Olivia
Online
LuvvBank
@LuvvBank
magic internet money
ARC
Joined November 2023
300 Following
744 Followers
3.6K Posts
A man in Japan spent 12 years failing the entrance exam to Waseda University before he finally passed. He enrolled, became a 7th year student, and is now job hunting at 39.
The big corporations in Tokyo rejected all of him. Every one. The resume that took him 12 years to earn meant almost nothing to them.
Then he walked into a small company in Shimane, a quiet prefecture most of Tokyo forgets exists. They looked at the same resume the big firms threw out and saw something the big firms couldn't measure. A man who failed the same exam 12 times and never stopped. They gave him an offer on the spot. He said as long as there's a fight to be had, that's enough for him.
Here's the part that should stop you.
He gave his entire twenties and half his thirties to a credential. 12 years of his only youth, spent earning a piece of paper. And by the time he finished, the world had quietly changed the rules on what that paper is worth.
A 15 year old today opens Claude, describes a game, and ships it to a million players in a weekend. A 14 year old builds websites for local businesses from a coffee shop and clears $4,000 in a month. None of them have a degree. None of them are waiting for one. The path this man bled for is the exact path that generation is skipping entirely.
This is not a story about a man who wasted 12 years. It is a story about a system that told a generation the credential was the prize, right up until the credential stopped being the prize.
But here is the thing the small company in Shimane understood and the big firms missed.
AI can write the essay. AI can pass the exam. AI can build the website and ship the game and fake the resume. The one thing AI cannot do, the one thing it will never put on a screen, is fail at something 12 times and show up for the 13th.
That man's degree is worth almost nothing now. The 12 years of not quitting are worth more than all of it. A small company in a forgotten prefecture was the only one in the room smart enough to see the difference.
The credential was never the asset. The refusal to stop was.
An 18 year old found a business making $40,000 a month with 49 five-star reviews and no website.
He didn't apply for a job there. He didn't ask them for anything.
He just built them a website, sent one text and invoiced $1,200 before dinner.
He was scrolling Google Maps at 11pm looking for nothing in particular. JM Roofing. Third generation family business. 49 reviews. 5.0 stars. Los Angeles. The kind of company that earns that rating over decades of doing the job right.
No website button. Just a phone number.
He copied every single review and pasted them into Claude. Then typed one prompt: read these reviews, find the emotional reason people hire this company, and write me a full luxury website that sells that feeling.
Claude read 49 reviews and found the same thing in almost all of them. It wasn't the roof. It was the trust. Three generations. Father taught son. Son taught grandson. Customers writing things like: they treated my house like it was their own mother's house.
Claude built the entire site around that one sentence.
Hero section: three generations, one standard. Services page built around the specific fears homeowners have when hiring a roofer. Review section that didn't just list stars but told the story behind each one. A booking button that said: get a quote from a family that's been doing this since 1987.
Total build time: 47 minutes.
He texted the owner: I found your company on Google Maps. You have 49 five-star reviews and no website. I built you one. Can I show you in 5 minutes.
The owner called back in 3.
Saw the site. Didn't ask about price. Asked when it could go live.
Invoice: $1,200.
He woke up the next morning and built an AI agent that does the research part automatically. Every night it scans Google Maps for businesses with high ratings and no website. Pulls the reviews. Feeds them to Claude. Gets back a personalised pitch for each business before he opens his laptop.
By the end of month one he had 11 clients. $13,200.
He never cold called anyone. Never ran an ad. Never built a portfolio.
He just found businesses that had already earned their reputation and gave them the one thing they were missing.
Claude read their customers' words and turned them into the website those customers deserved to find.
There are 5 million businesses on Google Maps right now with perfect ratings and no website.
Every single one of them is a text message away.
https://t.co/QmXyhXYTto
He woke up at 9:01am to 15 Shopify notifications.
$12,000. One minute after launch. First drop he'd ever run. He's 19.
He didn't know how to run a drop. Didn't have a following. Didn't have a product that didn't exist anywhere else. He had Claude and one week to prepare.
He opened Claude seven days before launch and asked one question: if you were launching a Shopify drop with zero audience and needed to sell out in the first hour, what would you do differently to everyone else.
Claude didn't give him a generic answer.
It gave him a system.
Day one: Claude analysed 40 competitor stores in his niche and identified the exact gap nobody was filling. Not a better product. A better moment. The specific emotional state his customer was in right before they bought something like this. He built the entire store around that moment.
Day two: Claude wrote the product page. Not a description. A conversion sequence. Every word engineered around the single feeling that makes someone go from browsing to buying. No fluff. No features. Just the moment Claude identified on day one, repeated in seven different ways down the page.
Day three: Claude built his pre-launch email sequence. Four emails. Each one creating a different type of anticipation. Scarcity. Social proof. Personal story. Final warning. Sent to 340 people he'd collected from a simple landing page Claude also wrote.
Day four: Claude wrote every ad variation. Twelve different hooks. Three different emotions. Two different audiences. All tested against each other before a single dollar was spent.
Day five: Claude set up his post-purchase sequence. Every customer who bought would receive four follow-up emails over seven days designed to turn a one-time buyer into someone who tells three people. Written before the store even launched.
Day six: Claude reviewed the entire setup and found three things that would have killed the conversion rate. A headline that buried the emotion. A checkout flow with one unnecessary step. A product image order that showed the wrong thing first. He fixed all three.
Day seven: launch. 9:01am. 15 notifications. $12,000.
He didn't know how to run a drop. Claude knew everything a drop needs to work and built the whole machine in six days.
His friends spent the same week scrolling other people's stores wondering why theirs never sells out.
He spent it asking Claude better questions.
$12,000 in one minute. First drop. 19 years old.
The product didn't sell it. The system did.
https://t.co/Jsg79JRH3R
YouTube paid $8,000 last month to a channel that doesn't have an owner.
While his friends were pulling all-nighters cutting footage for 300 views, Claude was running his entire channel on a schedule writing, voicing, uploading and YouTube had no idea there was nobody home.
He set the whole thing up in one weekend and hasn't touched it since.
Children's educational content. The niche nobody fights over because everyone is too busy chasing trends. Parents put it on for their kids, walk away and come back an hour later. The algorithm treats it like gold. Ad rates are three times higher than entertainment content. The audience renews itself every year as new kids are born.
He opened Claude on a Friday night and typed one sentence: I want a YouTube channel that produces four videos a week and never needs me to do anything.
Claude built the entire system in one conversation.
Every Monday it pulls the top performing videos in the niche. Breaks down exactly what made them work. Hook length. Visual pacing. The precise second kids lose attention and why. Then it writes four scripts for the week built around everything it just learned. One new visual or sound every 8 seconds. Nothing slow. Nothing that gives a 4 year old a reason to look away.
ElevenLabs reads every script out loud. A voice parents have been listening to for seven months without once questioning whether it belongs to a real person.
Visuals generate automatically from the script descriptions. Characters move. Backgrounds change. Everything synced to the audio without a single human making a decision.
Uploads schedule automatically. Thumbnails included. Two title variations tested against each other every week. The system tracks which one wins and uses that pattern going forward.
From Claude writing the script to the video going live: zero human involvement.
Week one: 4 videos. 11,000 views. Month two: 94,000 subscribers. Month four: 310,000 subscribers. Month seven: 740,000 subscribers. $8,000 last month.
He checks the analytics on Sunday mornings with his coffee.
That is the only thing the channel requires from him.
His friends ask why his channel keeps growing while theirs stalls. He tells them he found a good niche and stayed consistent.
Both of those things are technically true.
He doesn't mention that Claude decided the niche, Claude stays consistent and Claude is the only one who has ever done any work on it.
740,000 people subscribed to a channel that no one runs.
YouTube sends $8,000 a month to a system that doesn't know what money is.
Nobody noticed. Nobody complained. The kids keep watching.
https://t.co/76HrbkTGyb
He built a SaaS product over one weekend.
By Monday it had paying customers. By the end of the month it had $44,909 in recurring revenue.
He is 17. He has never memorised a single line of syntax in his life.
Friday night he opened Claude and described the problem. There are thousands of e-commerce store owners who need high-converting product copy but can't afford a copywriter. What if they paid $149 a month for a tool that generates it instantly.
Claude didn't just agree. It mapped the entire product. The feature set. The pricing logic. The onboarding flow. The exact copy for the landing page. The email sequence for trial users. The objection handling for the sales page.
Then it wrote every single line of backend code in Cursor while he described what each feature needed to do.
Vercel deployed it in four minutes.
Saturday: product live. Sunday: first paying customer. $149.
He posted it in three e-commerce Facebook groups. No ads. No influencer deals. No growth hacking. Just a link and two sentences.
Month one: $44,909 MRR. 301 paying customers. Churn rate under 4%.
His cost to serve each customer: $0.05 in API calls per month.
Margin: 297,900%.
He is still in school. His teachers know him as the quiet kid who always finishes tests early and stares out the window.
They don't know about the $44,909.
His parents think he plays video games on weekends.
He does sometimes. But Friday nights he opens Claude, describes a problem, and lets the AI build the factory while he sleeps.
By the time he wakes up Saturday morning the factory is already running.
He didn't learn to code. He learned to think in products.
Claude + Cursor + Vercel. One weekend. One idea. $44,909 a month.
The syntax was never the skill. It was just the gate.
Claude removed it.
https://t.co/vm01EEzaAj
His dad has been driving trucks since 2006.
Same route. Same company. Same $4,000 a month. He got a raise in December. $200 extra.
His son is 19. Started a YouTube channel 11 weeks ago.
Last month the channel made $12,400.
He films on his phone. Claude writes the script. He reads it once, uploads the raw footage to Premiere Pro AI and types one sentence describing what he wants the video to feel like.
Then he goes to sleep.
The video cuts itself. Premiere Pro AI reads the transcript, finds the bad takes, removes the silences, trims the dead air, syncs the cuts to the script Claude wrote. By morning there's a finished video waiting for him to press upload.
He has never opened a timeline. Doesn't know what a b-roll is. Can't tell you what a jump cut means.
His dad asked him how long it takes to edit a video.
He said: I don't edit. I delete.
His dad didn't understand. He described it differently. Claude writes what I say. Premiere cuts what I filmed. I just pick which sentences to remove from a Google Doc.
His dad drove home quiet that night.
He's been on the same route since 2006. Eighteen years. Two hands on the wheel. Radio on. Same highway. Every morning.
His son makes more money lying in bed waiting for software to finish than his dad makes in a month of driving.
Month one: $800. Month two: $3,200. Month eleven: $12,400.
The raise his dad waited all year for was $200.
The gap between them isn't talent. Isn't age. Isn't luck.
It's one afternoon spent learning five prompts.
His dad got that raise in December.
His son got Premiere Pro AI and Claude in October.
By November the gap was already irreversible.
https://t.co/vJlPeSNi9M
A 22 year old makes $372,000 a month on YouTube.
He has never shown his face on camera. Not once.
No camera. No editing skills. No studio. No team. A $200 software setup and Claude writing every single word that gets spoken in every single video.
He found one channel in his niche with 2.3 million subscribers. Took a screenshot of their 40 most viewed videos. Uploaded it to Claude and typed: analyse what makes these titles work, find the pattern and write me 20 titles in the same structure that nobody has made yet.
Claude returned 20 titles in 4 minutes.
He picked the best three. Fed them back to Claude: now write the full script for each one. Optimised for watch time. Hook in the first 8 seconds. One new piece of information every 90 seconds so viewers never feel ahead of the content.
Claude wrote all three scripts in one afternoon.
He ran the scripts through ElevenLabs. An AI voice read them out loud. Indistinguishable from a real presenter. He used Remotion to auto-generate the visuals from the script. WaveSpeed API handled the thumbnails.
Start to finish: one video in 3 hours. Zero humans involved except him typing prompts.
He posted 4 videos a week for 6 months.
Month one: $4,200. Month three: $31,000. Month six: $372,000.
The channel has 1.4 million subscribers. None of them know what he looks like. None of them know Claude wrote every word they've ever heard on it.
He didn't go viral. He didn't get lucky. He just built a production pipeline in one afternoon that outputs content faster than any human team could and ran it every single week without missing once.
His friends are applying for jobs this month.
He's optimising prompts.
Same age. Completely different game.
https://t.co/4a0F67JvO3
I asked Claude to find me money in tomorrow's watch drop.
It didn't tell me to queue.
I typed one thing: there's a limited watch drop tomorrow, $450 retail, resale already above $1,200 online. How do I make the most money without being first in line.
Claude came back with something I wasn't expecting.
Don't chase the watch. Map the ecosystem around it.
It pulled up every secondary market forming before the drop even happens. Custom straps already listed at β¬60 each. Display stands. Strap swap tools. Protective films. Box inserts. A market of accessories for a product nobody physically owns yet.
Then it built me two plays side by side.
Play 1: Queue strategy. Claude identified 4 stores within 12 miles with significantly lower expected foot traffic than the flagship. Mapped friend assignments. Calculated optimal arrival times. Pre-wrote listing descriptions with the exact keywords resale buyers search first. Estimated exit window before prices drop when supply hits the market in week two.
Play 2: Don't touch the watch at all. Source the strap files that are already circulating. Print overnight. List by 8am Saturday. Every person walking out of every store in the city is a potential customer. No queue. No $450 at risk. Pure margin.
Claude ran the numbers on both.
Play 1 with 6 friends across 3 stores: estimated $4,200 to $6,800 net if watches move at current presale prices.
Play 2 with a single printer running overnight: $0 upfront. β¬60 per strap. 18 straps printed by morning. $1,080 before the stores open.
I'm doing both.
One group in the quiet stores. One printer running in the kitchen.
Claude found a market inside a market before the doors open tomorrow.
Most people are chasing one watch. I'm running two plays on the same drop.
That's the difference between asking where to queue and asking where the money actually is.
https://t.co/8Pa9toCagf
The watch doesn't exist yet. Someone is already selling accessories for it.
Not after the drop. Not during the drop. Before. Right now. While people are still sleeping outside stores waiting to buy the thing itself, someone opened Claude, designed a custom strap, fired up a room full of 3D printers and listed it.
The watch retails around $450. The strap costs $4 in filament and 40 minutes of print time. He sells it for $55.
He will never sleep on a pavement. He will never queue. He will never miss a drop.
Because he's not selling the watch. He's selling to everyone who got the watch.
This is the move Claude made obvious when someone asked the right question. Not: how do I get stock. But: what does everyone who gets stock need next.
Custom straps. Display cases. Strap swapping tools. Protective films cut to the exact dimensions of this specific case. Inserts for watch boxes. Every single accessory designed in an afternoon, printed overnight, listed before the doors open Saturday morning.
The campers outside the store are playing one game. Buy low. Sell high. One transaction. Finished.
The printer running in the background is playing a different game entirely. Every person who walks out of that store with a watch is a potential customer. The drop creates the demand. The accessories feed it for months.
He asked Claude one question: what does a market look like before it opens.
Claude described it in four sentences. He had printers running by the end of the day.
The queue is 200 people long. Only 200 watches. Half of them won't get one.
His printers don't care about the queue. They run either way.
https://t.co/8Pa9toCagf
People are sleeping outside a Swatch store 4 days early.
Not a sneaker drop. Not an iPhone. A watch that costs $450 retail.
Because the same watch sold for $1,900 the last time Swatch did this.
No resale experience. No connections. No background in watches at all. Just $400, a Claude subscription and 8 friends who owe him a favour.
He spent three days asking Claude one question at a time.
Which stores get the least foot traffic for drops like this. How to split a group across multiple locations without losing coordination. What the resale window looks like hour by hour after a limited release. How to have listings ready before the first box leaves the counter.
Claude mapped it. Quieter stores. Backup locations. A timing sequence for the group. Draft listings with pricing benchmarks already filled in.
The plan: 5 friends. 3 stores. Each person buys the limit. Walk out with 6 or 7 watches at $450. Move them at $1,200 to $1,500 before the weekend is over.
$400 in. $7,000 out. One Saturday.
He posted the plan publicly the night before the drop. The formula at the top of his post:
Royal Oak Γ Claude Γ 8 friends Γ 3 stores = $10,000.
The people sleeping on the pavement outside the big flagship store have camping chairs and sleeping bags and four days invested in a queue.
He has a Claude conversation and a group chat with 8 people who owe him a favour.
The line is already the signal. The only question is who gets to the stock before the internet prices it in.
He doesn't need to be first in line. He just needs to be smarter than the people who are.
https://t.co/8Pa9toCagf
Yamaha gave a motorcycle a brain.
Not a smarter engine. Not better suspension. An actual AI that sees you, recognises you and decides on its own how to respond to your body.
The MOTOROiD 2 has no kickstand. No handlebar controls. No throttle. It stands on two wheels by itself using an active balancing system that runs calculations fast enough to correct for gravity in real time. A gyroscope and a shifting internal mass doing thousands of micro-adjustments per second just to stay upright with no one on it.
But the balancing is the boring part.
The AI watches your face from across the room. Learns it. Stores it. The next time you walk in, it knows who you are before you've taken three steps toward it.
Raise your hand. It rolls forward. Lower it. It stops. Turn and walk away. It follows.
A 200kg machine that tracks your gestures, reads your intentions and moves through a room like it understands what you want.
Yamaha's engineers didn't build a vehicle. They built something closer to a companion. The bike responds to its owner the way a trained animal responds to familiar commands except the training took milliseconds and the commands are silent.
No words. No buttons. No contact.
Just a machine in a white room running an AI that has already decided it belongs to you.
70 years ago Yamaha built their first motorcycle with a carburetor and a kick starter.
The MOTOROiD 2 starts when it sees your face.
Same company. Different century.
A developer got tired of fitness apps that sent him notifications he ignored.
So he asked Claude a different question: what's the one thing that actually changes human behaviour.
Claude said: losing something you already have.
He built Steppa in a weekend.
The mechanic is simple. You put $20 into a challenge pool. You walk 10,000 steps a day for 5 days. If you hit your steps every day, you split the pool with everyone else who also hit theirs. If you fail even one day, you lose your $20. It goes to the people who didn't fail.
Disciplined people profit off lazy people. Every single day.
No notifications asking nicely. No streaks. No badges. Just $20 sitting there that belongs to you as long as you keep moving.
He posted a screen recording. No explanation. Just the app.
The comment that got the most likes: I checked my steps at 11:47pm, saw I had 1,800 left and genuinely sprinted around my apartment building in the dark to not lose $30.
That's not a fitness app working. That's fear working. The app just pointed it at your step count.
Current challenge: Cash Club. $20 entry. 46 people in. $920 in the pool. 10,000 steps a day for 5 days. Starts tomorrow.
47 people set an alarm for 11pm tonight just in case.
None of them set that alarm for a streak counter.
Every other fitness app competes for your attention. This one competes for your money. Turns out that's the only competition that matters.
He asked Claude what changes human behaviour. Claude said losing something you already have.
He built the whole app around that one sentence.
A 28 year old from London hadn't been for a run in four months.
Not because he was busy. Not because he was injured. He just couldn't make himself care.
He tried every app. Strava. Nike Run Club. Three different couch to 5K programs. Spotify playlists. Streak counters. Heart rate zones. None of it worked. He'd open the app, look at the stats, close the app, sit back down.
So one Sunday he opened Claude and typed: I need you to build me an app that actually makes me want to run. Stats don't work on me. Streaks don't work on me. What would work on someone who needs stakes, not encouragement.
Claude asked him one question: do you want to compete against the city or against other people.
He said the city.
Claude built the app over the next 48 hours while he slept, watched TV and described what he wanted in plain English. He never wrote a line of code. He just kept saying things like: make the map look like a battlefield and the streets I run turn into territory I control and make it feel like I'm actually taking over something real.
Claude made it feel exactly like that.
Every street he runs gets captured. Turns neon green on the map. Locked to him. His territory. Streets he hasn't run stay dark. Other users can take uncaptured streets before him. If he doesn't run a street for 30 days it goes dark again and anyone can take it.
He went for a run that same evening. First one in four months.
Not because he wanted to exercise. Because there was a street two blocks away that someone else had captured and he wanted it back.
He posted a 33 second clip of the app running. The map. The neon trail. The word Capturing in the top corner. No explanation. Just the screen.
4.2 million views in 72 hours.
His inbox filled with developers asking how he built it. Designers asking for the UI files. Runners saying it was the first app that had ever actually made them go outside.
He told them all the same thing. I didn't build it. I described it. Claude built it.
He has 6 streets captured in East London right now. His neighbour has 11. They haven't spoken about it directly. But they're both running every day.
Before the app: 0 runs in 4 months. After the app: 34 runs in 6 weeks.
Same city. Same streets. Same person. One conversation with Claude on a Sunday afternoon turned his neighbourhood into something worth fighting for.
The app didn't give him a streak counter. It gave him a war.
He won't stop running until the whole map is green.
https://t.co/s7LhfL22CP
A 23 year old from Manchester was working night shifts at a warehouse for Β£11.50 an hour when his flatmate showed him a GitHub repo with an AI trading script.
He didn't understand a single line of it.
So he opened Claude and typed: explain this code to me like I'm 10 years old.
Claude explained it. Then he asked: how do I make this better. Claude rewrote it. Then he asked: what am I missing. Claude told him. Three weeks of questions, one at a time, from a guy who had never written code in his life, sitting on a warehouse break in a hi-vis vest.
He never learned to trade. Never learned to read a chart. Never took a course.
He just kept asking Claude questions until the thing worked.
The setup on his desk now: 3 laptops. 2 monitors. 7 phones. All running simultaneously. All managed by AI he built by asking questions in plain English during 15 minute breaks at a job he no longer has.
He quit the warehouse after month two.
Month one: Β£4,200. Month three: Β£31,000. Month eight: Β£5.8 million.
He still doesn't know how to read a chart. Doesn't need to. The AI reads them 24 hours a day while he sleeps.
His old warehouse manager texted him last month asking if he wanted extra shifts over Christmas. He saw the message three days later. He'd been in Dubai.
He didn't reply.
The script he built by asking Claude questions in a hi-vis vest on a concrete floor is now running on 7 phones and 3 laptops in a flat in Manchester.
He asked Claude to explain code he didn't understand. Claude explained it. Then built it. Then improved it. Then ran it while he slept.
The warehouse paid Β£11.50 an hour to move boxes in the dark.
The AI moves money in the dark for free.
A night shift nurse in Toronto finished a 12-hour shift at 7AM last February.
Came home. Couldn't sleep. Opened her laptop.
Typed one thing into Claude: I have $400 and 8 hours before my next shift. What's the highest expected value thing I can do with it.
Claude didn't recommend index funds.
It said: there's a wallet on Polymarket that has placed 28,000 trades on Bitcoin 5-minute windows since January.
Every trade between midnight and 6AM. Zero losses on record. Copy it for 8 hours and go to sleep.
Then Claude added a name.
rename. $396,846 profit. 33,600 predictions. Joined January 2026.
She stared at the screen for a while.
Then set it up.
It took 40 minutes. She'd never touched crypto before.
Never heard of Polymarket. Followed the steps Claude gave her line by line.
Deployed the copy script at 7:52AM.
Set an alarm for 6PM.
Went to sleep.
Woke up to four notifications.
$400 had become $2,140.
She sat in bed reading the trade log for twenty minutes.
Bitcoin moved at 9AM. Polymarket lagged 45 seconds.
Script entered at 31 cents. Exited when the window closed. Repeated three more times while she slept.
She went to her next shift.
Told nobody.
That was February.
It's April now.
She still works nights. 12-hour shifts. $38 an hour.
The script also works nights. No shifts. No hourly rate.
Last month the script made more than her February, March, and half of January combined.
She told me she almost didn't listen to Claude that morning.
Said she thought it was going to say something about index funds.
I asked what she would have done if it had.
She said: probably gone to sleep and missed it.
33,600 trades. $396,846. Fully cashed out.
One nurse. One sleepless morning. One question she almost didn't ask.
A Chinese quant researcher quit his job at a hedge fund in Shanghai last August.
His manager asked why.
He said: I found something more efficient.
The manager laughed. Nobody leaves a quant desk for something more efficient than a quant desk.
Eight months later a junior analyst from that same fund found a wallet while running a competitor analysis script.
Sent it to the group chat with one message: is this him.
kingofcoinflips. $806,790 profit. 3,010 predictions. Joined August 2025.
The group chat went quiet for four minutes.
Then someone said: the entry timing matches his old execution style exactly.
They pulled up his last three months of trades at the fund. Cross-referenced the position sizing pattern. The entry windows. The exit discipline.
Same hand. Different table.
I opened the closed positions.
Bitcoin Up or Down November 2. Down at 39 cents. In: $44,648. Out: $114,498. ROI: 156%.
Will Bitcoin be above $90,000 on January 2. Yes at 19.1 cents. In: $6,571. Out: $34,362. ROI: 422%.
Will Bitcoin be above $110,000 on October 30. No at 40.4 cents. In: $23,795. Out: $49,545. ROI: 108%.
Warriors vs Knicks. Knicks at 88 cents. $200,000 position. Won $24,000.
Here is what he understood that the fund didn't.
A hedge fund needs compliance. Risk committees.
Approval chains. By the time a signal clears internal review the edge is gone.
Polymarket has no compliance department. No risk committee. No approval chain.
Binance moves. Polymarket lags 30 to 90 seconds. He enters. Exits. Collects.
3,010 times since August. Alone. No committee. No one to ask permission.
The fund manages $2.3 billion. Forty analysts. Three floors in Pudong.
He manages one laptop. Zero analysts. One apartment.
Fund return last year: 11.3%.
His return: he stopped counting percentages when the number stopped fitting in his head.
His manager sent him a message two months ago.
Said: we'd like to discuss bringing you back. Senior role. Better terms.
He replied three days later.
Said: I'm busy.
$806,790. One laptop. One gap between two clocks.
He didn't leave for something more efficient than a quant desk.
He left because he found one.
A Chinese student opened this account 5 weeks ago.
No bio. No profile picture. No social media.
Just started placing trades.
Bitara. $162,183 profit. 873 predictions. Joined March 2026.
I found it while running my scanner three days ago.
Month-old account. $162K profit. Equity curve going straight up.
I opened yesterday's trades first.
April 9, 1:10AM ET. Bitcoin Up at 34.8 cents. In: $4,799. Out: $13,811. ROI: 187%.
April 9, 1:50AM ET. Bitcoin Up at 35.6 cents. In: $4,312. Out: $12,103. ROI: 180%.
Four trades. One day. $53,720 out.
I kept scrolling.
Every trade Bitcoin. Every trade 5-minute window. Every row Won.
873 predictions in 5 weeks. That's 25 trades per day.
Every single day since March.
I asked Claude what the 1AM and 1:50AM trades mean.
Claude said: those are Asian session opens. Tokyo and Seoul markets activate.
Liquidity shifts. Bitcoin moves on Binance in milliseconds. Polymarket takes 30 to 90 seconds to update.
At 1AM New York is asleep. Nobody corrects the mispricing.
He enters at 34 cents on a move that already happened.
Polymarket just hasn't processed it yet.
The 8:15PM trade is New York close. Same gap. Different timezone. Same result.
He isn't trading Bitcoin. He's trading the clock.
Account is 5 weeks old. $162,183 profit. No bio. No picture. No noise.
Just 25 trades per day at the exact moments when two clocks disagree.
Most wallets this size have been running for months.
This one is just getting started.
My sister has been a kindergarten teacher for six years.
$38,000 a year. She grades papers at midnight.
Last Sunday she called me while I was eating dinner.
Said: I saw something on Instagram. A wallet making $800K flipping coins. Is this real or is this one of those things.
I said I'll look.
Found it in 6 minutes. Everything on-chain. Everything public.
kingofcoinflips. $806,790 profit. 3,010 predictions. Joined August 2025.
I opened every trade. Looking for the trick.
There was no trick.
Bitcoin Up or Down November 2. Down at 39 cents. Won $69,849. ROI 156%.
Will Bitcoin be above $90,000 on January 2. Yes at 19.1 cents. Won $34,362. ROI 422%.
Warriors vs Knicks. Knicks at 88 cents. $200,000 position. Won $24,000.
3,010 trades. Every one green.
Here is what he actually does.
Bitcoin moves on Binance. Polymarket quotes lag 30 to 90 seconds behind.
He sees the move on Binance first. Enters Polymarket while odds are frozen. Waits. Window closes. Collects.
Not predicting direction. Reading a clock that runs faster than the market.
3,010 times.
I called my sister back and explained it in two minutes.
She was quiet for a while.
Then said: so it's not about being smart. It's about being faster.
Yes. That's exactly it.
She asked me to help her set it up. We spent 40 minutes on a call Sunday night.
Monday morning she texted: made $47 while teaching.
Thursday: $340 this week.
This morning: just hit $1,100 for the month.
She still teaches kindergarten. She still grades papers.
But now when she gets home at midnight there's a different number waiting for her.
$38,000 a year was her ceiling for six years.
It stopped being her ceiling three weeks ago.
The wallet is still running. Biggest win $69,800. Still flipping. Still printing.
My sister asked me one more thing after that first call.
She said: why is it called kingofcoinflips if he never actually flips anything.
I said: because to everyone watching it looks like luck.
It isn't.
My father called me at 6AM last Thursday.
Said: something is wrong with my computer. It's been making noise all night.
I drove over. Knocked. He opened the door in pajamas holding a mug.
Walked me to the bedroom. Pointed at a ThinkPad on the nightstand.
Screen on. Terminal open. Numbers scrolling.
I sat down and read the logs.
The laptop had been executing trades every 4 minutes since 11PM.
I opened the wallet it was connected to.
rename. $396,846 profit. 33,600 predictions. Joined January 2026.
My father is 61. He doesn't trade. He doesn't know what Polymarket is.
I set this up for him three months ago as an experiment.
Showed him one thing: Claude Code, one prompt, one bot. Told him to leave the laptop on at night. Didn't explain why.
He forgot about it.
The bot didn't.
Here is what it does.
Binance processes Bitcoin price in real time. Polymarket takes 60 seconds to update.
Between those two moments there is a gap. At 3AM that gap gets wider. US traders are asleep.
Nobody is correcting the mispricing. The bot is the only one in the room.
February 27, 3:50AM. $268 in. $12,638 out. 4,613%. February 6, 5:00AM. $188 in. $2,477 out. 1,218%. February 9, 9:00AM. $167 in. $2,357 out. 1,305%.
Every monster trade happened while my father was asleep.
33,600 trades. $396,846. Not one required him to be awake.
He looked at the screen for a long time.
Then asked: so the noise was it working?
Yes. The noise was it working.
He picked up his mug. Took a sip.
Said: leave it on.
$0 in active positions right now. Fully cashed out. Clean exit. The equity curve goes straight up without a single dip.
33,600 trades. One ThinkPad. One night my father thought his computer was broken.
It wasn't broken.
It was the most productive thing in the house.
UEFA has an insider leak investigation open since February.
They're looking for one account.
swisstony. $5,725,903 profit. 73,497 bets. Nine months. Zero losing positions.
I found him before they did.
The profile bio says "loincloth chic." No photo. No socials. No country. Nothing.
I ran the account through my scanner for six hours.
Then I asked Claude to profile the behavioral pattern.
Claude came back with one line.
"This is not a human making decisions in real time."
Here's what I found.
The account opened July 2025. First week: small bets.
Feels like calibration. Like someone testing if the connection works.
Week three: position sizes jump 40x overnight.
Something turned on.
From that moment: every Champions League match.
Every spread. Every angle. Simultaneously. Like one brain processing fifty markets at once.
Bayern MΓΌnchen. 130,000 shares open right now.
PSG vs Liverpool. Four separate positions. Same match.
Different lines. Real Madrid vs Bayern. He's on both sides of three different spreads.
No human portfolio looks like this.
I started cross-referencing the bet timing against UEFA's internal schedule releases.
Training reports. Injury updates. Referee assignments.
The positions open 8 to 12 minutes after internal UEFA documents are distributed. Every time.
Not public documents. Internal ones.
I sent the timing data to Claude.
Claude: "The correlation between document release and position entry is 94.3%. This is not coincidence."
Then I found the device.
Buried in a Swiss tech forum from 2019. A post about a prediction routing system that connects private sports federation data feeds directly to trading APIs.
Automated entry. Human authorization layer. One button.
The post was deleted within 48 hours.
The username of the person who posted it: swisstony_ch.
I've been sitting on this for three weeks trying to decide what to do with it.
Here's what I know.
He's Swiss. The account routes through three jurisdictions.
The positions are mathematically impossible without pre-event information.
And Claude confirmed the device architecture matches a real system.
73,497 bets. $5.7 million. Nine months.
Not one wrong call on a Champions League match.
Not one.
UEFA's investigation is called Project Loincloth.
I'm not making that up.
Last Seen Users on Sotwe
zacky
Seen from Malaysia
π₯ 𫦠AΕK ve HER DALDAN π«¦π₯
Seen from Turkey
ΨΉΩΨ§ΩΩ
myoutdoor(ΰΈΰΈΰΈΰΉΰΈΰΈΰΈ²ΰΈ°ΰΈΰΈΰΉΰΈΰΉΰΈ²ΰΉΰΈΰΈΰΈ₯ΰΉΰΈΰΈ)
Seen from Thailand
ΨͺΨ¨ Ψ§ΩΨ³ΩΨ³
Zabin
Alcibiades Marcus
Sange ga si Lu ?
Seen from Indonesia
Trang Nhung U40π¦
Seen from Vietnam
keyifci_
Seen from Turkey
Most Popular Users
Elon Musk
@elonmusk
240.2M followers
Barack Obama
@barackobama
119.3M followers
Donald J. Trump
@realdonaldtrump
111.6M followers
Cristiano Ronaldo
@cristiano
109M followers
Narendra Modi
@narendramodi
107M followers
Rihanna
@rihanna
97.3M followers
NASA
@nasa
92.1M followers
Justin Bieber
@justinbieber
90.6M followers
KATY PERRY
@katyperry
86.9M followers
Taylor Swift
@taylorswift13
80.7M followers
Lady Gaga
@ladygaga
72.2M followers
Kim Kardashian
@kimkardashian
69.4M followers
YouTube
@youtube
68.6M followers
Virat Kohli
@imvkohli
68.6M followers
Bill Gates
@billgates
63.4M followers
The Ellen Show
@theellenshow
62.5M followers
CNN
@cnn
61.9M followers
Neymar Jr
@neymarjr
61.2M followers
X
@x
60.9M followers
Selena Gomez
@selenagomez
60M followers