90 Pulls and a 21-Day Rhythm: The Gacha Revenue Architecture Seen Through an Esports Lens

Last night I sat down with my data board again, and one number kept jumping out at me. An open-world role-playing game, with no professional circuit, no teams, no transfer contracts, is running a revenue engine that most esports leagues can only look at with a different kind of eye. Three gears assemble that engine: a 90-pull pity threshold, a 50/50 split between limited and standard rewards, and a two-phase rhythm per version, each phase roughly 21 days. They mesh into what I call a scarcity architecture, measured by spending rules rather than play rules. Revenue structure determines competitive structure. I wrote a blog from a rented room in Nha Trang; now probability takes me everywhere, and everywhere I go I see the same law: whoever pays defines the rules. A league that lives on sponsorship, media rights, jersey sales, and prize money behaves very differently from a game that lives on direct top-ups from players. Esports still stands on four legs: sponsors, broadcast rights, in-game items, and prize pools. All four lean on one shared asset — audience attention. When attention rises, all four legs lengthen. When attention stalls, all four jam at once. That is why a league can explode for two years and shrink for the next two, with the cause sitting nowhere near competitive quality. The gacha engine runs on a different logic. It does not wait for an audience. It waits for players to open their wallets. And it is designed to make them open those wallets on the version rhythm, not the season rhythm. This is the point I want to dissect with data, because it carries a lesson many esports teams are trying to copy without fully understanding the mechanics. The title I am referring to is Genshin Impact, published by HoYoverse, and every mechanic below is publicly disclosed in the game's own system descriptions. I did not take them from rumors. I took them from the very lines the publisher is obliged to print so players know where their money goes. The first gear is the 90-pull pity threshold. In a gacha system, players spend to pull for characters. Each pull is a roll of the dice. Without a backstop, this is pure chance, and players walk away when a cold streak drags on. The 90-pull threshold breaks that psychology: players know for certain that within 90 pulls at most, they will get the character they want. It is an expectation-framing mechanism, turning a gamble into a plan. Statistically, it is a clever design. Players feel a spending ceiling, so they dare to start. But the actual probability of pulling the character usually arrives earlier than 90, so most players never touch the ceiling. That ceiling exists to lower fear, while the bulk of revenue sits in the middle. The second gear is the 50/50 mechanic. The first limited-character pull has a fifty percent chance of hitting the advertised character and a fifty percent chance of hitting a standard one. Miss, and the next five-star pull is guaranteed to be the limited character. This creates two emotional states: euphoria on a win and frustration on a loss. Both drive spending, but in different ways. Winners continue because they feel hot. Losers continue because they believe in the guaranteed pull coming next. What matters in data terms: if the 90 threshold is a hard variable, the 50/50 split is a variance variable. It turns player spending into a long-tail distribution — most spend little, a small group spends a great deal. In revenue economics, this is the structure every operator wants, because it does not require everyone to spend heavily, only a small share to spend enough. The third gear is the two-phase rhythm per version. Each version splits into two phases, roughly 21 days each, each with its own character set. It is a countdown clock that loops forever. Players are never in a finished state, because the next phase always arrives before they can rest. Compared with a sports season running several months and then pausing, the 21-day rhythm is far denser, and evenly dense. Three gears mesh into a loop: framing creates willingness to spend, variance creates emotion, rhythm creates repetition. This loop needs no audience, no stadium, no commentators. It needs only a player with a digital wallet and a countdown clock. This is where the story gets interesting for esports people. Many leagues try to manufacture scarcity around in-game items, but they do it crudely: limited sale for a few days, then permanent closure. That creates one spike and then silence. The gacha engine does the opposite: it never closes, it only rotates. The scarcity sits not in never coming back, but in not coming back right now. The gap between the two approaches is huge in long-term revenue. One more detail deserves a pause: the rerun schedule for older characters is not fixed. Some characters are absent for more than a year before returning; others come back after only a few versions. This uncertainty is a tool. When players do not know for sure when their favorite character returns, they tend to spend the moment it appears, out of fear of a long wait. This is a deliberately designed fear-of-missing-out mechanism, not publisher caprice. Beside it sits a second lane for older characters, running on its own rulebook. It lets the publisher re-monetize characters that have stopped rerunning on the main lane without disturbing the main lane's rhythm. This is the same problem many esports teams face: how to re-monetize an old asset without diluting its value. Gacha solves it by splitting the two lanes outright, each with its own rules. One more technical point: pity counters of the same type share their threshold. Pulls made on one character carry over to another of the same category. This lowers the switching cost between windows, and behaviorally it makes players more likely to keep pulling than to stop. It is a form of revenue smoothing across phases — not raising total spend, but flattening it. I tried placing these three gears next to the revenue structure of a typical esports league. The result was fairly clear: esports is seasonal, gacha is short-cycle. Esports depends on outside sponsorship, gacha on direct player spending. Esports needs audiences to stay, gacha needs players to return. Two models demand two different sets of operational skills. This explains why many esports teams try in-game item models but never match the performance of gacha titles. Esports has an audience, but the audience and the payer are not the same group. Gacha has payers, and the payers are also the content consumers. When those two groups overlap, the loop closes and the efficiency is far higher. But this is where I have to speak against the crowd, and speak with numbers. First point: correlation between good design and high revenue does not equal causation. The gacha engine works partly because it is attached to a game with tens of millions of players and a vast community of content creators. Strip the design away from that community and efficiency drops sharply. I do not have enough data to isolate the two variables, so I will only state it probabilistically: a large share of the engine's success comes from the base product, more than from the spending design itself. Second point: the model itself carries a legal risk esports rarely faces. Paid random mechanics are in the crosshairs of regulators in many markets, with probability disclosure and minor protection requirements. If the legal framework tightens, some of these gears may have to change, and the whole engine would need redesigning. Esports, with its traditional sponsorship and media-rights structure, is less exposed to this kind of risk. Third point, and perhaps the most important: I am not sure esports should copy this model. The two communities have different relationships with the idea of luck. Sports fans follow because they want to see skill rewarded. Gacha players enter a system they themselves know is random. Treating sports fans' emotions as a variable to optimize spending is the shortest path to losing them entirely. I have seen community backlashes when operators push addictive mechanics too far, and those backlashes are not small. In other words, an engine that runs well is not necessarily an engine that runs long. And an engine that runs long is usually not the loudest one in the short term. An empty stadium does not need spectators; it needs an analyst willing to look — and in this case, willing to look at numbers that never appear on a scoreboard. The match ends, but the data remains. For me, the value of the gacha engine lies in three questions it raises for the sports industry, more than in its three gears: how to frame expectations so fans dare to believe, how to set a rhythm even enough that no one forgets, and how to keep trust intact when the system itself is random. Of those three, the third is the hardest, and it is the one most esports teams have yet to answer. If next season brings another team testing a random-item model, I will look at the rate of players returning after their first miss before I look at revenue. That is the earliest signal, and also the most honest one.