Crypto play-to-earn games: the full research

The research was done by AI agents that open web pages. Some sites refuse them; where that happened we say so. The small numbers point to the source list at the end. Where a figure is our own sum or guess, it is marked “our estimate” and the basis is given.

The claim

Our collection for this scheme holds one tweet, not the usual dozens¹. It was posted at the end of July 2026 by a small account and had a few thousand views when collected¹. Our rule is to name an account only if it has about 10,000 followers or sells something tied to the claim, with a source. This account meets neither test, so we do not name or quote it. We paraphrase instead. We could not establish whether the account has any connection to the game.

The tweet says the author made $9 in one video of a game while risking $3, that the game is fun and runs in a web browser, that players are paid for beating other players, and that the author could do it all day. It tags the game, Trenches¹. The tweet uses the word “clip” twice. In shooter slang, to clip someone is to shoot them; a clip is also a short video. We did not watch the video attached to the tweet, and we did not view any replies or thread under it.

With one tweet, this page cannot rest on what many accounts say. It rests on how this kind of game is promoted generally: on the game that tweet promotes, read from the game’s own pages and public data, and on the record of earlier “play-to-earn” games. We did not research the token games that are running today, such as Pixels or the current version of Axie Infinity. The page judges the scheme, not the person who tweeted.

What the scheme is

“Play-to-earn” covers two different things. They pay in different ways.

Wager games. You deposit money, play against other people, and the winner takes the loser’s stake. The game in the tweet is this kind. Trenches is a first-person shooter that runs in a browser. Players deposit SOL, the coin of the Solana network. Each room has a fixed stake per kill. The game’s own page says: “Drop an operator and their stake moves to your balance the moment they hit the dirt … Die, and whoever dropped you collects yours.”²

So the money a winner takes is money another player lost. The game adds none. The company running it (the “house”) keeps a cut, called a rake. Because of the rake, players as a group must lose money. This is a bet on your own skill against other people, not a wage.

Token-reward games. You play, and the game pays you in its own token, a digital coin the game creates. The token is worth whatever buyers will pay for it. Axie Infinity (2021) and the tap-to-earn games on Telegram (2024) were this kind. No one stakes money against another player, but in some of them you had to buy game items first.

Trenches also has a token, $TRENCH, and two other ways to be paid besides winning fights: posting video clips of the game, and referring new players²¹⁴. We cover all three.

Is it real money? The evidence mostly says yes, with one loose end.

The landing page still contains a notice that says: “Deposits & withdrawals aren’t live yet — we’re still running on devnet.”² A devnet is a test network where the coins are worth nothing.
The game’s wallet script says that notice has been switched off: “LIVE / mainnet: wallet opens directly for everyone (real deposits + withdrawals).”⁵⁹ Mainnet is the real Solana network. A wallet here means the account that holds your coins.
The site’s settings file and its status address both name the real Solana network and a production (live) environment⁷⁶¹. The tournament feed also names the real network¹³.
Against that, one line in the landing page still sets the network name to the test network when the site’s newer settings are present, and they are present². With those settings present the page also does not load the older wallet and staking scripts we read; it loads a replacement script that we did not read².

The 50 accounts on the game’s profit board also exist on the real network, with transactions from late July 2026 onward⁶³. We did not deposit or withdraw, so we cannot confirm it first-hand.

The arithmetic

The tweet gives one session: $9 made, $3 risked. To turn that into an income you need the stake, the house cut, how often the player wins, and how many fights there are. The tweet gives none of these. The game’s pages give most of them.

Step 1: the stake

The room-creation script offers four preset stakes per kill: $0.05, $0.25, $1 and $5³. The form also accepts any custom stake³. That fits the landing page, which speaks of rooms from “a $1 trench” to “a $49 one”². The smallest deposit is 0.01 SOL². SOL traded at $109.77 when we read it, so that is about $1.10 (our estimate: 0.01 x $109.77)¹⁷.

Step 2: the house cut

The site’s scripts state a 4% cut of each kill’s stake. One says: “every kill charges a 4% house rake”⁴. Another carries a fallback setting of 40 per thousand, which is 4%⁵. That script says the fallback “only drives display, never a transfer”, and the live page may not load it at all (see above)²⁵. The room script is built to show “winner keeps” 100% minus a rake figure the game sends while running³. We did not see the live value.

The landing page words it differently: “A flat commission is taken when you enter the arena — kills, extractions and withdrawals are never taxed.”² These two descriptions do not match. The money logic runs on the game’s servers where we cannot see it, so we could not settle which is right. We use 4% per kill below, because it is the figure the code states.

Step 3: how much you can lose

The page says: “You can never lose more than 10× the room’s stake in one session: in a $1 room the worst run costs $10, in a $49 room it’s $490.”² Nothing we read limits the number of sessions.

The session limits at the four preset stakes would be $0.50, $2.50, $10 and $50 (our estimate: ten times each preset)³. The tweet’s “$3 risked” matches none of them. But a custom stake of $0.30 would cap a session at $3, and three deaths in a $1 room would also cost $3 (our estimate)²³. The tweet does not say which.

Step 4: what a kill is worth

With a 4% cut, each kill pays you 96% of the stake. Each death costs you the full stake. So over many fights:

Profit per kill = stake x (0.96 - deaths per kill)

This is our estimate, built from the 4% rake⁴⁵.

A player with one kill per death loses 4% of the stake on every kill.
Break-even needs just over 1.04 kills per death (1 / 0.96).
A player with two kills per death keeps 46% of the stake per kill (0.96 - 0.5).

Step 5: the volume needed

The tweet names no monthly figure. We use $1,000 a month as a modest test of “income”. The table shows the kills needed in a month, at three skill levels. All figures are our estimate from the formula above.

Stake per kill	1.5 kills per death	2 kills per death	3 kills per death
$0.05	about 68,200	about 43,500	about 31,900
$0.25	about 13,600	about 8,700	about 6,400
$1	about 3,400	about 2,170	about 1,600
$5	about 680	about 435	about 320
$49 (custom)	about 70	about 44	about 33

The $49 row is the landing page’s own example of a high-stake room². It is not one of the four presets, and we did not see such a room in use.

For $10,000 a month, multiply by ten: about 21,700 kills in $1 rooms at two kills per death.

Two things limit these numbers. First, every kill needs an opponent who stakes the same amount and loses. Second, the higher the stake, the fewer and better the opponents are likely to be. We found no data on how many players use the higher-stake rooms.

Step 6: check against the game’s own records

Trenches publishes leaderboards through a public data feed. We read them on 10 October 2026.

Profit. The site’s code describes the main profit board as covering “the DEPLOY $0.05 lobby”, ranked by “net staked SOL”⁶. DEPLOY is the code’s name for the main mode, and the site’s settings have a five-cent mode switched on⁶⁰. So this board appears to show the lowest preset stake only. On it, the best all-time result of any player is 0.27148 SOL⁹. That is about $29.80 (our estimate: 0.27148 x $109.77)¹⁷. At a five-cent stake that is a gain of roughly 540 stakes (our estimate: 0.27148 / 0.0005 SOL), a strong result for the stake³⁹. Tenth place is about $5.17, 25th about $0.70 and 50th about $0.06⁹. The 50 accounts listed are up 1.65 SOL between them, about $181⁹. Five are at 0.1 SOL or more⁹.

Kills. The account with the most kills all-time has 10,251¹⁰. The top 50 accounts by kills total 63,101 kills; the middle account among them has 658¹⁰. But the kills feed returns the same list whatever mode is asked for, so it mixes staked and free play¹⁰. The room script describes free rooms as “no stake, bots fill”, meaning computer-run opponents fill the empty places³. So we cannot say how many of those kills were staked or were against people.

Like for like. $1,000 in one month at the five-cent stake needs about 43,500 kills at two kills per death (the table above). The most active account has 10,251 kills of all kinds since launch, an unknown share of them unstaked¹⁰. That account’s profit on the five-cent board is about $3.09, but because its kills are mixed, the two figures cannot be read as profit per staked kill⁹¹⁰. Six of the ten accounts with the most kills are not in the top 50 by profit⁹¹⁰.

Recent play. The “this week” and “today” profit boards returned no rows¹¹. The weekly kills board listed 10 accounts with 833 kills between them, and the daily kills board one account with 2 kills⁶². So people played this week. Either none of it was staked in the five-cent lobby, or the profit board’s date filter is broken. We could not tell which.

When the board’s accounts were last active. We looked up each of the 50 profit-board accounts on the real Solana network. All 50 exist there. For 45 of them the latest transaction falls between 25 July and 12 August 2026; 13 of those are dated 31 July. The other five are dated 30 August to 21 September, and none is later⁶³. The game holds players’ money itself, so these transactions are most likely deposits and withdrawals, not single fights. Our reading (an estimate): real-money accounts were active at the end of July, most of the board’s profit was made in a few weeks around the tournament, and staked play has been thin since mid-August. That fits the empty weekly profit board.

One-on-one duels. The duel board lists five players all-time. Three are ahead, the best by 0.105 SOL (about $11.50). Two are behind. The five results add up to minus 0.0056 SOL¹². That small shortfall is what a rake looks like: winners’ gains match losers’ losses, less the house cut.

Limits of these boards. The site’s code says of its boards: “Everyone is shown: net-losing players rank below the winners (in red), not hidden.”⁶ The duel board does show losing players¹². But the public feed for the main board stops at 50 rows and all 50 are positive, so the losing accounts fall below the cut⁹. At least 50 accounts are ahead; we cannot see how many are behind or by how much. Results in $0.25, $1, $5 and custom rooms are not on any board we found, so the best result in the game could be higher. The figures are the operator’s own records. We did not reconcile them with transfers on the Solana network.

Step 7: the clip route

Trenches runs a campaign that pays people to post gameplay videos on X, TikTok or Instagram¹⁴. Payment is in $TRENCH. The campaign page says: “Every approved clip $2.80 just for posting 100,000 $TRENCH · paid on approval”¹⁴. Bonuses follow by views: 100,000 more tokens at 500 views, 300,000 more at 5,000 views and 10,000,000 more at 50,000 views¹⁴. Clips are approved by hand¹⁴. The campaign page is undated. We do not know whether the campaign was running at the end of July 2026, or whether the tweet in our collection was submitted to it or paid.

The page’s dollar figures assume the whole token is worth $28,000¹⁴. At the price we read today, $0.00003683 per token¹⁵:

One approved clip pays about $3.68 (our estimate: 100,000 x $0.00003683).
A clip that passes 50,000 views would pay about $387 in all (our estimate: 10,500,000 tokens).
$1,000 a month from the base payment alone needs about 272 approved clips, around nine a day (our estimate).

The catch is the token. $TRENCH had a total market value of $36,554, with $20,942 in its trading pool, the money available to buy the token from sellers¹⁵. About $620 to $630 traded in 24 hours, in six trades, in two readings on 10 October 2026¹⁵. The total supply is about 992.4 million tokens¹⁵. So the 10,000,000-token bonus is about 1% of all tokens that exist (our estimate). With six trades a day, a few people selling their clip rewards would push the price down. A dollar figure on the campaign page is not the same as dollars you can take out.

We could not see how many clips have been approved or paid. The campaign page shows that only after a crypto wallet is connected¹⁴.

Step 8: the referral route

A referrer earns a share of the house cut on kills made by people they bring in. The landing page says: “you earn a live slice of the kill rake — paid in real SOL, for as long as they keep fighting.”² The script says the default is half the rake, and that it can be changed⁴. Half of 4% is 2% of each stake.

For $1,000 a month a referrer needs recruits to win $50,000 of kill stakes a month (our estimate: $1,000 / 0.02). In $1 rooms that is 50,000 kills. At the five-cent stake it is 1,000,000 kills. The whole top 50 on the kills board totals 63,101 kills since launch, and staked kills are fewer than that by an unknown amount¹⁰.

What people who tried it report

Trenches itself

We found no independent player of Trenches who published tracked results. Two searches returned nothing usable. Every figure above comes from the operator’s own pages and data feed.

The one tournament on record had 41 entrants¹³. It was created on 22 July 2026 and its finals deadline passed on 30 July 2026, though the feed still labels its stage “finals”¹³. Entry cost 5,000 $TRENCH, with 10,000 more from each of 20 finalists and a 5% house cut¹³. Fees were at most 405,000 tokens, about $15 at today’s price (our estimate)¹³¹⁵. The feed shows 7 finalists paid, which makes 275,000 tokens, about $10 (our estimate)¹³¹⁵. Only 5 of the 41 entrants scored any points¹³. That suggests paying players numbered in the tens at the time.

The landing page shows “341 in the trenches right now”². That number is written into the page itself, and the scripts we read do not update it. One script carries a note marked “PLACEHOLDER until wired to trenches-api”, saying the “footer field-report stats” are display data that is not yet live⁶. We did not prove the 341 counter is one of those, and we did not read every script. We report it as seen: the figure could not be confirmed as live.

Other shooters that pay per kill

Ev.io (2023). A Solana browser shooter that paid per kill with no stake. A Decrypt reviewer bought an item for about $22 to qualify and earned 124 units of the game’s currency in an hour, about six cents³⁶. This is one reviewer’s own account, from 2023.
Bonk Arena (2025). The same stake-per-kill model as Trenches. Players paid a fee to enter and took the stake of players they killed, “with a variable service fee from 0% to 10% applied”³⁷. We found no player results for it.

We found no published data for any stake-per-kill crypto shooter on what share of players end ahead.

Skill games with a house cut, in general

The best measured comparison we found is daily fantasy sports, a paid contest of skill. McKinsey analysed operators’ data for the first half of the 2015 baseball season and found “91 percent of DFS player profits were won by just 1.3 percent of players”³⁸. The top 11 players paid $2 million each in entry fees and made $135,000 each on average. The rest of that top 1.3% paid $9,100 and made $2,400 each. The 80% it called “minnows” lost $25 on $49 of fees³⁸. It is a different game and an old study. We read it on McKinsey’s German site because the main site refused our agents.

A poker coach estimates that “only around 30% of poker players are winners over the long term”⁴¹. He gives no sample size and sells poker courses, so treat it as a rough guide only.

Where every player’s result can be counted on a public blockchain (the shared record of crypto transfers), the picture for chance-based products is clear. Bitquery counted all wallets that bought a card-prize product on Solana: “17,544 wallets bought pulls during the period. Of those, 3,895 got more money back than they spent. That is 22 percent.” The middle wallet was down $50³⁹. A study in PLOS ONE of 2.2 million bets on Ethereum gambling apps found “modest losses for the typical player”, a median of 5.3% of the amount bet⁴⁰. These are games of chance, not skill, so they are comparisons only.

Token-reward games: Axie Infinity

Axie Infinity is the best-documented case. Players collected creatures called Axies and earned a token called SLP. Many players could not afford the creatures, so owners (“managers” or guilds) lent them out and took a share. The borrowers were called scholars.

It did pay, for some months. The largest guild, Yield Guild Games, reported that by the end of July 2021 its scholars had earned “a lifetime gross revenue of approximately USD $8.9M”, with 70% going to the scholar²⁵. It had more than 4,000 scholars and had farmed 31 million SLP²⁵. That is roughly $2,200 gross and $1,550 to the player each (our estimate: $8.9 million / 4,000, then 70%), spread unevenly because they joined at different times. These are the guild’s own figures.
Buying in. An analysis of public blockchain records by Covalent put the middle wallet’s spend on creatures at $867 and average token claims at “about $192 per wallet per month”, about 4.5 months to earn the cost back at 2021 prices²⁶. The $192 is an average, not a middle value. The same analysis found the middle deposit was $820 and the middle withdrawal $137²⁶. We read an archived copy; the live page no longer shows the article.
By November 2021. The research firm Naavik wrote that “scholars are very close to realizing that playing Axie isn’t so different from a minimum wage job”, using a Philippine daily minimum wage of $7.03 and an entry cost of about $350²¹. One of the report’s authors wrote on Twitter, as quoted by Forkast, that the daily earnings of the typical player “have fallen below the Philippines’ minimum wage line for all but the high-ranking players”²². Naavik’s figures are a model built from the game’s reward limits and the token price, not measured payouts.
By 2022. Rest of World reported the guild’s scholars at “around $20 a month” in late April 2022, against player reports of up to $2,000 a month at the peak²⁴. A guild owner told Forkast in July 2022 that a scholar “is probably only making about US$30 to US$60 per month now”, down from $300 to $500²⁹. By May 2022 all the SLP the guild had ever farmed, 195 million tokens, was “worth about $1.1 million” at that day’s price²⁸. The first 31 million alone had been valued at about $8.9 million when earned²⁵.
By 2023. The Thomson Reuters Foundation reported one manager’s scholars had earned 450 pesos a day against a 470-peso local minimum wage, before the token fell 99%²³. Daily users fell from 2.7 million to about 250,000, according to a crypto-gaming site the article cites²³.

None of the articles we opened carried a reply from Sky Mavis, the company behind Axie, to the minimum-wage findings. We did not find one.

Token-reward games: tap-to-earn

These were free games in the Telegram chat app where players tapped the screen to collect points, later swapped for tokens. An airdrop is a free handout of tokens.

Notcoin (2024). Decrypt reported that the token “now has 11.5 million holders, according to the Notcoin team”, and that “At least 2.5 million of them are on-chain holders”, from a game with 35 million players³¹. The token was valued at $1.8 billion a month after launch³¹. As a mean, that is about $51 per player or about $157 per holder (our estimate: $1.8 billion / 35 million, and / 11.5 million). Both are upper-end guides: the value covers the whole supply, not only tokens handed to players, and holders include people who bought the token. One writer’s informal survey put Notcoin payouts at $100 to $200 “for many ordinary drop participants”³⁵. We found no published middle value.
Hamster Kombat (2024). The team said that of more than 300 million players, “131 million qualified for the airdrop”³³. So more than half received nothing. Decrypt reported that “many users complained about small allocations worth $10 or less”³². The same writer’s informal survey put payouts at $5 to $20 for people who had not recruited others, and $100 to $200 for those who had³⁵. The token’s value peaked at about $646 million³². As a mean, that is about $2 per player or about $5 per qualified player (our estimate: $646 million / 300 million, and / 131 million).

Most games do not last

CoinGecko counted the games: “Around 2,127 web3 games have failed in the last five years since the GameFi niche emerged, representing 75.5% of the 2,817 web3 games launched.”¹⁸ “Web3 games” and “GameFi” are trade names for crypto games. “Failed” means active users fell 99% or more from the peak¹⁸. DappRadar reported “Over 300 gaming dapps went inactive this quarter” for April to June 2025; a dapp is an app that runs on a blockchain¹⁹. CoinDesk, reporting a study by the trading firm Caladan in April 2026, wrote that “roughly 93% of so-called GameFi projects are now effectively dead, with token values down about 95% from their 2022 peaks”²⁰. We did not open the Caladan study itself.

The tokens tell the same story. Read today, the Axie reward token SLP was 99.8% below its peak, Notcoin 98.3% and Hamster Kombat 97.7%¹⁷.

What the rules allow now

The game’s own terms

We found no terms of service on the Trenches site. Four guessed addresses for a terms page returned “not found”, and the only legal page linked is a privacy policy²⁸. We found no named company, no age limit, no list of restricted countries and no gambling licence on the pages we read²⁸. Terms may appear after sign-in, which we did not test.

The privacy policy says the game holds the keys to wallets it creates: “Wallets created and managed by Trenches require us to protect their signing keys.”⁸ Whoever holds a wallet’s key controls the money in it. The policy also says: “Automated security and fair-play checks may restrict gameplay, place withdrawals on hold or flag an account for review,” and that the policy “does not guarantee that every restriction will be removed.”⁸ So a deposit carries the risk of a held withdrawal on top of the risk of losing fights.

Computer-run opponents. The room script shows that free rooms are filled with computer-run opponents (“no stake, bots fill”), that rooms containing them are marked “+bots” in the room list, and that one-on-one duels are described as “private duel, no bots”³. We could not see whether staked rooms with more than two players can contain them, or what a kill of one pays.

Lowering the cut. A staking script says a player can lower the 4% cut by depositing $TRENCH with the game for a time. It lists steps from 1 million tokens for a 20% discount up to 50 million tokens for no cut at all, with a seven-day wait to get the tokens back⁵. At today’s price that is about $37 to about $1,840 (our estimate)¹⁵. Fifty million tokens is about 5% of all that exist (our estimate)¹⁵. The live page does not load this script when its newer settings are present, so these steps may not be what the live game uses². The site also describes a one-time fee in $TRENCH to unlock paid play, skins bought by destroying tokens, and tournament entry in tokens²⁵¹³. We could not see the size of the unlock fee.

App stores and platforms

Trenches runs in the browser, outside the app stores. The stores’ rules help explain why games of this kind do.

Google Play: “Google Play generally prohibits apps from facilitating real-money gambling and gaming.” Its rules for blockchain apps add: “You may not promote or glamorize any potential earning from playing or trading activities.”⁴⁷
Apple: real-money gaming apps “must have necessary licensing and permissions in the locations where the app is used, must be geo-restricted to those locations”⁴⁸.
Steam, the largest PC game store, does not accept “Applications built on blockchain technology that issue or allow exchange of cryptocurrencies or NFTs.”⁴⁹ An NFT is a tradable digital item recorded on a blockchain.
Telegram requires the small apps that run inside it to use only its chosen blockchain, called TON, with a deadline that passed in February 2025⁵⁰. A platform can change which tokens a game may pay in.

The US games ratings body gave another Solana shooter with small skill bets an “Adults Only” rating “due to its real money wagering elements”⁵⁸.

The law

We are not saying Trenches or its players break any law. We did not establish who operates it or where, and terms may be shown after sign-in.

Whether staking money on your own skill is legal depends on where you live. We read some statutes. We did not read court rulings on how they apply to video games, and we found no regulator action against Trenches or any similar stake-per-kill game.

Florida: “Whoever stakes, bets, or wagers any money or other thing of value upon the result of any trial or contest of skill … commits a felony of the third degree”⁵¹. A felony is a serious crime.
Washington State: knowingly sending or receiving “gambling information” over the internet is a felony⁵³. We did not check whether a staked shooter falls under the state’s definition of gambling.
Great Britain: a “game of chance” includes “a game that involves both an element of chance and an element of skill”⁵⁴. Whether a shooter has an element of chance is untested in what we read.
South Korea: a law firm’s note says the games regulator “has consistently refused to rate P2E games, thereby effectively banning P2E games for distribution in Korea”, and a Seoul court upheld a refusal in January 2023⁵⁵. P2E is short for play-to-earn. We did not check the position in 2026.

A US gaming lawyer’s overview from 2020 says that in the restrictive states “a head to head bet on the outcome of a game is generally prohibited”, while “paying to enter a tournament with pre-announced prizes may be lawful”⁵². It is six years old and is not a regulator’s view.

Tax

In the United States, the tax authority says “Gambling winnings are fully taxable and you must report the income on your tax return,” and that losses can be deducted only up to the amount of winnings⁵⁷. It also says that selling crypto creates a taxable gain or loss⁵⁶. Whether stake-per-kill winnings count as gambling income is our reading, not a statement by the tax authority.

Regulator and court cases in the wider field

Each case below concerns only the companies and people named in it. None concerns Trenches.

FBI warning, March 2023. “The FBI warns of criminals creating fake gaming applications (apps) to steal millions of dollars in cryptocurrency. Criminals advertise the apps as play-to-earn games offering financial incentives to players.”⁴³ This is about copies and imitations, not about any named game. It is a reason to check a link from a tweet before connecting a wallet.
Skillz v. Papaya Gaming, 2026. Papaya runs paid skill games on phones; it is not a crypto game. The case was about whether its advertising implied that players faced people when some opponents were computer-run. A law firm’s note says a New York federal jury in April 2026 “returned a verdict in favor of Skillz”, finding Papaya should pay $420 million in damages. Its update says that in July a federal judge ordered Papaya “to disgorge $719 million and pay $10.1 million in attorney fees” on top⁴². To disgorge is to hand over gains. We could not open a statement from Papaya and do not know whether it has appealed. The case matters here only as a reminder to ask who the opponents are; what Trenches’ own script shows about computer-run opponents is set out above.
Axie’s network, March 2022. The FBI said North Korean groups “are responsible for the theft of $620 million in Ethereum reported on March 29”⁴⁴. Rest of World identifies this as the theft from Ronin, the network that carried Axie players’ funds²⁴. Money held by a game’s systems can be lost wholesale.

One more event, not a regulator case. Before its token handout, the Hamster Kombat team said “We’ve banned over 2.3 million cheaters”, giving the example of one person who “connected over 400 accounts” to one address³⁴. Decrypt reported that 6.8 billion tokens were removed from those accounts, half to be given to other players and the rest destroyed³⁴. The point for a player: the operator alone decides, at the end, who qualifies.

Who makes money from it

In the game the tweet promotes, the pages show five kinds of party that are paid, whoever wins a given fight.

The house. The scripts state a 4% cut of every kill⁴⁵.
Referrers. By default, half of that cut goes to whoever referred the player: “a cut of the fee on every kill they make — paid straight to your balance, forever. No cap.”⁴ A person with a referral link earns on every kill a recruit makes, whether or not the recruit ends up ahead overall.
Clip posters. Posting a clip is paid work under the campaign¹⁴. So a video of someone winning can be a paid post, and it is not a sample of how players do. We do not know whether the tweet in our collection was submitted to the campaign or paid.
The token’s creator and the launch site. $TRENCH was launched on pump.fun, a site for creating tokens¹⁵. That site’s fee page says: “On every trade the creator receives some portion of the total fees.” At this token’s size the creator’s fee is 0.300% and the site’s own fee 0.930% of each trade¹⁶. We did not find out who created the token or how much has been collected.
Better players. Their winnings come from weaker players, as the duel board shows¹².

The only collected tweet promotes the game itself, not a course, a paid group or a newsletter¹. We found none of those around this game.

Earlier games paid the same kinds of parties. Axie’s marketplace took 4.25% of each sale, raised to 5.25% in March 2022³⁰. Guilds kept 30% of what their scholars earned: “70% going to the player, 20% going to their community manager, and 10% to YGG”²⁹. One manager told the Thomson Reuters Foundation he “earned as much as 600,000 pesos a day from the game by taking a 60% cut of his players’ income”, with 200 players²³. That is about $10,800 at the article’s exchange rate (our estimate). It is a peak day by his own account, and it does not fit the 450 pesos a day the same article gives for his players: at that rate 200 players would yield about 135,000 pesos a day (our estimate, taking 450 pesos as the players’ 40%)²³. Another manager told Forkast in 2021 that he made $20,000 a month at the peak from the players he managed, also his own account²².

A study of all 9.4 million Axie addresses sorted them by behaviour. Its authors infer that 99.5007% of addresses, including the barely active ones, were scholars playing with borrowed Axies, while “the remaining 0.4993% of addresses have the most substantial control over Axies in the game”²⁷. About 70% of addresses showed minimal activity and rarely claimed tokens²⁷.

The upside

Some people have been paid. The cases differ a great deal in size and in how well they are checked.

In Trenches. The game’s profit board is checkable evidence that some players are ahead. At least 50 accounts show a net gain; the feed stops at 50⁹. The best result is about $30, the middle of those top 50 is about $0.67 (our estimate: 0.00615 SOL x $109.77), and all 50 together are up about $181⁹¹⁷. This is the five-cent lobby, where $30 is a gain of roughly 540 stakes³⁶⁹. Most of it appears to have been won within a few weeks in July and August 2026, not over a long period (our estimate)⁶³. Results at higher stakes are not published. These are the operator’s records, not independently verified. We found no documented case of a player earning more than this from the game.

In skill contests with a house cut. McKinsey’s fantasy sports data shows what the top looks like where a market is large: 11 players averaging $135,000 profit in half a season on $2 million of entry fees each, and the rest of the top 1.3% averaging $2,400 on $9,100³⁸. Those results came from operators’ data, not from the players’ own claims. They needed a market with many thousands of weaker paying players. The Trenches boards do not show such a market⁹¹²⁶².

In Axie, 2021. This is the best case for ordinary players. Scholars put in no money. By the guild’s own figures they received very roughly $1,550 each over the first months (our estimate)²⁵. Players reported up to $2,000 a month at the peak, by their own account²⁴. High-ranked players stayed above the Philippine minimum wage after typical players fell below it²². The address study found that among those who kept playing, “it is relatively fair for players to compete for rewards” whether or not they were rich in Axies²⁷. The window was roughly mid to late 2021.

In free tap games, 2024. Notcoin players paid nothing. The mean was on the order of $50 per player if sold in the first month, more for the most active (our estimate from the team’s own counts)³¹. We found no published middle value. The far larger game that followed left many with $10 or less³².

The largest sums went to people who owned the game assets and took a share from many players. The $10,800-a-day and $20,000-a-month figures above are both the managers’ own statements²²²³.

A realistic good result, first year (our estimate, not sourced). This estimate is drawn from one small game. In a stake-per-kill shooter with a 4% cut and a small pool of opponents, a strong player at the lowest stake might end a few dollars to a few tens of dollars ahead in total. That matches the top of the Trenches five-cent board⁹. The sum scales with the stake and with the number of weaker opponents willing to play at it, and we have no data on either at higher stakes. An average player loses slowly, at about 4% of each stake. In a free token game, a competent player might get one small payout if the game succeeds and nothing if it does not. We did not research the token games running today. In the searches we did run, we found no checkable report of a player earning a continuing income from a crypto game in 2025 or 2026.

What it takes to compete

Skill well above the room. In a wager game the only source of profit is being better than the people you meet. Break-even is just over 1.04 kills per death; a modest profit needs clearly more⁴⁵. In fantasy sports the profit went to about one player in a hundred³⁸.

Enough weaker opponents. Skill is worth nothing without people to win from. The Trenches duel board lists five players, and the five-cent board recorded no staked profit this week¹¹¹²⁶³. This is the tightest limit we found. No amount of skill can win more than the other players bring.

Volume. The top fantasy sports players put through $2 million in fees each³⁸. In Trenches, $1,000 a month means thousands of kills a month at $1 stakes and tens of thousands at five cents (our estimate, Step 5).

Timing, in token games. Those who did well in Axie and the tap games were early, in the one game that grew, and turned tokens into money before the price fell²⁴²⁵³¹. All the tokens we checked are down 97% or more from their peaks¹⁷.

Capital and people, for the largest sums. The big Axie incomes belonged to owners running many players²²²³. Entry cost each player-seat $350 to $867 at different times in 2021²¹²⁶.

An audience, for the referral and clip routes. Both pay for bringing in or reaching other people, not for playing well⁴¹⁴.

What it costs. In Trenches: a deposit from about $1.10, up to ten times the stake per session, the 4% cut, and for those who want a lower cut, money tied up in the game’s token²⁵¹⁷.

How to tell early which side you are on.

Count your kills and deaths over a few hundred staked fights at the lowest stake. Under about 1.04 kills per death, you are paying the others. This follows from the 4% cut⁴⁵.
Look at the public profit board. It shows the size of the lowest-stake market only: tens of dollars at best, whatever a video shows⁶⁹. Higher-stake results are not published.
Check that the weekly profit board has entries. No recorded staked profit this week means few people to win from¹¹.
Look for the “+bots” mark on a room before staking in it³.
Test a small withdrawal before depositing more. The game holds the keys and can hold withdrawals⁸.
Value token payments at what the token’s daily trading can absorb, not at the figure on the campaign page¹⁴¹⁵.
Check whether staking money on a game of skill is lawful where you live⁵¹⁵²⁵⁴.

What we did not verify

The tweet. One agent confirmed that it exists through the public feed X provides for showing tweets on other sites; X itself refused our agents. We did not watch the video and did not see replies. We could not establish whether the account has any connection to the game.
The session in the tweet. We do not know the stake, kills, deaths or time played.
Whether real-money play is live. Most of what we read says it is, but one line in the landing page says otherwise, and we did not read the script the live page loads in place of the older wallet script²⁷⁵⁹⁶¹. We did not deposit, play or withdraw.
The fee actually charged. The 4% comes from script comments and a display-only fallback setting in a script the live page may not load. The landing page describes a flat fee on entry instead²⁴⁵. The live values need a login.
The full spread of results. The profit feed returns 50 rows, all positive. The number of losing players, the middle player’s result, total deposits and total house takings are unknown. The board appears to cover only the $0.05 lobby⁶⁹.
Staked and unstaked kills. The kills board mixes them, and may include kills of computer-run opponents³¹⁰. Whether staked rooms for more than two players can contain such opponents was not established.
The operator’s records against the blockchain. We checked only the date of each board account’s latest transaction⁶³. We did not reconcile board figures with transfers, and the dates do not show individual fights.
The “341” counter and why the weekly and daily profit boards were empty while the kills boards were not.
Who runs Trenches, where, and with what licence, if any. Whether terms, an age check or country blocks appear after sign-in. We did not ask the operator.
Clip payouts. How many clips were approved or paid, whether the largest bonus has ever been paid, and when the campaign began.
The token’s price history. A third-party listing seen only in a search summary puts the peak at $0.0005 on 4 August 2026, a fall of about 93% since. The page refused our agents directly and through a reader, so we do not rely on it.
Legal status of stake-per-kill play in any specific place. We read statute text for three places and no court rulings.
Exact quotes from some pages. Quotes from Google, Apple, Steam, Telegram, the IRS, the statutes and several news articles reached us through a tool that summarises pages. They should be re-opened before being relied on word for word.
Underlying studies. We did not open the Caladan study behind the CoinDesk article²⁰. Naavik’s daily earnings figures are in charts we could not read²¹. McKinsey’s study was read on a mirror site³⁸. Covalent’s was read from an archive, is undated, and one of its totals looked inconsistent²⁶.
Self-reported figures. The guild’s $8.9 million, Notcoin’s holder count, Hamster Kombat’s player counts and every manager’s income are the parties’ own statements. The tap-game survey figures are one writer’s informal poll³⁵.
Company replies. We found no Sky Mavis reply to the wage findings. We could not open a statement from Papaya or any notice of appeal.
Other games. Move-to-earn apps, Pixels, today’s Axie and other current games were not researched. With one tweet collected, other play-to-earn pitches on X may differ from this one.
A US rule limiting gambling-loss deductions to 90% from tax year 2026. One agent recalled it from memory and did not find it on the page read, so it is not used.

Sources

Does It Pay collection of 1 tweets on this scheme, collected 2026-10-06. Tweet dated to the end of July 2026 from its ID. Read from the project’s own file on 2026-10-10.
Trenches, landing page. https://www.playtrenches.com/ Page last modified 2026-10-07. Read at source on 2026-10-10.
Trenches, room-creation script. https://www.playtrenches.com/TemplateData/lobbies.js Undated. Read at source on 2026-10-10.
Trenches, referral script. https://www.playtrenches.com/TemplateData/affiliate.js Undated. Read at source on 2026-10-10.
Trenches, token staking script. https://www.playtrenches.com/TemplateData/staking.js Undated. Read at source on 2026-10-10.
Trenches, landing script. https://www.playtrenches.com/TemplateData/landing.js Undated. Read at source on 2026-10-10.
Trenches, live settings file. https://www.playtrenches.com/TemplateData/inhouse-config.js Undated; names the environment “production” and the network “mainnet-beta”. Read at source on 2026-10-10.
Trenches, privacy policy. https://www.playtrenches.com/privacy.html Undated. Read at source on 2026-10-10.
Trenches data feed, all-time profit board, main mode. https://trenches-api.vercel.app/api/leaderboard?metric=pnl&mode=deploy&period=all&limit=5000 Read at source on 2026-10-10, about 14:05 UTC.
Trenches data feed, all-time kills board (same rows for every mode). https://trenches-api.vercel.app/api/leaderboard?metric=kills&mode=deploy&period=all&limit=5000 Read at source on 2026-10-10.
Trenches data feed, weekly and daily profit boards. https://trenches-api.vercel.app/api/leaderboard?metric=pnl&mode=deploy&period=week Read at source on 2026-10-10.
Trenches data feed, duel profit board. https://trenches-api.vercel.app/api/leaderboard?metric=pnl&mode=duel&period=all&limit=200 Read at source on 2026-10-10.
Trenches data feed, tournament status. https://trenches-api.vercel.app/api/tournament/status Tournament created 2026-07-22, finals deadline 2026-07-30. Read at source on 2026-10-10.
Trenches clip campaign page. https://trench-clipping.vercel.app Undated. Read at source on 2026-10-10.
DexScreener data feed for the $TRENCH token; supply from the Solana network. https://api.dexscreener.com/latest/dex/tokens/DWQJRoA5hQBGafXCbbtNPNnNh8FfL4S9hiVabrK6pump Trading pair created 2026-06-29. Read at source on 2026-10-10, at about 14:05 and 14:20 UTC.
pump.fun, fees page. https://pump.fun/docs/fees Updated 2026-10-08. Read at source on 2026-10-10 through the r.jina.ai reader.
CoinGecko data feed: SOL price and token falls from peak. https://api.coingecko.com/api/v3/coins/markets Read at source on 2026-10-10.
CoinGecko, “How many GameFi are dead”. https://www.coingecko.com/research/publications/how-many-gamefi-dead 2023-11-29, page marked updated 2026-04-30. Read at source on 2026-10-10 through the r.jina.ai reader; the direct route was refused.
DappRadar, “State of Blockchain Gaming in Q2 2025”. https://dappradar.com/blog/state-of-blockchain-gaming-in-q2-2025 2025-07-10. Read at source on 2026-10-10 through the r.jina.ai reader.
CoinDesk on the Caladan study of web3 games. https://www.coindesk.com/markets/2026/04/23/more-than-90-of-web3-games-failed-after-usd15-billion-boom-as-gamers-never-showed-up-caladan 2026-04-23. Secondary; read on 2026-10-10.
Naavik, “Axie Infinity: Infinite Opportunity or Infinite Peril?”. https://naavik.co/digest/axie-infinity-infinite-opportunity-or-infinite-peril/ 2021-11-12. Read at source on 2026-10-10 (text only; charts not read).
Forkast, “Earnings in Axie Infinity fall below minimum wage”. https://forkast.news/earnings-axie-infinity-below-minimum-wage/ 2021-11-16. Secondary; read on 2026-10-10.
Context (Thomson Reuters Foundation), “Stung by losses, Filipino players ditch Axie Infinity crypto game”. https://www.context.news/big-tech/stung-by-losses-filipino-players-ditch-axie-infinity-crypto-game 2023-04-26. Read at source on 2026-10-10.
Rest of World on Axie Infinity after the Ronin theft. https://restofworld.org/2022/axie-infinity-hack/ 2022-06-22. Read at source on 2026-10-10.
Yield Guild Games, “YGG Community Roundup July 2021”. https://yieldguild.substack.com/p/ygg-community-roundup-july-2021 2021-08-12. Read at source on 2026-10-10.
Covalent, Axie analysis part two, archived copy. http://web.archive.org/web/20240415054958/https://www.covalenthq.com/blog/axie-part-two/ Undated, about January 2022; archived 2024-04-15. Read at source on 2026-10-10 from the archive.
Lai, Fan and Cai, study of Axie Infinity wallets, IEEE MetaCom 2023. https://faculty.washington.edu/weicaics/paper/papers/YimingLFC2023.pdf 2023-06. Read at source on 2026-10-10.
Cointelegraph on Yield Guild Games scholarships in early 2022. https://cointelegraph.com/news/game-on-yield-guild-games-scholarships-on-the-rise-through-q1 2022-05-24. Secondary; read on 2026-10-10.
Forkast, “Earnings from play-to-earn games dwindle”. https://forkast.news/earnings-play-to-earn-games-dwindle-upskill-players/ 2022-07-08. Secondary; read on 2026-10-10.
Decrypt on Axie Infinity’s marketplace fee rise. https://decrypt.co/94134/axie-infinity-players-push-back-on-fee-hike-as-play-to-earn-economy-lags 2022-03-01. Secondary; read on 2026-10-10.
Decrypt on Notcoin after its airdrop. https://kakapo.decrypt.co/235669/notcoin-loses-11-after-the-not-airdrop-claim-ends 2024-06-17. Secondary; read on 2026-10-10.
Decrypt, “What is Hamster Kombat”. https://decrypt.co/233407/what-is-hamster-kombat-telegram-game-token-airdrop 2024-06-02, updated 2025-01-02. Secondary; read on 2026-10-10.
The Defiant on Hamster Kombat’s airdrop exclusions. https://thedefiant.io/news/tokens/hamster-kombat-faces-backlash-for-excluding-57-of-users-from-airdrop 2024-09-23. Secondary; read on 2026-10-10 through the r.jina.ai reader.
Decrypt on Hamster Kombat’s account bans. https://decrypt.co/250940/why-hamster-kombat-purged-millions-players-telegram-airdrop 2024-09-24. Secondary; read on 2026-10-10.
HackerNoon, “How much did Hamster Kombat earn and why are users unhappy”. https://hackernoon.com/how-much-did-hamster-kombat-earn-and-why-are-users-unhappy 2024-10-09. Secondary (one author’s informal survey); read on 2026-10-10.
Decrypt review of Ev.io. https://decrypt.co/149902/frenetic-shooter-evio-earn-solana-while-fragging-enemies 2023-07-24. Read at source on 2026-10-10.
Decrypt on Bonk Arena. https://decrypt.co/323296/solana-meme-coin-bonk-kill-to-earn-shooter-game 2025-06-02. Secondary; read on 2026-10-10.
McKinsey, “For daily fantasy sports operators, the curse of too much skill”. https://karriere.mckinsey.de/industries/technology-media-and-telecommunications/our-insights/for-daily-fantasy-sports-operators-the-curse-of-too-much-skill 2015-09-01. Read at source on 2026-10-10 on McKinsey’s German site; mckinsey.com refused both routes.
Bitquery, investigation of Collector Crypt’s card product. https://bitquery.io/investigations/collector-crypt-jupiter-gacha 2026-07-13. Read at source on 2026-10-10 through the r.jina.ai reader.
Study of Ethereum gambling apps, PLOS ONE. https://pmc.ncbi.nlm.nih.gov/articles/PMC7592737/ 2020 (month not confirmed). Read at source on 2026-10-10 through the r.jina.ai reader.
BlackRain79, “What percentage of poker players are winners”. https://www.blackrain79.com/2019/09/what-percentage-of-poker-players-are.html 2019-09. Read at source on 2026-10-10; the author’s own estimate.
Kelley Drye, note on the Skillz v. Papaya Gaming verdict. https://www.kelleydrye.com/viewpoints/blogs/ad-law-access/game-company-hit-with-420-million-false-advertising-verdict 2026-04-28, with an update on the July 2026 order. Secondary; read on 2026-10-10.
FBI Internet Crime Complaint Center, alert I-030923-PSA. https://www.ic3.gov/PSA/2023/PSA230309 2023-03-09. Read at source on 2026-10-10.
FBI statement on the $620 million theft. https://www.fbi.gov/news/press-releases/fbi-statement-on-attribution-of-malicious-cyber-activity-posed-by-the-democratic-peoples-republic-of-korea 2022-04-14. Read at source on 2026-10-10 through the r.jina.ai reader; the direct route was refused.
Texas State Securities Board, Emergency Cease and Desist Order ENF-22-CDO-1860. https://www.ssb.texas.gov/sites/default/files/2022-04/Order_ENF_22_CDO_1860_.pdf 2022-04-13. Read at source on 2026-10-10. No longer cited on this page: the matter is not about a play-to-earn game.
US Securities and Exchange Commission, Litigation Release 26223. https://www.sec.gov/enforcement-litigation/litigation-releases/lr-26223 2025-01-16. Read at source on 2026-10-10. No longer cited on this page: the matter is not about a game.
Google Play, real-money gambling policy and blockchain content policy. https://support.google.com/googleplay/android-developer/answer/9877032?hl=en and https://support.google.com/googleplay/android-developer/answer/13607354?hl=en Undated. Read at source on 2026-10-10.
Apple, App Store Review Guidelines. https://developer.apple.com/app-store/review/guidelines/ Undated. Read at source on 2026-10-10.
Steam, onboarding rules for developers. https://partner.steamgames.com/doc/gettingstarted/onboarding Undated. Read at source on 2026-10-10.
Telegram, blockchain guidelines for mini apps. https://core.telegram.org/bots/blockchain-guidelines Undated; cites deadlines in February 2025. Read at source on 2026-10-10.
Florida Statutes, section 849.14. http://www.leg.state.fl.us/statutes/index.cfm?App_mode=Display_Statute&URL=0800-0899/0849/Sections/0849.14.html Current text. Read at source on 2026-10-10.
Lawrence G. Walters, overview of skill gaming in US states. https://www.firstamendment.com/skill-gaming-states/ 2020-08-14. Secondary; read on 2026-10-10.
Revised Code of Washington, section 9.46.240. https://app.leg.wa.gov/rcw/default.aspx?cite=9.46.240 Current text. Read at source on 2026-10-10.
Gambling Act 2005 (Great Britain), section 6. https://www.legislation.gov.uk/ukpga/2005/19/section/6 Current text. Read at source on 2026-10-10.
Bae, Kim & Lee, note on South Korea’s play-to-earn ruling. https://www.bkl.co.kr/en/law/insight/newsletter/5356 2023-01-31. Secondary; read on 2026-10-10 through the r.jina.ai reader.
US Internal Revenue Service, questions and answers on virtual currency. https://www.irs.gov/individuals/international-taxpayers/frequently-asked-questions-on-virtual-currency-transactions Reviewed 2026-06-30. Read at source on 2026-10-10.
US Internal Revenue Service, Topic 419, gambling income and losses. https://www.irs.gov/taxtopics/tc419 Reviewed 2026-09-24. Read at source on 2026-10-10.
Decrypt on the shooter OverTrip’s age rating. https://decrypt.co/295657/battle-royale-shooter-overtrip-epic-games-store-solana 2024-12-09. Secondary; read on 2026-10-10.
Trenches, wallet script. https://www.playtrenches.com/TemplateData/wallet.js Undated. Read at source on 2026-10-10.
Trenches, run-time settings. https://www.playtrenches.com/runtime-config.js Build tag production-20261006. Read at source on 2026-10-10.
Trenches, status address. https://www.playtrenches.com/inhouse-health Live reading. Read at source on 2026-10-10.
Trenches data feed, weekly and daily kills boards. https://trenches-api.vercel.app/api/leaderboard?metric=kills&mode=deploy&period=week Read at source on 2026-10-10, about 14:25 UTC and again later the same day.
Solana network public lookup, latest transaction of each of the 50 accounts on the profit board in source 9. https://api.mainnet-beta.solana.com Our own check, read at source on 2026-10-10; the reading of what the dates mean is our estimate.

Corrections

If you are quoted or named on this page and think something is wrong, want your reply shown beside it, or want to be removed, write to [email protected]. We aim to reply within 14 days, and always within one month. Factual errors are corrected with a dated note here. A sentence that is seriously disputed comes down while we check it.

Change log: (empty)