Pity, 50/50 and the Revenue Machine That Runs on Probability

A player opens their 87th pull and notices something odd: the five-star rate suddenly jumps. The published base rate is 0.6% per pull, a figure close to zero. But from roughly pull 74 onward, the curve changes shape, and by pull 87 the actual probability is many times the baseline. Players call it "luck turning." The system calls it soft pity — a function designed to distribute emotion pull by pull.

I have tracked this metric for years, not to play, but to understand how a probability system manufactures a feeling of certainty in a user's mind. The 87th pull is where mathematics meets psychology, and that is the point worth pausing on.

Context

The game here is Genshin Impact, published by HoYoverse. By genre it is an open-world role-playing game, single-player or co-op. It has no professional circuit, no club system, no transfer market in the esports sense. Labeling it as esports is a misclassification that should be corrected before any analysis is put to other use.

But its monetization model is worth the esports industry's attention, because it solves the problem every league struggles with: keeping money flowing steadily without a stadium. Esports lives on sponsorship, broadcast rights, skin-revenue sharing and prize pools — all dependent on a third-party ecosystem of sponsors, broadcasters and streaming platforms. Gacha lives on direct, recurring spending held entirely in the publisher's hands.

Esports is not slower than football — it simply runs on a different clock. But gacha does not even run on esports' clock. It runs on its own, and that is what makes it both worth studying and worth caution.

In Vietnam, mobile gaming is one of the largest digital entertainment spending categories. Vietnamese players sit within the wider regional flow, and understanding the architecture behind each banner means understanding how money leaves a wallet according to a predictable rule. This is a matter of system design, not morality.

I have tracked many major tournaments and many banner cycles side by side, and what caught my attention was not scale but risk structure. In esports, a team wins or loses on true talent plus a share of variance. In gacha, no true talent is measured — only spending behavior is. The total absence of performance data turns any character-strength analysis into pure speculation.

Analysis

Each Genshin Impact version splits into two phases of roughly 21 days, each with its own banner. This is a content-release cadence, not a match schedule. But that cadence is designed to create recurring, time-boxed spending windows — a kind of "season" for the wallet.

The pricing structure lies in the pity system. A five-star character is guaranteed within 90 pulls. On an event banner, the first five-star has a 50% chance of being the featured character and a 50% chance of being a standard character; if a standard character appears, the next five-star is guaranteed to be featured. This is the 50/50-plus-guarantee architecture, a design that maximizes both perceived accessibility and revenue variance.

The gap between base and consolidated rate is the place to look. The baseline is 0.6% per pull, but the consolidated rate across the whole curve is commonly recorded at around 1.6%. That gap is soft pity — probability shifted toward the end of the curve. Technically, it is a way of redistributing probability over time to turn a rare event into a near-certain one within a defined spending window.

One under-noticed detail: pity is shared across banners of the same type. The marginal cost of switching from one banner to another therefore falls. Behaviorally, this raises spending frequency, because players no longer feel they must start over each time they pivot.

On expected cost, calculating by the guarantee threshold, the average number of pulls to secure a featured character sits somewhere between 60 and 90, depending on whether the player wins or loses the first 50/50. That spread is designed variance — and also the hardest part of revenue to forecast.

The current banner structure runs on two lanes. The main lane serves new and returning characters. The secondary lane serves older characters under a separate rule set. The secondary lane exists so the publisher can re-monetize characters that have stopped appearing on the main banner without disrupting the main lane's cadence.

Data does not lie, but it learns to hide what matters most. What is hidden here is the rerun policy. The game does not publish a fixed rerun schedule. Some characters are absent for more than a year, others return within a few versions. That uncertainty is not an operational flaw — it is a deliberate scarcity mechanism, the equivalent of limited-time event design in entertainment. And when an older character returns through a separate banner type, the publisher gains an extra revenue lane for an asset assumed to be dormant.

I once built a simple tracking table for banner cycles to see whether they followed any pattern. The result: no fixed cycle, only correlation with the content calendar. The key point is here — this model's revenue does not depend on any external cultural or sporting event. It runs itself, and is therefore less exposed to calendar shocks than esports, but more exposed to regulatory shifts.

In that value chain, the publisher is simultaneously the game operator, the gacha rule-maker and the information authority. This concentration of power is markedly higher than in most esports ecosystems, where at least a few independent parties exist: tournament organizers, teams, media outlets. Here, no independent arbiter verifies the published numbers.

Contrarian angle

The common reaction to gacha is to call it gambling. That label is convenient, but it misses the important point. Technically, players always receive something after each pull, and there is always a guarantee threshold. That structure is closer to a subscription with variance than to a wager. The real risk lies in the predictability of the FOMO architecture: thresholds, countdown clocks, and a fear of missing out programmed in advance.

There is a warning about the source itself. Most information about future banner schedules in articles of this kind has no clear origin. Only a small share comes from official announcement channels; the rest is speculation. Some character names and version numbers circulating online cannot be reconciled with the known game state. For readers, that is the risk of acting on false information.

Many articles of this type carry a promotional tone and little verification. It is a traffic-filter content type — written to attract attention ahead of a version, not to provide verifiable information. Recognizing this content type matters no less than reading the content.

Variance is not the enemy — it is the mirror that shows prediction its own arrogance. When an article says "save for the next version," readers should ask themselves: is this a conclusion drawn from data, or merely a schedule read aloud? A schedule is not value. Knowing "when" does not mean knowing "whether it is worth it."

Takeaway

What is worth tracking next is not which character name appears, but two signals: official confirmation from the publisher's channel, and any change in probability-transparency rules. A season is a statistical sample. A decade is the evidence. With models that run on probability, the most important question is who holds the rules — and whether anyone else can verify them.