Pity, 50/50 and an Unfixed Rerun Schedule: The Revenue Mechanism Inside Genshin Impact's Banner Machine

They sit and count. Every time the 'wish' button is pressed, a hidden counter advances by one. At the ninetieth pull, if no five-star has appeared, the system is forced to return a guaranteed result. There is no emotion in that calculation. Only a threshold, a probability, and a silent contract between a publisher and tens of millions of accounts waiting for exactly this moment.

I keep an old habit: when a phenomenon draws in enough attention, the first move is to strip it of its media packaging and look at the structure underneath. The banner cycle of Genshin Impact — version 7.0 phase two, then 7.1 with its two subsequent phases — is one such structure. It is not a match, there is no extra time, no scoreboard. But it runs on a rhythm no less rigorous than any league I have ever watched from an empty stand.

What made me stop was not the character list. It was the way a system is engineered to turn time into money — steadily, predictably, and almost independent of any external factor. That is a machine. And I have always been drawn to machines.

According to published information, each version splits into two phases, usually around 21 days each, and each phase carries its own banner. With the 7.0–7.1 cycle, the picture is fairly clear: the second phase of 7.0 leans toward reruns — re-offering characters launched earlier — while 7.1 phase one is designed for two new characters simultaneously, and 7.1 phase two returns to reruns. Within that structure, peak spending pressure lands on phase one of 7.1, when two new characters force players to allocate a scarce resource: accumulated premium currency.

Let me state plainly here something many articles overlook. This is not esports in the sense I normally work in. Genshin Impact is an open-world action role-playing game, single-player or co-op, running on a gacha monetization model. It has no official professional tournament circuit, no franchised club system, no transfer market, and its version updates are PvE content drops, not competitive balance patches. Calling it esports is a category error. But precisely because of that, it becomes a valuable comparison object: an entirely different revenue model, placed beside the esports model I analyze daily.

The core lies here: esports lives on sponsorship, broadcast rights, skin-revenue sharing and prize pools — a value chain with many participants. Gacha lives on direct, recurring, in-game spending performed by the players themselves, with no third party required. These two systems have fundamentally different risk structures.

Start with the hardest part of the machine: the guarantee mechanic, or pity. A five-star character is guaranteed within a maximum of 90 pulls. On an event banner with a limited character, the first five-star has a 50% chance of being the limited character, and the other 50% falls to the standard-character pool. If that attempt misses, the next five-star is guaranteed to be the limited character. Players call this the 50/50.

I spent considerable time looking at the number 90 and the number 50 the way an analyst looks at a passing chart, not the way a player looks at their luck. And what I found was a design that is nearly perfect psychologically. The threshold of 90 creates a sense of reachability: players know that however unlucky they are, there is a stopping point. The 50% chance on the first attempt creates variance: the outcome is unpredictable, and that very unpredictability is what keeps people going. And the guarantee on the next attempt turns a loss into an investment rather than a blank write-off — it keeps players in after a miss instead of pushing them away.

That is the architecture. Three layers, interlocking, serving a single purpose: sustaining spending across multiple cycles, not just within one.

But the machine does not stop there. There is a rarely mentioned detail with high technical value: pity is shared across banner types of the same category. That means if a player has accumulated pull progress on one banner and moves to another of the same type, the guarantee progress is preserved. Economically, this is a mechanism that lowers the marginal cost of switching between same-type banners. It reduces the psychological barrier that makes players hesitate to start a new banner, thereby potentially increasing overall spending frequency. A design that appears to benefit the player, but is simultaneously a subtle revenue lever.

Now look at the rerun schedule. The notable point is that it is not fixed. Some characters are absent from banners for more than a year, while others return after only a few versions. That uncertainty, from a system-design perspective, is a deliberate scarcity mechanism. It produces fear of missing out — what the industry calls FOMO — in a controlled manner. Players cannot plan long-term with certainty, so they tend to seize the opportunity the moment it appears.

Add to that the Chronicled Wish mechanic, a separate banner type with its own rule set, usually aimed at older characters. Its existence implies a secondary monetization lane for characters past their life-cycle peak. It reduces the pressure to bring old characters back to the primary banner, while opening a new revenue channel for seemingly dormant assets. For an analyst, this is a sign of a system mature in exploiting product life cycles.

I recall the summer of 2026 in Seoul, coding fourteen matches of an opponent into a pressing map. Back then, I learned that every complex system has a deviation point outsiders cannot see, because they only look at the final result. With this banner machine, the deviation point lies here: the public looks at a character list and calls it news, while the real story lies in the financial structure behind that list.

Every dynasty carries the gene of its own collapse; the tournament is merely the day that gene expresses. But here, there is no dynasty waiting to fall — only a wheel designed never to stop. That is the fundamental difference between a competitive system and an exploitative one.

I do not write about the plays, I write about how time evaporates across each half. Here, time does not evaporate across a half. It evaporates across 21-day phases, steadily, like a countdown clock wound by the publisher itself.

There is an aspect I am obliged to raise, however unglamorous. Most of the information in the underlying document lacks verifiable sourcing. Of 28 information points, 20 carry no source, only one cites the publisher's official announcement, and three are recorded as the author's opinion. Several character names and version numbers cannot be cross-checked against known game state. This is a reliability warning, not a content one.

In twenty-three years in this trade, I have learned that the most dangerous thing is not false information. It is half-true information presented with full confidence. Because the true half makes readers believe, and the false half makes them act on that belief.

The document itself concedes one thing: the exact banner schedule is still awaiting confirmation. That is an honest signal, but also a confession that the entire content is provisional. Every forward-looking claim in it should be treated as a hypothesis, not an event.

From a contrarian angle, I want to pause on a common misconception. Many people, including in media, tend to file Genshin Impact alongside esports because it is a video game with a massive following. But similarity in audience scale does not mean similarity in operating structure. Esports is defined by organized competition: teams, players, coaches, transfer rules, competitive integrity, champions and losers. The gacha machine has none of these. It has no winners and losers in a competitive sense — only spenders and non-spenders.

The second misconception, deeper, is the notion that gacha is simply legalized gambling. That view ignores design. Gacha is a structured monetization system, with calculated thresholds, compensatory mechanics, and scheduled cycles. It does not rely on pure luck, but on expectation management. The difference between a game of chance and a revenue machine lies here: a game of chance lets the outcome decide, while a revenue machine is designed so players decide to continue.

I once wrote that a cold dressing room in 2026 taught me intuition is no longer king. That lesson applies here differently. Crowd intuition suggests a new version is an event. But the data shows it is merely a beat in a pre-programmed cycle. Excitement and mechanical fact are two different things.

One point deserves attention. If we view this machine as a risk analyst, the greatest risk to the reader is not losing money, but making decisions on unverified information. When an article says 'this is a good time to prepare for version 7.1' yet offers no data on a character's strength or kit, it is advice about scheduling, not advice about value.

Vua Bong said: reason is also a kind of passion; it simply does not know how to celebrate. Here, reason demands one simple thing: distinguish 'when' from 'is it worth it.' The banner machine answers the first question very well. It barely answers the second.

Let us return to the financial dimension, where the story truly lies. The publisher is simultaneously the game's operator, the gacha rule-maker, and the official announcement authority. Three roles concentrated in one entity. In the esports model, power is usually more distributed: publisher, tournament organizer, teams, player associations, sometimes even a sports regulator. The concentration of power here is notably higher, both in supply and in information.

That means no independent arbitration mechanism verifies the published rules. The publisher defines, publishes, and benefits. This structure is operationally stable, but thin in oversight.

The only external factor that can affect this machine is regulation. Rules on probability transparency and protection of vulnerable players, especially minors, are tightening in many markets. The pity thresholds and disclosed probabilities in the document partly reflect the transparency requirements imposed by law in some jurisdictions. Because the rulebook is written and published by the publisher itself, there is no independent arbitration verifying that disclosure — a tension similar to what I have seen in esports governance.

The stadium in 2026 was empty, but I still heard footsteps in the data maze. In 2026, with no spectators, I discovered that the rules still operated, just differently. Here too. With no stands, no scoreboard, no commentators, the machine still runs. Only numbers moving through a pre-designed pipeline.

I want to stress one thing about this topic's analytical value. By competitive criteria, it is empty. No rosters, no form, no injuries, no balance patches. But by industry criteria, it is a clean case study of a monetization model based on direct user spending. For someone tracking money flows in esports, this is a worthwhile comparison point.

Esports depends on schedules, seasons, sponsors, streaming platforms. The gacha machine depends on one thing only: players' habit of recurring spending. That makes it less vulnerable to calendar shocks than esports, but more exposed to gacha regulatory shifts.

One more point worth recording. The two-phase rhythm, roughly 21 days each, is a monetization-rhythm design. It creates recurring, time-boxed purchase windows. Players face not only what to buy, but when to buy. Deadlines turn a decision from an open choice into a framed one. And framed choices are always made faster than open ones.

I once said transfers are not where people are bought and sold; they are where clubs reprint their own fate. Here, rephrase: a banner is not where characters are sold; it is where the publisher reprints its own revenue cycle. Each banner reaffirms a silent contract with the player: you will have a chance, within a fixed window, at a fixed probability.

And because that contract is reprinted regularly, players rarely get a chance to stop and ask whether they should participate at all.

I think of the concept of a war of attrition I use when analyzing teams in the final twenty minutes. The rule allowing multiple substitutions turns the endgame into a battle of stamina and reserves. The banner machine has a similar logic. It does not attack the player with a single blow, but with a small decisions chain, continuous, each appearing reasonable in isolation. The system's power lies in its repetition, not in any single banner.

This is why I believe banner-cycle analysis has archival value. An article about a single banner expires in three weeks. An article about the machine's structure lasts longer.

A word on data. In the document, phase one of 7.1 introduces two new characters at once, while the other phases are reruns. Architecturally, placing two new characters in the same time window creates a peak of resource-allocation pressure. Players must choose, or spend more to get both. Both choices benefit the publisher, just in different ways. This is an observation about revenue architecture, not competition.

With this focus, let me return to a principle I always follow: without evidence from footage or data, make no judgment. Here, I have no footage, and very little verified data. That means what I present is structural analysis, not result prediction. I describe how a system is designed, not claim it will lead to a specific outcome.

In the document, one observation is recorded with medium confidence: the pity mechanism combined with an unfixed rerun schedule can produce spending spikes around unpredictable reruns. This is conceptually close to how esports games exploit limited-time events — for example, selling skins or items only within a short window. The difference is that in esports, buyers usually know the item's price. In gacha, buyers only know a probability. That is a structural difference in information, and information structure shapes behavior.

There is one more aspect I want to view objectively. The document mentions Chronicled Wish as a monetization lane for older characters. If true, it is a mechanism that lets the publisher re-monetize assets past their life-cycle peak without disturbing the primary banner rhythm. In product life-cycle management, this is a sound strategy. For consumers, it is a reminder that in such a system, no asset truly disappears. Every character can be returned to market when conditions are favorable.

This leads to a counterintuitive conclusion. In esports, a player past their career peak sees value decline, usually irreversibly. In this system, a supposedly obsolete character can be brought back and generate revenue. The life cycle here is not a downward curve but a repeatable loop. This is the fundamental difference between a talent market and an asset market.

I want to devote the closing of this analysis to signals worth tracking, because a beat keeper always ends with what to observe next, not with a summary. The first signal is the publisher's official announcement of the specific banner list for each phase of 7.1. It will confirm or refute current information. The second is verification of the character names and version numbers through official materials. The third is regulatory developments on gacha in major markets, as that is the only external factor capable of altering this machine's structure.

The beat keeper knows silence has a beat too — especially when the stadium has no spectators. Here, the silence lies in the gap between official announcements. Between a banner just closed and one about to open, there is a pause in which every speculation can sprout. That pause is where unverified information breeds fastest.

And in that pause, I choose to keep a single principle from twenty-three years in the trade: trust only what can be verified, and record what cannot as a variable to track, not a conclusion to publish.

The banner machine will keep turning. It will keep printing 21-day cycles, keep setting 90-pull thresholds and 50% probabilities, keep leaving pauses that people fill with expectation. The only thing that changes is the list of names. The structure does not.

When a system runs smoothly enough that people forget it is running, it has succeeded. The next question is not what the next banner contains. The next question is: how long before people start reading the structure instead of the list?

I have watched enough matches to know people usually only look at the scoreboard. And in this field, there is no scoreboard. So they look at other numbers — the numbers the system itself has chosen to place before their eyes. It is a scripted performance, and I always want to look at the wings, where the cables connect to the controller.