TRENCHPAPER

Two players, 10 turns each, highest market cap at the end. Every turn you are given a marketing budget, and every card costs some of it to play. The budget grows: it starts at $40K and reaches $400K on the last turn.

Unspent budget comes off your market cap. Not metaphorically — it is subtracted when the turn ends. Sitting on money is not caution here, it is a payment, and it is the rule that stops a match being a race to hoard.

Only a project opens a position, and only a position pumps. A position pays its pump into your market cap at the end of every turn, so a project played on turn two is worth far more than the same card on turn nine — it is not what it costs, it is how many payouts are left.

Turns

10

Cards in hand

5

Positions at once

6

Budget, first turn

$40K

Budget, last turn

$400K

Throwing a card away, or banking a position

$50K

A full portfolio does not block a project — it asks you to close one. What that position already made stays yours; what you give up is everything it would have paid from here. That choice is the reason the cap exists.

750 cards, covering 155 real projects. A project that mattered is often several cards — a launch, a peak, an aftermath — because the interesting thing about most of them is not that they existed but what happened to them.

Projects

610

Influencers

60

Tactics

34

Tools

26

Events

20

Rarity is what a card does, not what it is worth. A mythic is a card that changes a match; it is not a card about a project with a bigger market cap. Rarity also sets the price: a common costs $20K to play and a mythic $280K, so an expensive hand is one you cannot afford early.

Common

189 cards · $20K to play · 50% of a pull

Rare

203 cards · $40K to play · 35% of a pull

Epic

177 cards · $80K to play · 9% of a pull

Legendary

103 cards · $200K to play · 5% of a pull

Mythic

78 cards · $280K to play · 1% of a pull

Card text is generated from what a card actually does rather than written beside it. A card cannot promise something the engine does not do, because the sentence is produced by the same object the engine reads.

40 cards, one copy of each, out of the cards you own. There is no deck budget — a card is paid for when you play it — so the question is not what you can afford to own but whether you have something worth doing with every turn’s money.

You draw a median of 31 cards in a match, so at this size you see nearly all of it. What you build is what you get.

One thing the numbers are blunt about: decks drawn from packs run from 21% to 74% against each other, measured. So a rank here is deck plus player rather than player. That is right for pairing people and a poor basis for anything that reads as merit.

Cards come out of packs and belong to a wallet, not a browser. There are three products: a single card, which guarantees nothing and is the only one with a price so far at 0.05 SOL; a pack of 10 with one slot guaranteed rare or better; and a deck mint of 60, which is twenty more than a deck so there is something left to leave out.

Every draw is against the whole set, so a card you already hold can come out again. Odds that stop being true once your collection fills up are not odds.

Holding $TCG never changes what you may put in a deck.

That was measured before it was decided. When decks had a points budget, one built on 110 points beat one built on 80 in 86% of matches. Selling deck power in a game people bet on is not selling a stronger deck — it is selling the result of the bet.

Matches are correspondence: a day a turn, so a match runs over a week and nobody has to be anywhere. When the clock runs out your turn ends — not the match. At one minute, forfeiting would mean a bad connection costs a stake, and losing a whole turn’s budget is punishment enough.

Every match therefore ends. A match both players abandon still resolves, as a draw, after twenty windows. There is no stuck-funds case in this system and so no refund path, no dispute window and no timeout to argue about — which is normally the hard half of putting money on a game.

The engine decides on the server. In the maker’s previous project the card check was a filter in the interface, and a direct call could play anything.

Rank is a plain Elo starting at 1000, and only staked matches move it. Friendly matches leave it alone: dropping a rank has to cost a stake rather than an afternoon.

The attack on a rating with money behind it runs downward. Nobody inflates a rating, because a higher one buys harder opponents. What pays is losing on purpose until the rank is low and then taking an expensive seat against somebody who genuinely belongs there. Three things are arranged against that, and none works alone.

The rank you are matched on

mean of your last 20 staked matches

Equal, and unrefusable

within 100 points

The stake is the weight

1×1 5×1.6 10×2 0.1×0.4 0.5×0.8

The 5 SOL seats

10 matches at 1 SOL or above

The 10 SOL seats

10 matches at 5 SOL or above

The lagging mean is the substance of it. Throw five matches and a quarter of the fall registers; decline honestly over thirty and all of it does. Falling is free when it is real and slow when it is chosen.

You may refuse a challenger far above you, and far below you — that second half is the one people forget. Taking a much weaker player’s stake is exactly what the band exists to stop. What you may not refuse is an equal, or the band becomes a way to sit in a seat and only take the matches you fancy.

Stakes are 0.1, 0.5, 1, 5, 10 SOL. Both stakes are in escrow before the match exists — not after the first move, because a player who has seen their opening hand and then decides whether to pay is not staking anything.

The escrow is custodial, and that is not dressed up.

One wallet holds both stakes. An on-chain program would hold them perfectly and still have to be told who won, because the match is decided off-chain — so the trust moves to a key rather than away from it. The only way round that is reimplementing every rule in a contract, which would be a second implementation of the rules and the thing the maker’s last project died of.

What is offered instead is that every payment and every payout is a transaction you can look at, linked from the match that produced it.

Supply is 1 billion, launching on pump.fun. Four things earn, and most of what they earn is spent buying $TCG and burning it.

pump.fun creator fee

50% The maker's own wallet · 20% Marketing and the people doing it · 20% Buy and burn $TCG · 10% Tournament prizes. Paid out, not spent

Paid mints

75% Buy and burn $TCG · 25% The maker's own wallet

NFT royalties

75% Buy and burn $TCG · 25% The maker's own wallet

Staked matches

100% Buy and burn $TCG

Every buy-and-burn runs through one wallet, so all of it lands somewhere anybody can watch. The wallets are published on the burn page with their balances read off the chain, and the burn total there is the sum of transaction signatures rather than a number in a database.

A smaller cut of what you win. Every staked match burns a share of the pot, and how big that share is depends on what the winner holds — your stake is gone either way when you lose, so a discount on it would only ever be a discount for the person who beat you.

RETAIL

anything, including nothing · 25% burned

BAGHOLDER

from 100,000 $TCG · 15% burned

HOLDER

from 1,000,000 $TCG · 10% burned

WHALE

from 10,000,000 $TCG · 5% burned

A balance that cannot be read is retail. A discount nobody could verify is a discount nobody earned.

5 things to do, one point each, and a point for every one of those 5 that somebody you bring in finishes. So a referral is worth up to 5, and somebody who does three of five earns you three.

Link your X account

checked

Link your Telegram

checked

Join the Telegram group

checked

Follow us on X

taken on trust, checked by hand

Play a demo match

checked

Points are spent, not scored. At 1500 you are choosing between both token lots and ten starter packs, and 1500 is 299 people plus your own 5.

One card

5 points · 0 people

A booster pack

25 points · 4 people

A starter pack

100 points · 19 people

100K $TCG

500 points · 99 people

250K $TCG

1000 points · 199 people

Wallets are free, so neither making one nor typing a code can be the bar. An X account is not free and this site attaches each one to a single wallet — that is what makes a referral cost something. Anything that looks botted is removed along with the points it earned, on both sides.

The list every paper leaves out. All of it is true at the time this page was last deployed, because it is read from the same switches the site runs on.

The token

not launched — there is no contract address

Paid mints

a single card is priced at 0.05 SOL and there is nothing that can take it

Cards as NFTs

not on chain yet

Live matches

correspondence only — live needs a connection that stays open

Claiming points

switched off until there is something to hand over

Tournaments

a wallet exists for the prizes and nothing else does

The escrow is a hot wallet. The server pays out without a human, so it holds a key that can move funds. Whatever is on it is the whole of what is at risk, which is why it holds what is in play and not more.

A rank is deck plus player. Somebody climbing may have minted better rather than played better. Fine for pairing people; worth remembering before prizes hang off it.

Sandbagging is expensive, not impossible. The lagging mean and the tier gate make a staged fall cost sixty SOL of real matches at weights that move a rank the whole way. That is a price, not a wall.

One task is taken on trust. X charges per follower read, so following cannot be checked automatically. It is marked as declared rather than verified, and it is removed by hand when it is not true.

The art references real projects and real people. The NFT metadata stays editable so a picture can come down if a rights holder objects. That is a repair and not a promise: marketplaces cache images hard, and some caches never refresh.