Hamster Kombat: 300 million players, a 60-billion-token airdrop, and the month they left

Three hundred million players, one promise

The game was a button. You opened a Telegram mini-app, tapped a cartoon hamster, and a coin counter went up; cards and “combos” raised the rate; referrals raised it more. The in-game coins were explicitly not the token. The token, HMSTR, would come later, and the whitepaper published on 30 July 2024 said what would come with it: 60% of the supply to players in an airdrop, the rest reserved for liquidity, partnerships, grants and squads. That same day the makers said the game had 300 million players. Pavel Durov had put it at 239 million on 5 July and called it an internet phenomenon.

Read those two facts together and the game is clear even before the token exists. Three hundred million accounts are competing for a fixed 60% of a fixed supply. Every tap raises your own count and, at the same time, the denominator everyone is divided by. Nobody could compute what an hour of tapping was worth, because the two numbers that fix it, the token’s price and the formula converting coins to tokens, did not exist yet. People played anyway, for months, on the promise that the number would be large.

Player counts are the project’s own, read from its mini-app and its announcements; nobody outside the project audited them, and BitPinas cites an August peak of 400 million that the others do not. Decrypt’s eligibility figure is “approximately 129 million”; ForkLog and BitPinas report 131 million. The listing-day price is reported as $0.011 by ForkLog and $0.012 by Cointelegraph.

What 60 billion tokens are worth, divided by 131 million

The airdrop was described in advance as the biggest in history, and by headcount it may have been. By value per head it was something else. Sixty percent of a 100 billion supply is 60 billion tokens. Spread evenly over the 131 million accounts that qualified, that is about 458 tokens each. At the listing-day price of $0.011 that is five dollars; by the end of the first day, under four; at the 1 November low, $1.15. The distribution was not even, because the formula rewarded coin balances, referrals and card purchases, so most players got less than the average and a few got much more. Decrypt reported players receiving ten dollars or less and calling it dust.

Average share = 60 billion ÷ 131 million ≈ 458 tokens. “30 hours” assumes ten minutes a day for the six months between launch and airdrop; a player who tapped more earned a lower hourly rate, not a higher one, because the pool was fixed. 11.25% of the allocation was vested to July 2025 and is ignored here.

This is the arithmetic of a fixed pool. A casino game pays each bet by a rule that does not depend on how many other people are playing; a fixed pool pays each player a share that shrinks with every new entrant. Growth, the thing the project advertised daily, was the thing that made each player’s expected payout smaller. Three hundred million was not a sign that the reward would be large. It was the denominator.

The price did the rest. A token whose entire reason to exist is to be handed to 131 million people on one day meets 131 million potential sellers on that day. The 30% fall in the first 24 hours is not a mystery and does not need a villain; it is what happens when supply arrives before demand. The developers said the drop was expected because the game runs in seasons, announced 15 billion more tokens for a second season and a second airdrop for summer 2025, and gave Decrypt an end-of-October date for the new season that passed without a launch.

The month they left, and what a counter can tell you

The useful number in this story is not the price. It is the player counter on the mini-app, which anyone could read. Over 100 million monthly players in late September, by Decrypt’s reading; under 44 million by the start of November; 40.9 million on ForkLog’s reading a few days later; 38 million on 6 November. Against the summer figure of 300 million, that is a fall of about 86% in six weeks. Against the late-September monthly figure, 56%. Either way, the people who stayed through the airdrop left in the month after it, as soon as the payout they had been playing for became a number.

Note what that curve is not. It is not a hack, a rug or a crime; the token was delivered as described, the exchanges listed it, the bans were for cheating. It is the shape of a game whose payout is decided after you play, by the issuer, at a price set by the people receiving it. For as long as the payout is unknown, every hour has an imagined value; the moment it is known, every hour has a real one, and the real one was a few cents.

That is the contrast, and it is worth stating without flattering ourselves. A provably fair original is also a game that pays in money for time, and it also has a cost per hour: in the pricing article it is computed as edge × stake × hands per hour, a number you can read before the first roll. What it does not have is a denominator. The multiplier on a dice roll is the same whether a thousand people are playing or a hundred million, the result is fixed by a seed committed before the bet, and the house cannot decide after the fact what the round was worth. The honest comparison is this: in one game you can price an hour before you spend it, in the other you find out six months later, and the player counter shows what people did when they did.