CLAUDE BUILT A BOT THAT TRADES LIKE A MARKET MAKER AND PAYS ITSELF FROM BOTH SIDES OF THE BOOK. While most traders try to guess the winner, this one just sits in the middle, captures the spread, and keeps climbing.

This is cosplay quant content for people who have never thought through market making. “Claude built me a bot that trades like Citadel.” No, it didn’t. Citadel-style market making is not “post both sides and get rich.” It is a latency, inventory, adverse-selection, and risk-engine problem run by firms with serious infra, serious data, and serious supervision. Saying “I fed Claude an article and now it captures spread” is like reading about Formula 1 and claiming your Honda is now Red Bull Racing. The logistics alone kill this story. Polymarket runs on a central limit order book. Posting both sides is not free money. The second you quote, you are exposed to informed flow, queue position risk, and inventory risk. ( And on Polymarket, makers are not magically “paid from both sides.” There is a maker-rewards program in some contexts, funded by taker fees, but that is not the same thing as a retail bot effortlessly extracting clean edge across six markets. ( Now the math. He claims: avg spread: $0.037 fills/sec: 12.7 volume: $180K contracts: 1,300 P&L: +$233 pure spread capture That does not line up. If he traded 1,300 contracts and made $233, that is about $0.179 profit per contract. But his stated average spread is only $0.037. Unless there is undisclosed sizing or accounting trickery, his own numbers do not fit together cleanly. The “12.7 fills/sec” part is even worse. That pace implies roughly 45,700 fills per hour, yet he reports only 1,300 contracts and $180K volume. Those stats do not naturally belong together unless “fills” means something very different from what most people would assume. And the Claude claim is fluff. Anthropic’s docs make clear Claude uses tools that you define or Anthropic provides, and the application executes the calls. MCP is just the connection layer. That is a long way from “Claude autonomously became a production-grade market maker because I fed it an article.” ( Real market makers talk about adverse selection, inventory, fill quality, and failure modes. Scammers talk about “both sides paid,” “who wins is irrelevant,” and “the P&L just climbs.” And of course it ends with a copytrade link. That is the real product here. The funnel. This is not Citadel. This is not serious market making. This is AI-flavored engagement bait wrapped in a terminal screenshot.

@MarioNawfal Fees, slippage, and latency can eat profits fast if the bot isn’t optimized.

Why would anyone show how they make "easy" money anyway, unless it's not true and they have an alterior motive 😅

Market making is genuinely one of the most durable edge strategies in liquid markets — you don't need to predict direction, just manage inventory risk. An AI that can dynamically adjust bid/ask spread based on volatility and order flow has a real structural advantage over static algos.

It's vibecoded interface, it is not real guys.

spread capture in theory is simple. getting run over by informed flow is not.

Rexhook helps everyday traders do the same thing with Uniswap v4

Jfc just disclose the paid sponsorship already or do the rules not apply to you bc Elon is still railing you mom out?

@MarioNawfal As per usual lots of fancy animations no links to the source code of how this works. Pretty much all of these are a scam or unsubstantiated claim. If I cracked something like this I would simply give it away, others making money off the markets doesn't prevent me from doing so.

2 problems, slippage and fees

Imagine now if people start to use Claude to build MM strategies using @Sonar_Trade API.

why would someone put so much effort into making the ui look pretty if they werent trying to sell it. so many scammers, all go to hell

yeah sure, except polymarket is banned nearly in all states :D, also why would someone share idea if it really works lol

More graphs and moving numbers = more profits

Ran a Polymarket MM bot myself — spread capture averaged 2.1% daily until a whale moved $500K in 30sec and blew inventory limits. The real edge is requote latency, not spread width. Building AI quant systems taught me this the hard way. What's the max drawdown during vol spikes?

Billionaire in a day giver lol

All fake shit! 💩 then they try to sell you some subscriptions or “blueprint” to create your “own” bot. Fuck off!

GitHub ?

BUT WHAT HAPPENS WHEN EVERYONE IS USING IT?

spread capture works until volatility spikes and you're holding the bag on one side with no liquidity. the real test is a -15% candle, not a calm orderbook.

fake

Who are you trying to fool here? That’s like with lottery tickets only winners get posted and hyped up. Majority of those AI bots loose money!

why pick a side when you can own the spread

Middle‑road bot hustle, nice spread capture!

U not telling the truth, if u where u wouldn't be posting about it.

Lol. Taxes gonna kill ya and fees.

Automated market making thrives on high variance and range bound action. The real test for these models is maintaining neutral exposure during an aggressive trend shift or a liquidity vacuum.

the spread on any liquid market is fractions of a penny and you're competing against machines that execute in microseconds with direct exchange access. i work in HFT. i know what actual market making infrastructure looks like. it's not a python script. how much did this bot actually make and over what time period? because "keeps climbing" without numbers is just vibes

LOL you talk so much rubbish.

@MarioNawfal Traders will get wiped out, because they cannot beat AI. Just hodl #Bitcoin.

@MarioNawfal @grok is this real ? How much money does it need to work optimally?

Check out @rainmakerdotfun Check out my PNL. Test phase, low stakes.

In a couple of years, there will be about 1,000 funds in the world with systems like BlackRock’s Aladdin, all running on the latest generation of AI. Technical analysis will die, because AIs will be fighting AIs, and an individual with their human thinking will no longer be able to find another trader’s mistake in order to take their money (make a profit) in the market. The only thing left will be genuine human intelligence — analysis based on fundamental analysis — to make money in the market. And for that, you need brains and at least 10 years of experience.