you want to see, and how many cards you will have seen. The result is a mathematical probability under those inputs.
Presets cover questions such as three land drops by turn three, a turn-one play in the opener, fast mana in the opener, or a tutor by turn three. The exact-copy bars show how the probability is distributed rather than only reporting “at least one.”
The model assumes a randomized library and the stated number of cards seen. It does not simulate London mulligan decisions, scrying, tutors already in hand, card-selection sequencing, opponent disruption, or cards returned to the library. Adjust the inputs when your question differs from a preset.
Commander and Canadian Highlander estimates
Commander decks can show an estimated bracket based on detectable signals such as Game Changers, mass land denial, extra turns, two-card combos, and tutors. Some criteria depend on external combo data, and unresolved cards can make an estimate change. Rule Zero intent is not fully machine-detectable, and the strongest competitive bracket is a player-declared, metagame judgment. Use the estimate to prepare a conversation, not replace it.
Canadian Highlander decks show cards on the current points list and their total against the 10-point cap. Because the governing list can change, a refreshed rules snapshot and the official format source should settle event questions.
A practical review order
- Confirm format, zones, quantities, commander, partner, and companion.
- Check total cards and lands before interpreting averages.
- Read the curve together with colored demand and actual sources.
- Open the cards behind any role count that looks wrong.
- Use probability inputs that match the decision you are making.
- Treat prices, bracket signals, and role classifications as estimates with visible assumptions.
- Finish with the format-legality result, then use the current official rules for any event-level decision.
If a statistic appears inconsistent with the behavior described here, send the deck URL, the affected panel, and the card or quantity involved through the contact page.
