Сarm1ne's banner
Сarm1ne's profile picture

Сarm1ne

@carm1nee • 18,783 subscribers

✦ Researcher & Analyst ✦ Wall Street / AI / News | @beyond_xai

Shorts

this is the most DANGEROUS repo in trading right now hedge funds pay millions for models that forecast price - someone trained one on 12 BILLION candles and put it on GitHub for 100% FREE it's called Kronos. Yu Shi and 6 co-authors built it, the paper got into AAAI 2026, and the repo has 41k stars first understand how it works: > ChatGPT turns words into tokens & predicts the next word > Kronos turns every candle into tokens & predicts the NEXT CANDLE > it was pre-trained on 12 BILLION candles from 45 exchanges > in the paper's benchmarks it beat the best rival model's forecasts by 93% here is how to run it yourself for FREE: 1. open Claude Code and ask it to clone and install everything 2. load Kronos-small from Hugging Face. 24.7M params, it runs on a normal laptop, no GPU 3. pull the last 512 candles from Binance's free public API. any coin, any timeframe, no account 4. ask it for the next 24 candles, exactly like an LLM finishing a sentence 5. sample 100 futures one by one, not one guess. the spread between them shows how risky the next move really is 6. read the regime: most paths pointing one way means TRENDING, paths scattered both ways means CHOPPY 7. backtest it walk-forward on past windows with real fees before you trust a single path 8. plot the forecast next to real candles and paper trade it first i won't say anything more:

this is the most DANGEROUS repo in trading right now hedge funds pay millions for models that forecast price - someone trained one on 12 BILLION candles and put it on GitHub for 100% FREE it's called Kronos. Yu Shi and 6 co-authors built it, the paper got into AAAI 2026, and the repo has 41k stars first understand how it works: > ChatGPT turns words into tokens & predicts the next word > Kronos turns every candle into tokens & predicts the NEXT CANDLE > it was pre-trained on 12 BILLION candles from 45 exchanges > in the paper's benchmarks it beat the best rival model's forecasts by 93% here is how to run it yourself for FREE: 1. open Claude Code and ask it to clone and install everything 2. load Kronos-small from Hugging Face. 24.7M params, it runs on a normal laptop, no GPU 3. pull the last 512 candles from Binance's free public API. any coin, any timeframe, no account 4. ask it for the next 24 candles, exactly like an LLM finishing a sentence 5. sample 100 futures one by one, not one guess. the spread between them shows how risky the next move really is 6. read the regime: most paths pointing one way means TRENDING, paths scattered both ways means CHOPPY 7. backtest it walk-forward on past windows with real fees before you trust a single path 8. plot the forecast next to real candles and paper trade it first i won't say anything more:

69,493 次观看

Videos

carm1nee's profile picture

Claude Fable 5.2 + TradingView... just turned every candle into a trade decision & it feels ILLEGAL this FREE TradingView indicator turns every closed candle into a JSON packet for AI agents, & Fable 5.2 reads it & calls SHORT, LONG or WAIT on its own i rebuilt the same setup as a paper desk & replayed the last 2 weeks of BTC, ETH & SOL: 19 packets, Fable took 7 trades, said WAIT 5 times & all 6 closed trades ended green (+9.01R) first understand how this indicator works: > you see candles, an AI agent can't read a chart > the indicator breaks every closed candle into JSON: regime, trend, momentum, volume, structure, levels & long/short scores out of 100 > Fable 5.2 reads that packet, makes the call & writes exactly why the trade could fail how to use it with Fable 5.2: 1. paste ai-agent-lens.pine from the repo into the Pine Editor on TradingView & hit "add to chart" 2. every closed candle gets a long & a short score out of 100, built from 5 factors, with no repainting 3. each setup comes with an entry, an ATR stop & 1.5R / 3R targets, drawn on your chart like TradingView's position tool 4. create ONE alert with "any alert() function call" & paste your webhook URL 5. run the webhook receiver from the repo & open a free tunnel to it 6. every packet lands in packets/latest.json the moment the candle closes 7. open Claude Code with Fable 5.2 & give it agent/AGENT_PROMPT.md as its instructions 8. ask for the signal: it returns the decision, confidence, entry, stop, targets, size & exactly why it could fail 9. one alert per chart: add BTC, ETH & SOL & ask it to rank the setups it's FREE & open source, the repo is in the first reply:

Сarm1ne

185,740 次观看 • 2 天前

carm1nee's profile picture

this is f*cking INSANE... citadel & two sigma quants basically LEAKED their hedge fund stack on gitHub zipline’s #1 author now runs equities engineering at Citadel, and pandas was led for years by a Two Sigma managing director They’re called zipline, pandas and Lightweight Charts, and they already have 88,500+ GitHub stars together First, get how it works: 1. zipline is “a Pythonic algorithmic trading library”, “an event-driven system for backtesting”, the engine that ran Quantopian’s crowd-sourced hedge fund 2. pandas gives you “fast, flexible, and expressive data structures”: every price, signal and P&L of the backtest lives in its DataFrames 3. Lightweight Charts is TradingView’s own chart engine, “one of the smallest and fastest financial HTML5 charts”, so every trade lands on your screen Here’s how you can run it for free yourself: 1. Install Claude Code: npm install -g @ anthropic-ai/claude-code, open a new folder and run claude 2. Ask it to install the stack: pip install zipline-reloaded pandas 3. Tell Claude Code: “download BTC/USD daily candles from Kraken’s public API and save them as a CSV” 4. Ask it to register the CSV as a zipline bundle on a 24/7 calendar and run zipline ingest -b btc 5. Tell Claude Code: “write a zipline strategy for BTC with a trend signal, a stop and a target” 6. Backtest it: zipline run -f strategy. py --start 2024-1-1 --end 2025-12-31 -b btc -o results.pickle 7. Ask Claude Code to open results.pickle with pandas and print the returns, the drawdown and every trade 8. Build your chart screen: npm install --save lightweight-charts, then a page that draws the candles, entries and exits 9. Put it on your big monitor, and let Claude Code rerun the backtest every time you change the strategy i won’t say anything more:

Сarm1ne

100,161 次观看 • 9 天前

carm1nee's profile picture

Watch what happens when emotions override common sense Louis on Deal or No Deal turned down $333,000 Two cases left, his and one other, $750,000 or $5 The banker offered $333,000 to walk away, he said no deal His mother, brother, and the entire audience begged him to take the money His wife said, "Keep playing" He opened his case → $5 Nobel Prize-winning economist Richard Thaler studied 151 contestants on this show across three countries And published the results in the American Economic Review // > He found two patterns that destroy people's decision making: • Break-even effect → after a series of bad outcomes contestants take bigger risks Trying to get back to where they started, the more they lose the more reckless they get • House money effect → after a series of good outcomes contestants also take bigger risks Because the gains feel like "free money" that isn't really theirs Both patterns are irrational, both lead to yhe same result worse decisions under pressure Thaler's team said the show "almost appears to be an economics experiment rather than a TV show" // Thaler studied a TV show, but he described every Polymarket trader who's ever blown up Since I’ve found myself in this situation more than once I feel it’s important to say something everyone should keep reminding themselves of No matter how obvious or simple it may sound, your losing streak should not influence your future decisions The exact same principle applies to winning streaks as well the clip below is the moment she turned down $333,000 & watch what happens when emotions override math in real time ↓

Сarm1ne

2,216,431 次观看 • 7 个月前