
Ventry
@ventry089 • 3,235 subscribers
still won’t stop
Shorts
Videos

this kid is 18. his app makes $1.4 million a month one function: take a photo of your food - get the calories the entire product is one API call. photo goes in, JSON with calories comes out. frontend shows the number Cal AI. 15 million downloads. MyFitnessPal acquired them marketing: tiktok. influencers film themselves photographing their food. viral by default - everyone eats every day. CAC close to zero one vision API request: $0.01-0.03. user subscription: $5.99/month. margin 80%+ someone already built an open-source version of the same thing: -> photo and text analysis, meal history, weight tracking, model selection, PWA, works offline a $1.4M/month app is one vision API call and the right distribution. the code is open. distribution is your problem
Ventry420,327 views • 1 month ago

I made 4 bots kill each other for my money gave each $500 and said "in a week one survives" bot 1- copies whales. just mirrors top-5 wallets bot 2 - contrarian. waits for crowd to be 80%+ confident and bets against bot 3 - scalper. catches delay between binance and prediction markets bot 4 - weather guy. trades only temperature. three meteo models, enters when all three agree stats for 7 days: day 1: scalper ahead. +$180 others around zero day 2: nothing interesting day 3: contrarian caught skew on sports market. +$740 on one trade. took the lead day 4: scalper started dumping. delay shrank, edge gone. $500 → $310 day 5: weather guy silent for 4 days. then made TWO trades. both profit. +$420 day 6: whale copier steady in the middle. no fails, no highs. boring but alive day 7: scalper dead. $82 left results: contrarian - $1,740. one explosion decided everything weather guy - $1,120. two trades in a week copier - $780. survived but unimpressive scalper - $82. corpse the dumbest bot won. it just waited for everyone to err and entered once and the smartest died first cuz its edge was temporary
Ventry1,449,256 views • 3 months ago

youtube is paying $50,000 a month to channels running motion graphics. no face. no voice. just graphics and numbers the niche is ranking videos. countries by GDP, population, area. WorldData_3D - 700K subscribers. Global Data - 660K subs. one video at 5 million views = $15,000 from ads alone this used to be rendered by a 3D artist in Unity and Blender. weeks of work per video now it's one prompt - finished animation in seconds. you describe, AI generates, you tweak the colors via chat, export mp4 the trick isn't the tooling. it's the format ranking works because the brain has a built-in curiosity timer. who's next? who's first? - people watch to the end. youtube sees retention, pushes harder if you want it self-hosted --> open source motion graphics in React. 70+ effects. prompt claude code, get a finished video 5 channels in parallel = $150-$350K a month from one pipeline youtube doesn't check who rendered the video. it checks retention. the barrier used to be the artist. now the barrier is who hits publish first
Ventry234,075 views • 26 days ago

A former quantum from Jump Trading sent me the source code of the bot he said they fired him for running it from the work server when i opened the code i didn’t sleep til 4 am gist of the code: bot holds two websocket connections simultaneously direct feed from Binance spot orderbook and Polymarket CLOB API BTC price on Binance updates ~200ms polymarket recalculates contracts with 3-7s lag in this window bot calculates implied probability compares spot divergence with contract price over 8% - enters btc does +0.4% on spot up contract at 53¢ model probability - 87% bot buys in 5 mins contract resolves at $1 20-50% per trade 100+ times a day $284,719 per month off $1,300 bankroll this isn’t prediction it’s latency arb between CEX and prediction market no need for $50k colocation $20 vps and clean code suffices
Ventry403,227 views • 4 months ago

$5,280 a year on AI subscriptions. Claude Max, ChatGPT Pro, Gemini, Cursor. $440 every month going to the cloud. a mini PC for $1,700 does the same thing on your desk. GMKtec EVO-X2 - a box the size of a hardcover book. inside it the chip that Lisa Su personally brought on stage at CES and signed with her name. 128GB of memory in one pool. runs models that cloud subscriptions behind $200 paywalls throttle during peak hours. here there are no limits - your hardware, your rules. NVIDIA's RTX 5080 at $1,000+ lost to this box by 3x on a real AI workload. a discrete GPU beaten by a mini PC the size of a lunchbox. electricity - $9 a month. pays for itself in 10 months. after that $440 every month stays in your pocket. the cloud made sense when nothing on a desk could compete. now it can.
Ventry26,897 views • 8 days ago

this video explained to me why my cs2 skin arbitrage bot actually works guy shows a working method - csfloat → skinport through his own google sheets calculator logic is right, formulas are correct only thing - while you're calculating one skin in excel, skinport refreshes another 200 listings $7.82 billion market 12 spreads appear every 4 seconds ak-47 redline currently gives +17% roi awp dragon lore +$250 net the window stays open exactly until someone runs an automated parser on these same routes i already did same math he has in google sheets just async across 5 marketplaces in parallel full skinport catalog (47,823 items) scanned in one request like + bookmark - you'll need this when you build your first arbitrage bot through claude
Ventry117,046 views • 1 month ago

Former bookmaker leaked his approach to polymarket profit under $2m. without a single bot he got fired when the company closed he transferred everything he knew to prediction markets his method: he opens a sports market where one outcome costs above 75¢ then goes to pinnacle and bet365 compares bookmakers odds with the contract price on polymarket if bookmakers give the underdog 30-35% and Polymarket values it at 15-20% he enters on the underdog side the difference between the bookmaker line and the contract price bookmakers spend millions on analytics polymarket - crowd betting on emotions he just buys the difference between professional valuation and amateur
ventry46,401 views • 3 months ago

8 open source repos to take 8% off the $8 billion CS2 skin market. -> async scanner across 5 marketplaces in parallel. skinport, csfloat, dmarket, buff163, steam. one full scan in 4 seconds. 12 spreads, $695 potential profit per scan -> async wrapper for skinport API via aiohttp. all endpoints. rate limiting out of the box -> steam trading library. market price history since july 2014, sell/buy orders, inventory management -> simple wrapper for steam market. get_lowest_price + get_volume in two lines. no auth required -> CSFloat arbitrage in python with Flask dashboard. multi-strategy analysis, undervalued items detection ->golang alternative. dmarket + skinport -> buff163. requires 2captcha and discord webhook -> curated list of 50+ repos. CSFloat helpers, buff2steam, SteamDT. start research here -> auto-buy bot for steam market. monitors specific skins and buys when price drops below target before: team on VPS + paid APIs for $50K/year now: git clone, pip install, python like + bookmark - you'll need this when you build your first arbitrage bot through claude
Ventry23,628 views • 1 month ago

built a weather bot. it finds the right signals. i taught it to throw them away 24 hours in paper mode 37 trades. 22% win rate. +$20.65 PnL looks like nothing. until you open the counter-factual block if the bot didn't close on stops and trailing - PnL would be +$1,761.77 money left on table: +$1,866.21 Ankara: actual +$27. if held to resolve - +$1,156 Buenos Aires: actual +$20. if held - +$520 Tokyo: actual +$12. if held - +$51 the strategy works. signals are valid. the bot catches real mispricings my risk management chokes them first iteration. trailing stop too aggressive. take-profit triggers before the market has time to reprice pipeline is alive. rewriting exits this week the best thing that can happen to a new bot is to find a concrete break you can see in the data
Ventry10,853 views • 1 month ago
No more content to load