Inner Workings of Arena

Common misconceptions and questioned rules

Here are clarifications on some of the most common points where new players get confused about Arena and Magic. This section will repeat many of the things said in other parts of the guide.

Arena-specific:

• 60 cards is the minimum in Standard/Historic Constructed, not a requirement. It’s 40 in Limited. Playing more makes your deck less consistent.
• You can craft cards that you don’t have a copy of. Use the Craft button or “Not Collected” filter in the Collection.
• Receiving a 5th copy of a card never produces a wildcard for you. That would break the economy. Wildcards in packs are just randomly rolled.
• 750-Gold daily quests do not award 750 Mastery XP. It’s 500 XP in either case. You also don’t get three quests every day, just one.
• Generally, to respond to your own spells and abilities, you have to be in Full Control or set a stop on the current turn phase or step.

Rules-related:

• A blocked creature remains blocked even if no blocker is left on the field (imagine it as a distraction, not an obstacle). Only a creature with Trample will deal damage to the original target if blockers are removed (e.g. killed in the first strike step).
• Fight isn’t attack or combat, it’s an ability keyword like many others. It just means creatures dealing non-combat damage to each other.
• Indestructible prevents being destroyed by damage or effects that destroy. Exiling or lowering toughness to 0 deals with indestructible. Damage doesn’t lower toughness (another slap to Arena for displaying that in an unclear way).
• Hexproof and Protection deny targeting. Neither entails immunity. Things that don’t target can affect Hexproof or Protected permanents.
• A part of Protection is preventing damage. That part doesn’t work is something says damage cannot be prevented.
• Mana abilities are the ones that produce mana (though not all of them).

Arena quirks and deviations

Overall, Arena strives to be “true” Magic, but there are some limitations and some improvements, most importantly — attempts at reducing variance and the number of non-games in best-of-one.

• Arena uses “hand smoothing” in best-of-one games: the game shuffles the deck thrice and draws three hands and leans to choose the one with the land/spell ratio closest to the deck’s ratio (no color checks). Wizards tested other land consistency tech but confirmed to use only this one.
• Deck size in Arena is limited by 250 cards.
• Arena limits the number of tokens on each side of the field by 250.
• It’s enough to have 4 copies of a card to put any number you want in a deck (for cards that break the 4-of rule).
• You don’t put basic lands into your Limited sideboard, but you can fetch them during a game with “wish“-like cards.
• Arena has its base set (also possibly this and this) that is legal for “Arena Standard“ best-of-one games even if not legal in “real“ Standard. It’s also legal in Historic modes.
• Arena doesn’t support ordering simultaneously received timestamps (rule 613.7k).
• Effects that put cards “on the bottom of your library in any order” like Anticipate only let you choose the order if you have fewer than 10 cards left in your library or activate Full Control, otherwise randomize (and hide) the order.
• (Obsolete) Some non-Standard cards were legal in Arena Brawl, e.g. Rhys the Redeemed. They “rotated“ out with Zendikar Rising release.

Reporting bugs

Word of warning: Arena is just a piece of software and surely can have bugs. However developers created an engine that interprets cards based on the Magic ruleset, thus avoiding the need to explicitly code all but the most non-trivial abilities and interactions. The engine is highly trustworthy, and if something worked not as you expected, most likely you misunderstood or missed something. Over 15 months since Arena went into the open beta, “bug” reports on Reddit that turned out to be rule-related would barely add up to a dozen. To give you an idea of the complexity that might confuse the engine, here’s an example: initially, Chandra, Acolyte of Flame‘s -2 ability didn’t allow casting Light Up the Stage for the Spectacle cost.

If you encountered an unexpected card interaction, it most likely isn’t a bug, even if it’s not listed above. If you’re sure it is, or if you observed a graphical glitch, the game halted or crashed, here’s where to report it: https://feedback.wizards.com/forums/918667-mtg-arena-bugs. If such an issue caused you to lose in a paid-entry event, feel free to also open a ticket and ask for a refund.

Shuffler and matchmaking

What do we know about shuffling in Arena

Not much. Basically, this:

To shuffle decks in MTG Arena we use Fisher-Yates, pulling numbers from a Merseene Twister (MT199937), which is seeded with 256 cryptographically secure randomized bits. We use the same approach for coin tosses, only we’re looking for a 1 or a 2 rather than a whole deck of cards.

In laymen words this means that shuffling is done close enough to true random for any practical purposes, given it is coded correctly. There were doubts about it being coded correctly: results deviated from expected enough to indicate presence of a common mistake, but with War of the Spark release in April 2019 these deviations disappeared and are not observed anymore.

There are no reasons to assume the game intervenes into shuffling itself, but as indicated above, not all shuffle results are used to avoid the number of “non-games” in best-of-one modes.

What do we know about matchmaking in Arena

Somewhat more than about shuffling, but the official information we have is very old, and for many modes that appeared later we are limited to deduction. The baseline information was given in this breakdown, give it a read.

Almost no changes were officially made, and most additions were had to be gathered bit by bit from Reddit and Discord comments made by the developers, including the very fact that no significant changes happened. The most impactful change happened almost immediately, with Limited MMR removed from Ranked drafts and win/loss record being made the primary matching parameter. Aside from that:

• Matchmaking in Brawl is adjusted based on commanders to match the strongest ones primarily with each other.
• Matchmaking in Brawl, like in the Play queue, considers deck weight, seemingly as a secondary parameter. This corresponds with player findings.
• No other mode and specifically no competitive mode considers deck weight, and once again.
• Little is known about deck weight. What developers said indicate that how frequently a card is crafted plays a role and how good/efficient it is plays a role (presumably in terms of being used in top decks). Deck weight is then calculated by adding up individual card weights.
• Matching with specific deck types in mind is not done in any mode.
• Matchmaking in all draft modes is primarily based on the win/loss record in the running event. In Quick and Premier drafts (that are ranked) rank is the secondary matching parameter. It means rank discrepancies are more likely to happen in drafts, but when possible the game still matches you within rank, creating a soft bias in favor of lower-ranked players (assuming matches versus those are generally easier).
• MMR (matchmaking ranking) is known to be “modified Glicko 2”.
• MMR in ranked is separate from rank and presumably untied from it. This creates a situation where you your MMR can vary drastically from players at the same rank (as you can’t drop to the previous one during the season), and then they encounter a very different opponent pool. Here‘s a long and detailed article on results of player-ran investigation of rank/MMR dual system in Arena.
• Your position in top #1500 Mythic directly corresponds to your MMR and thus may be very volatile. Below #1500 your rank is represented by a percentage of your MMR to the MMR of the player at the position #1500.