Loki, the Deceiver does one thing on the attack, and the whole deck is built to make it count. When Loki swings, you make a tapped, attacking, nonlegendary copy of another Villain you control, and that token gets sacrificed at the next end step. On top of that, whenever one or more of your Villains deal combat damage to a player, you draw a card. So the copy is a rental, and the deck lives or dies on whether the Villain you rent is worth having twice for one turn.
When Loki attacks, you get a free copy of one of your villains for the turn. Then it goes away. So the whole deck is about running villains that are great to have two of, even for a single turn.
What Loki actually rewards

Because the copy is gone at end of turn, a good copy target does its work the moment it arrives. A Villain with an enter trigger gives you that trigger a second time. A Villain with a static buff, like Doctor Octopus handing your other Villains +2/+2, applies the instant the token lands. A Villain with evasion actually connects, and that feeds Loki's card, since he draws when one or more of your Villains deal combat damage to a player. The one thing the token cannot do is set off an attack trigger. It is created already attacking, which means it was never declared as an attacker, so a "whenever this creature attacks" ability never fires on the copy. Copy for enter triggers, static buffs, and evasion, and leave the attack-trigger Villains to swing on their own.
There is a second lever hiding in that draw. Loki's card counts players dealt combat damage, not attackers, so one Villain and its copy both hitting the same opponent draw one card between them, while sending them at two different opponents draws two. When the board allows it, spread your Villains across the table and the same swing pays more.
A good villain to copy is one that does something the moment it shows up, or one that is hard to block so it gets through. The copy comes in already attacking, so villains that only do something "when they attack" do not work, they were never really declared as attackers. Loki draws you a card when your villains hit a player, and hitting different players draws more cards.
The second engine is connive. Several Villain-typal cards say that whenever another Villain enters, something connives, which draws a card, discards a card, and drops a +1/+1 counter if you pitched a nonland. Every copy Loki makes is a Villain entering, so a connive package turns each attack into card selection and a growing board.
Reading a Villain by its copy value
EDHREC's high-synergy list for Loki is a ranking of exactly this, the Villains the crowd copies most, and each earns its place a different way.
EDHREC shows the villains most Loki players copy: Chameleon (a copy itself), Prowler (hard to block and draws cards), Doctor Octopus (makes your team bigger and refills your hand), and Tombstone (brings villains back and makes them cheaper).
The payoffs and the mana

Above the copy targets sit the cards that make the whole engine cheaper and deeper. Baron Strucker, in seventy-nine percent of decks, drops your Villain spells by one and connives once each turn when a Villain enters, so it is a cost reducer and a card-filter in one slot. Loki's Scepter, at seventy-four percent, taps for any color and does something sneaky I will get to in a second. Villainous Hideout, at seventy-one, is a Villain-flavored land that both fixes your mana toward Villain spells and hands you another connive activation. The crowd stacks bodies and lowers the cost of running a whole Villain subtheme at the same time.
Some cards make the whole plan cheaper and smoother: Baron Strucker makes your villains cost less and draws cards, Loki's Scepter makes mana and a sneaky trick, and Villainous Hideout is a land that helps your villains.
Loki's Scepter, the sneaky line
Loki's Scepter is worth its own paragraph. When it enters, you gain control of any creature until end of turn, untap it, give it haste, and make it a Villain on top of its other types. Read that against Loki's attack trigger and the trick appears: you can steal an opponent's best creature, turn it into a Villain, and then have Loki copy it on the same attack. You get their threat and a rental clone of it for one big turn. Point the stolen creature and its copy at two different opponents and Loki draws two cards, since his card counts players hit; send both at the same player and it is one. It is the most Loki thing in the deck, borrowing a face and wearing it into combat.
Loki's Scepter lets you borrow an enemy's creature for a turn and make it a villain. Then Loki can copy it too. You get their big creature and a copy of it for one attack.
The swaps that feed the copy loop
This is the part a surface deck tech skips. Salubrious Snail never reviewed this precon, so the read here is his method applied by us, not his take on the deck. That method is a dead-draw audit from his general video: hold every card up to the two engines this precon runs, the discard drain on one face and the Loki copies on the other, and cut whatever neither engine can use. Nothing generically weak comes out. What leaves is off-plan, and what comes in feeds the copy loop directly. Three swaps show it best, and two of them exist because of Loki.
That is the difference the correction asks for. The copy loop carries a cost, a token that leaves at end of turn, and the smart upgrades either monetize that cost with Kingpin or cancel it with Sundial. Read the deck through Snail's audit and the swaps stop looking like power upgrades and start looking like plumbing for the engine you are already running.
Salubrious Snail checks every card against the deck's two engines and cuts the ones that do not fit. Green Goblin is a strong villain: when he attacks he discards and draws cards, which helps the Doom side. Kingpin makes treasure the first time a creature dies each turn, so Loki's copy going away actually helps. Sundial of the Infinite can end the turn at just the right moment so the copy never goes away at all.
The one rule
Run bodies that pay off the moment they arrive. An enter trigger, a static buff, or evasion makes the rental worth it; an attack-only Villain or a plain beater just swings and vanishes.
That single test sorts the deck. When you are deciding whether a Villain earns a slot, ask what the rented copy does before it is sacrificed. If the answer is an enter trigger, a static buff, an evasive connect for Loki's card, or a connive off the token entering, keep it. If the answer is an attack trigger that never fires or a plain swing that vanishes, the slot wants a better Villain.
What the meta is building now
Loki's live EDHREC page, about 1,263 decks at rank 1,390, shows the crowd solving the copy engine's one weakness, running out of things worth copying, by stacking redundancy. The high-synergy column is a wall of clones and copy targets: Chameleon at eighty-two percent, Spark Double at sixty-eight, Mysterio at sixty-five, Taskmaster at fifty-nine, even Loki, God of Mischief at sixty-two, so there is always a strong body to duplicate. Around them sit the Villain-tribal answers a token deck wants, Kindred Dominance and Blasphemous Act as wraths that clear the table while your board rebuilds off the next copy, with Black Market Connections and Chaos Warp for card advantage and a catch-all. The plan runs deeper than the precon: build a clone pile so wide that every Loki attack finds a worthwhile rental.
One caution before you copy that list. In his EDHREC Effect video Snail warns the site inflates already-popular cards and overrates synergy pieces that only shine while your commander is out. Loki is a build-around, so much of the high-synergy pile does nothing on a turn he is dead, which is why the page overrates it. Treat the percentages as popularity, and keep enough standalone Villains that a board without Loki still functions.
Looking at real decks, about 1,263 people build Loki. Most of them run lots of clone cards like Spark Double and Mysterio, so there is always a good villain to copy, plus wraths that spare their own team. Just remember, popular websites can make a card look better than it is.
Snail Tip
The audit above is Snail's, and it is worth hearing in his words. His Reducing Dead Draws video is the source for the two-question test, that a card should feed one of the deck's engines or come out. His EDHREC Effect video is why the meta list is a starting point rather than a decklist. We worked his dead-draw audit onto this whole box, in his own words: read the Doom dead-draws page.
The trick of cutting cards that do not feed your engine comes from Salubrious Snail. Here is his video, and here is our page that uses it on this box.