The 90-Pull Architecture: Reading Genshin Impact's Gacha Model Through an Esports Framework

In the described schedule, phase one of version 7.1 introduces two limited characters at once — Vesna and Vodyanitsa — while phase two carries only reruns. The game splits each version into two phases of roughly 21 days apiece. Packing two new slots into a single window is a design decision, not a calendar coincidence.

I opened this document as a transfer-data analyst, not a player. And I will say it plainly in the first line: the true domain of this material is gacha operations, and it does not belong to the esports ecosystem. Genshin Impact has no professional circuit, no clubs, and no transfer market in the competitive sense. The "Esports" label assigned to it by the classification system is an error, and a wrong label corrupts every calculation that follows it.

What kept me in the document was a different detail. While cleaning the data to strip out what could not be cross-verified, I noticed the gacha architecture here runs almost exactly like a transfer market — except that no players exist at all.

Context: a system engineered to sell timing

Genshin Impact's gacha engine has three core parameters. First, the pity floor: a five-star character is guaranteed within 90 pulls. Second, the 50/50 mechanic: the first five-star on an event banner has a 50 percent chance of being the limited character and a 50 percent chance of a standard character; if it lands on standard, the next five-star is guaranteed to be the limited one. Third, shared pity across banners of the same category, meaning accumulated pull counts carry over when a player switches between same-type banners.

Read together, those three parameters form a familiar pricing architecture. The 90-pull floor acts as a price floor — it turns a random process into a cost with a defined ceiling. The 50/50 mechanic produces variance. Shared pity reduces switching friction: players are not punished for changing targets between two same-category banners, so total spending frequency tends to rise rather than fall.

The document also describes a rerun schedule with no fixed cadence. Some characters are absent for more than a year; others return within a few versions. Alongside sits a separate banner lane — Chronicled Wish — operating under its own rule set, typically for older characters. This is the point I want to linger on longest.

Before going further, a note on sourcing. Of the document's 28 information points, 20 carry no source, only one cites an official publisher announcement, and three are the author's opinion. Several proper names — Odette, Flins, Ineffa, Vesna, Vodyanitsa — along with version numbers 7.0 and 7.1, cannot be cross-checked against any known state of the game. I record them as hypotheses, not established facts.

Core: four mechanisms, one engine

Start with the rerun schedule. A non-fixed cadence is not operational laziness. It is a scarcity mechanism. When players do not know when the character they are waiting for will return, the opportunity cost of skipping a rerun rises, and the spending decision is pushed toward "do it now." Every transfer is a murder case. The culprit is expectation; the weapon is timing.

Next, Chronicled Wish. This banner lane lets the publisher re-monetize characters past their cycle peak without consuming a slot on the primary banner. In transfer language, it is reselling a player past his prime without diluting the starting squad.

Third is the two-phase rhythm, roughly 21 days per phase. This cadence does not create competitive outcomes; it creates time-boxed, regularly repeating purchase windows. It is revenue-rhythm design, and it explains why currency-allocation pressure peaks hardest in phase one of version 7.1 — where two new characters arrive together. On the reverse side, those who spent on the phase-two reruns of 7.0, namely Flins and Ineffa, enter the new window with depleted resources. In any resource-allocation system, the loser is not the one who spent wrongly, but the one who spent correctly at a moment the system had already priced in.

Fourth is shared pity. It blurs the boundary between "new character" and "rerun" from a marginal-cost standpoint. A player who has accumulated 60 pulls does not face starting from scratch when switching banners in the same category. Low friction, high frequency. This is a design that flattens the decision curve: it does not persuade players to spend more at once, it persuades them to spend more often.

I used to think I was reading the match map; it turned out I was only looking at a mirror reflecting my own fear. Years of watching this industry taught me that every revenue system simulates a specific fear. Here, the simulated fear is missing out, and it has been programmed into a measurable variable rather than left as a vague emotion.

Contrarian angle: a schedule is not value

The document frames its story around one piece of advice: prepare for version 7.1. But it supplies no strength data whatsoever. No kit analysis, no comparison table, no roster evaluation. The only confirmed element is the schedule — and even that, the author concedes, is still awaiting official confirmation.

The market does not move on news. It moves on the gap between two reports. Here, the gap sits between an unconfirmed schedule and a fully confirmed expectation. Readers get the answer to "when" and are left short of "whether."

This touches a habit of the entire industry. When strength data does not exist, the schedule story fills the vacuum, carrying an implicit assumption that the new is always worth more than the old. That assumption has never been proven. Correlation between "new version" and "high demand" is not causation. The publisher controls both ends: it sets the schedule, sets the power level, sets the announcement. When one entity is simultaneously rule-maker, operator, and announcer, every claim about value passes through its hands before reaching the reader's.

And here is where I audit myself. If I had to find a flaw inside that "perfect system," what would it be? It is the absence of any independent arbiter verifying the published odds. There is no VAR for the die. The publisher publishes the rates, and the publisher is also the party profiting from those rates. In sports history, whenever those two roles merge into one body, disputes arrive later and resolve harder.

Takeaway: one map, one gap

I cannot apply a tournament, roster, or transfer-market framework to this document — not because the framework is wrong, but because the subject lacks those things. What I can extract is a lesson in revenue architecture. Four mechanisms — the 90-pull floor, the 50/50 variance, rerun scarcity, and the secondary banner lane — combine into a self-contained engine. It depends on no external cultural or sporting event, so it is less exposed to calendar shocks than most esports ecosystems. In exchange, it exposes itself to a different risk: shifts in gacha and paid-randomness regulation.

K League 2026 taught me that the pioneer does not fail for looking far, but for looking far while miscounting one column of data. The column this document miscounts is sourcing. Twenty of twenty-eight information points have no origin. Proper names remain unverified. A few future version numbers remain unestablished.

What I take from this piece is not a list of banners to wait for. It is a question about convergence. As esports ecosystems lean ever harder on direct fan revenue — skins, battle passes, limited items — how thin will the line between the esports model and the gacha model become? And if that line does thin, who will verify the metrics both sides publish?

Applause in an empty stand is not noise; it is a signal from a future we have not yet been brave enough to index.