The 90-Pull Pity Floor: A Money Machine Esports Has Not Dared to Copy

I worked out that number on a Miami evening, pulling apart the banner schedule for Genshin Impact versions 7.0 and 7.1. The 90-pull pity floor, the guaranteed five-star threshold, comes to roughly 200 US dollars at standard top-up rates. Add the 50/50 system: the first five-star on an event banner has exactly a coin-flip chance of being the promoted character, and a coin-flip chance of missing. Miss, and the next five-star is guaranteed. People call that luck. I call it a price list written in probability. People call it shocking when I say so. I call it a map, and that map runs straight from a player's thumb to the publisher's vault.

Context: a game sitting outside esports territory

Let me be blunt up front: Genshin Impact sits outside esports territory. It is an open-world action role-playing game from HoYoverse, run on a gacha model. It has no professional circuit, no international tournament, no club system, no player-transfer market in the esports sense. What it calls a version is a PvE content drop, not a competitive balance patch.

Its monetization architecture, though, deserves an esports audience. Each version splits into two phases of roughly 21 days, each with its own banner. The second phase of 7.0 is a rerun, re-offering older characters Flins and Ineffa. The first phase of 7.1 is reported to launch two new characters simultaneously, Vesna and Vodyanitsa, per the official Genshin Impact announcement channel. The second phase of 7.1 returns to reruns.

In the dataset I have, that is the single information point carrying an official source. Twenty of twenty-eight points carry none. The source material itself concedes the exact banner schedule is still awaiting confirmation. I flag this before analysing anything, because unsourced data can still be useful for dissecting structure, but it cannot be used to assert facts.

A closed revenue engine

This is where I stopped longest. Three mechanisms, the 90-pull pity floor, the 50/50 system, and pity sharing across same-category banners, combine into a closed revenue engine that esports currently has no equivalent for.

First, the pity floor. Because players know they will receive the character within a maximum of 90 pulls, the maximum cost is priced in advance, before anyone clicks a button. That is a ceiling, not a floor; the lucky pay less, the unlucky pay exactly the ceiling. The publisher never has to guess what a player is willing to spend. The architecture answers for them.

Second, the 50/50. On an event banner, the first five-star has exactly a half chance of being the promoted character. Miss it, and the next one is guaranteed. This preserves the illusion of reachability while pushing spending variance very high. Some pay less than planned. Some pay double. That variance is the revenue.

Third, pity sharing. Pity accumulates across banners of the same category. Money sunk into one banner does not evaporate when a player jumps to another of the same type. The marginal cost of switching targets falls, and spending frequency rises. It is a revenue-smoothing mechanism, keeping cash flowing evenly through both launch windows and rerun windows.

Then there is rerun policy. There is no fixed rerun schedule. Some characters are absent for more than a year; others return within a few versions. That uncertainty is deliberate scarcity, exactly the limited-time-event playbook. The Chronicled Wish mechanic opens a secondary monetization lane for legacy characters, letting the publisher avoid crowding them into primary banners while still extracting money from nostalgic players.

Set against esports, the difference is in the number of links. A team earns from sponsorship, broadcast rights, in-game item revenue share, and prize money. Every one depends on a third party: sponsors, broadcasters, tournament organizers. The gacha model does not. Here there are only two links: the publisher and the payer. The publisher is the seller, the rule-maker, and the rule-announcer. That concentration of power is higher than in almost any esports ecosystem I have covered.

Based on my experience tracking and dissecting sports revenue models over eight years, one detail stands out: this machine does not depend on a sporting calendar. No season gets postponed, no broadcast-rights crisis erupts. Revenue follows the version cycle, and the publisher owns that cycle. It withstands calendar shocks better than esports, but exposes itself to an entirely different class of risk.

Three places I could be wrong

First: I am reading a schedule explainer, not a power analysis. The source material carries not one line of data on kits, stats, or character strength. It answers when and who, not whether. A piece like that cannot support any conclusion about the value of any banner, and I will not pretend otherwise.

Second: source reliability. Most of the character names and version numbers I cited above cannot be cross-verified against known game state. This is very likely rumour-based content, or worse, machine-generated content. I wrote this piece because the monetization architecture is real and worth studying. I do not vouch for the individual names.

Third: legal risk. Probability-disclosure rules and minor-protection regulation are tightening across several markets. A machine that can be rewritten by a statute is not as durable as a broadcast-rights contract. That structural weakness is something anyone planning to copy the model must price in first.

And here is where I doubt myself: I am describing the efficiency of a machine I do not want to see inside esports. Efficiency is not the same as ethics. Esports is not the future. It is the present pretending hard to be the future, and sometimes the thing it is pretending to be looks cleaner than the thing it is about to become.

What I am waiting to verify

My prediction, and it is testable: within 24 months, at least one regional-tier esports property will attempt to graft a guaranteed-floor mechanism onto its in-game item sales, a maximum spending threshold for a randomised reward. If that happens, the machine I just dissected gets a second copy wearing an esports name. Then the question stops being who wins the title. The question becomes who is paying for the finals ticket, and whether they know what they are paying for.