the setup nobody prices, and the turn the free curve has already won

Issue #200, 2026-09-13. Constructed, Duel (485). Rules version 2026-07-16, opened in data/Riftbound-Core-Rules-2026-07-16.txt and data/Riftbound-Tournament-Rules-2026-07-16.txt; card text verbatim from data/corpus_flat.txt. Read the unopposed clock first for what a baseline is.

Subject: bottled-constellation-time-warp — the row the turn clock moved furthest in a single day, from T14 to T5, and the one that shows what the clock still cannot see.

1. Why it moved nine turns, and why that part was right

CLAUDE.md prices this line as "60 Energy and 18 Power of deployment, i.e. turn 14". The entry's own first step says otherwise, and has all along:

"Turn N: loop; play three Constellations (30E+6P); Time Warp."

The sixty Energy and the eighteen Power were never a deployment. lux-infinite-energy and lux-infinite-power are already running when they are paid, so the loop buys them. The clock was charging a rune curve for cards the rune curve never pays for; it now prices an engine's own output and the row reads T5. That change is correct and it is in scripts/adversarial-check.mjs.

And it is the smaller half of the story. Removing a nine-turn over-charge took away the padding that had been hiding a constraint underneath it.

2. What the clock prices, and what this line actually needs

The clock prices cards. lux-infinite-energy states two requirements in its own prerequisites.notable that no card cost can reach:

"12 runes in play (the deck's whole Rune Deck channeled); at least one Mind rune." "Main Deck EMPTY — the deck draws itself out first; every card not in the loop must be in hand or trash."

Twelve runes is turn six at the earliest. 315.3.b channels two a turn, 161.2.a fixes the Rune Deck at exactly twelve, and 161.2.b takes a rune off the board for every Power paid — so T6 is the best case and every Power spent on the way pushes it out.

Ten of the eighty rows in the clock carry that engine in their closure, and every one of them prints T5 — bottled-constellation-time-warp, grand-plaza-loop-time-warp, jayce-mesmerize-renata, lady-luminosity-loop-comet, lux-infinite-energy, lux-infinite-power, renata-bubble-bot-ready, renata-mastermind-points, swain-double-conquer, time-warp-hold-burst. Read all ten as no earlier than T6. That set is exact in a way the floor's source is not: it is "the closure contains lux-infinite-energy", and that entry declares both requirements in its own words.

The empty deck is the one with no number on it, and the catalogue already knows it is fragile. lux-infinite-energy carries the finding in its own words — "THE EMPTY MAIN DECK IS A ONE-TURN STATE THAT DESTROYS ITSELF" — because 315.4.b.1 makes an empty deck at your Draw Phase an automatic Burn Out, 431.2.b then "Recycles their trash into their Main Deck" randomized, and 431.2.c hands an opponent a point. Riot works the whole sequence at 431.2.d.

So the window is exactly one Main Phase: the Main Phase of the turn on which your last card is drawn. What no entry and no instrument prices is how long it takes to get there.

3. Pricing it

Tournament Rules 601.1.b: "In competitions, a player's Main Deck must be exactly 40 cards." 103.2.a.1 sets the Chosen Champion aside "at the start of the game", so 39 are in the deck. Rule 116 draws four. Thirty-five cards to remove, and the Draw Phase (315.4.b) removes one a turn.

Only two things take a card out of a Main Deck: drawing it and burning it. Recycling does not — 416.1 puts it on the bottom. So what matters per deck slot is the net: a card that draws two nets one, because one of those draws found the card itself.

Swept over the pool for cards legal in Mind/Order (103.1.b — domains a subset of mind + order, colourless included), folded by name, legends excluded: nine names remove more than one card from their own Main Deck.

One of the nine costs nothing and repeats. VEN-165 Shadow Temple is colourless, so any identity may bring it, and its burn carries no "may" — it is a fuse, which is also why overshooting it is a Burn Out under 431.1.b.

The timetable with the Temple, going first (485.7 gives the extra rune to the player going second, so this is the slower seat). A body played on T1 enters exhausted (143.4), moves in on T2 and Conquers — 190.3.a.1 applies Contested, 344.2 opens the Showdown at the next Cleanup, 348.2.a establishes Control. From T3 the Temple Holds every turn.

Four a turn from T3 — three burned, one drawn. Turn ten or eleven on the Temple alone, and adding a dozen net draws from the table above pulls it to roughly turn seven or eight, which is also when twelve runes and the loop's pieces can plausibly be in place.

4. The sting, and it is one line rather than ten

485.4.a has each player provide three battlefields and 485.5 selects them at random, so exactly one of your three is on the table. The Temple therefore competes with any battlefield the line itself needs.

Nine of the ten rows in this family name no battlefield, so they can simply run it. One cannot: grand-plaza-loop-time-warp needs OGN-293 The Grand Plaza on the board and an empty Main Deck, and the cheapest emptier in its identity is a battlefield it can never have at the same time. That line's deck has to empty itself out of the Energy-priced half of the table instead — E1 to E7 a card, paid on the same turns it is trying to reach twelve runes.

The deck that empties itself cannot be the deck that wins, for that entry and only that entry.

5. Verdict

Everything about the combo is intact. The loop works, the arithmetic in the entry is right, and the clock's T5 is the right answer to the question the clock asks. The question it asks is "when are these cards affordable", and this line's binding constraint is not affordability.

Priced as a game rather than as a combo, it is a turn seven-to-ten line, and the way it gets there is the thing to sit with. The cheapest route to an empty deck is holding a battlefield every turn from T3 — and holding a battlefield every turn from T3 is the contested Hold curve this project measures everything against, the one that pays a point a turn and reaches eight on T9. So the setup the loop requires is itself the plan the loop was supposed to improve on.

That is the unopposed-clock argument arriving from a direction it was never pointed at. It was derived there from what a finisher is worth against a free curve; here it falls out of a deck literally having to spend the same turns doing the same thing to become legal at all.

And it is not an accident of this one deck. 469 defines Scoring as Conquer or Hold and nothing else, and 315.2.b.2 Holds every battlefield you control at the start of every Beginning Phase — so a card whose trigger is a Hold cannot be used without scoring. Swept over the pool, 35 names carry a free, non-point, Hold-triggered ability, against 22 that fire once on a Conquer and therefore carry no rate at all. Setup time is not a pure cost in this pool: a deck accumulating from a Hold-triggered faucet for N turns has scored N points doing it, and the contested curve reaches eight on T9. Shadow Temple is one of thirty-five, not one of one.

What I am not claiming. Not that the entry is wrong — it is not, and its own notable already records that the empty deck destroys itself. Not that seven-to-ten is exact: it is a hand timetable over one shell, and a Mind/Order list built specifically to empty fast would beat it. And not that the clock should model any of this — a deck's emptying rate is a property of the list, not of the line, which is precisely why the instrument that prices lines cannot see it, and why this is a play rather than a patch.