Technomancy AI

You can just do things, relates to 'you can code the life you want' and @NousResearch and @Teknium have opened the door through Hermes.

Announcing Cofounder 2: Run an entire company with agents. It's the infrastructure for the one person billion dollar company - orchestrating agents across engineering, sales, marketing, ops, and design. (and yes that's my real grandma in the video)

Import AI 455: Automating AI Research jack-clark.net/2026/05/04/imp…

POLYMARKET JUST CONFIRMED IT'S A MATH TRAP You bet on an outcome. You’re sure of the result. The market moves in your direction, but you still lose money. Here's the part that should scare you: → 41% of all conditions on Polymarket are mathematically broken → The median logic deviation is $0.60 per dollar → Your "intuition" is useless against Bregman Projection → Standard bots are just "exit liquidity" for the elite Top traders use the Frank-Wolfe Algorithm to turn exponential complexity into pure profit. They aren't "guessing", they are projecting the market state onto an arbitrage-free manifold. They call it "market efficiency." We call it a retail wipeout.

Python made AI accessible. Rust can make parts of AI understandable. That’s the bet behind Category Theory for Tiny ML in Rust. We’re building tiny ML systems from first principles using: Rust types typed transformations composition training loops category theory as an engineering tool Not abstraction cosplay. Executable structure. Working draft. Public feedback welcome.

i build a polymarket trading bot. took an old abandoned repo, rewrote 90% of it. name: zostaff repo: github.com/zostaff/poly-t… mit. python 3.12+. py-clob-client with eip-712 signed orders. usdc.e on polygon. three strategies in the box: AI Directional (llm via openrouter, any model), Safe Compounder (pure edge math, no llm), Beast Mode (aggressive, kept for comparison). most "ai bots" for polymarket on github are abandoned templates with hardcoded losing strategies. this one ships with a streamlit dashboard, sqlite telemetry on every trade, fallback chain across llm models, circuit breaker at 15% drawdown, quarter-kelly default, paper trading mode by default - until you flip LIVE_TRADING_ENABLED=true, nothing goes live. the readme has a section called "Lessons From Live Trading". real numbers from real losses: three-quarter kelly destroys capital on a 45% win-rate strategy. category discipline beats ai confidence every single time. a 50% drawdown limit isn't a limit. more trades without edge = faster path to zero. this is not a turnkey bot. nobody on polymarket is selling a turnkey bot that works. it's a toolkit with three example strategies you fork to fit your own edge. if you don't have an edge - no llm will save you. if you do - run paper mode for a week before flipping live.

The more I look into the $GME deal... it's 100% possible. Ryan Cohen might be able to pull this off. If $GME price explodes above key levels they will be able to raise billions in cash, clear all their debt, and grow their market cap enough to make the buyout possible. Key Levels & Instruments: 1. $32 - GameStop Warrants ($GME.WS) 2. $38 - Convertible Notes (Zero Coupon) 3. $50 - Ryan Cohen Pay Package 1. 59.15M $GME October Warrants become exercisable above $32. If all warrants are exercised they raise $1.9B USD while share count grows from 448M > 510M, or $16.3B MC Dilution here isn't harmful. Warrant holders believe in the stock going higher and are willing to pay a premium to the current price in order to exercise. 2. Two tranches of zero coupon convertible notes were issued to buy $BTC. The convert and expiry is split between 2030 & 2032. These notes are economically in the money above $29-30, which is their convert. 2032 Notes convert: $28.91 2030 Notes convert: $29.85 Conversion is contingent on multiple specific stipulations (Note - this is extremely important later). Note holders can't convert to shares until $GME trades above 130% of the conversion price for 20 of 30 consecutive trading days in the prior quarter. Our trigger prices... 2032 Notes trigger: $37.58 2030 Notes trigger: $38.81 If we hold above $39 and these notes convert to shares, $4.2B in convertible debt is wiped from the balance sheet. Share count would grow 510M > 653M pushing the market cap of $GME to $25B+. $GME would be $25B with ~$11B cash on hand with zero debt... all the sudden buying $EBAY at $45B is possible. 3. Ryan Cohen's Historic $35B Pay Package will become official with a vote in June. The top milestones include $100B market cap and $10B EBITDA for $GME where he would vest 171.5M stock options @ $20.66 strike. Assuming $GME takes over $EBAY he could hit $50-70B market cap and ~$3B of EBITDA year one. If Ryan were to vest all of his shares before an ebay deal (very unlikely) it would add $3.54B in cash and bring the fully diluted share count to 825M ($41B+ MC). Now that you understand the instruments, let's take a step back to understand their plan. Why did they structure it this way...? In my opinion they are designing a squeeze on delta neutral bondholders. Bondholders are in a very unique spot. Once $GME is in the $32-$37 price range they are forced to add short positions to stay delta neutral even if they are below their trigger event. On issuance 65M-80M shares were shorted @ $22 by bondholders to go delta neutral on their positions. If GME is above $32, the $4.2B would need to short approximately 115M-125M total shares ($3.68B) in order to stay delta neutral. The shorts would need to 2x very quickly. Borrowing costs would skyrocket, available shares would disappear, and eventually bondholders will likely struggle to cover their positions as they tread water for 3+ months while their stipulations for exercise were satisfied. Here is where it gets interesting... in order for these books to stay delta neutral they arent just shorting the delta of $GME shares, they also need to short the delta of $GME warrants... which are massively deflationary as they are being exercised. In the $32-40 range for $GME they would be required to short 11-14M $GME+ Warrants of the 59M total available, its extremely reflexive. Quick reminder these bonds expire in 2030 & 2032, they could be forced to short with massive borrow rates for months or even years on both shares & warrants. $32+ warrants will vest, gamma will skyrocket, borrow fees jump on both assets, bondholders will rush to hedge delta, and eventually they run out of warrant supply to cover their positions. 14M of 59M Warrants are already held by Convertible Note holders today as a hedge, what happens if prices spike above $32 quickly? There wont be enough supply to hedge and the full squeeze begins. There are multiple angles here where $GME grows their market cap to $16-32B, grows their cash by billions, and can complete the takeover. As reported by WSJ, it's also extremely likely that they have already accumulated a stealth position in $EBAY. There is a decent chance $GME already owns ~3-5% of $EBAY through share and options purchases. I love the upside here... long $GME shares & warrants. Power to the Players! 🎮 (All Data and Probabilities for Thesis Were Built by Axion AI - axion.eternis.ai/invited/SCS)

A 27-year-old Chinese developer made $246,000 from 127 predictions with zero social media presence No Twitter. No Telegram. No public strategy. Just a wallet that kept opening positions seconds after news wire drops before the headline hit X. Someone reverse-engineered the architecture. 3 data layers. 15 Claude pipelines. A signal system that reads what news implies, not what it says. The edge was never the information. It was the speed of interpretation. And now the blueprint is public

I met a Jane Street trader in Zurich He called Polymarket "gambling for quants." That was before he saw the log-odds graph. He quit Jane Street three weeks later. He'd been pricing options for 8 years. Spotless record. Laughed when I opened the laptop. Stopped when he saw the numbers. $11,400 profit. 19 days. Sharpe 2.31. Copytrade here: @trackmind" target="_blank" rel="nofollow noopener">kreo.app/@trackmind He leaned in. No words. Scrolled the cluster map. github.com/warproxxx/poly… 86M trades scraped. One Claude prompt: scan 14k wallets, rank by 70%+ win rate, profit. Top 20 crushed the bottom 13k combined. Filtered for crypto domains only. 81% edge. github.com/Polymarket/age… 3 agents vote: news lag, whale flow, crowd bias. Two agree - Kelly size. One alone - half. All disagree - no trade. "This isn't trading. It's Bayesian inversion on market noise." github.com/Polymarket/pol… 500 markets - 35 survivors. Exits at 85% capture or 3x volume spike. Never holds to settlement. Day 3: +$890 Day 10: +$4,200. Sports dead, crypto alive. Day 19: +$11,400. 214 trades. "$25/month Claude. $5 VPS. Free repos." He rebuilt it overnight. DM'd me one line: "+$1,100 first run. How did I miss the sensor math?" Quit Jane Street. Now full-time on Polymarket. The repos are free. The data is public. The wallets are grinding right now. Scroll if you're retail. Build if you're next.

I did a thing and there is no turning back now. Time to cook and push the limits of my creativity. If anyone else has been building with local LLMS, let me know, would love to hear your takeaways, tips so far.

A small group of quants wrote one formula, launched it on Polymarket and quietly took $20m in a year Their secret fits in one equation: A = e^(-λ₁T - λ₂y - λ₃R) · (1 + γ₀y · 1_{y>0.7}) · e^(-φ₀R²) · sgn(σₕ - κC)(1-C) These aren't random symbols each factor is a filter: > how stale the position is over time > how overheated the market is by returns > how much volatility justifies the entry > which direction to look at all Only if all four factors align the trade exists No "I feel like it". No "I believe in this asset" Just a number positive or not But there is one tool that changes the game completely Top traders on Polymarket are active simultaneously across multiple high-probability markets every day Each trade brings a small win on its own But combining 5-6 of these trades into one parlay you can increase your return from 80% to 400% ROI instead of five small wins That's exactly what this bot does automatically Try here: t.me/poly_parlay_bo…

My M4 Max MacBook gets 3,756,165 tok/sec in pure C, compared to ~50,000 tok/sec with the FPGA. Try it yourself: github.com/AlexCheema/tal…

There is no better teacher than rock bottom.

10k clones in last 14 days They can't do it like me🤷‍♂️ Who needs a stacked dev

JUST TWO repositories that can change your life. -> For a more serious research to live setup. github.com/nautechsystems… - is one of the strongest open source trading engines to look at. It supports research, deterministic backtesting, live execution, and includes a Polymarket integration for loading market data and building proper strategy workflows. -> AI WORKFLOW github.com/randomness11/p… - is an AI powered trading framework for Polymarket. You define the strategy in plain English, and the framework handles market analysis, risk rules, position sizing, paper trading, backtesting, and real time data. These tools are a must have in your arsenal.

Here's what 96 concurrent @NousResearch Hermes Agents (using 382,745,618 tokens over 171,136 API calls to deepseek-v4-pro) can generate for you in less than 3 days. From your M4 Macbook Air 24GB, on hotel wifi! 81% cache hit rate in case you're wondering!

1/ Absolutely extraordinary must-read from my friend @scmallaby Every book of his from Greenspan (the man who knew) to hedge fund masters (more money than god) to best book on VC (the power law) captured with insane access + clarity the stories nobody else knew NOW...

Cancelled both my Claude Code Pro and ChatGPT Pro for this. Kimi K2.6 is just as good for my side projects as Opus or GPT 5.4 were. The price for this is crazy low, and there are a bunch of models I can try (like DeepSeek). Bonus: I'm moving away from building everything on Claude Code - now that both @opencode and @cursor_ai have their SDKs open, I feel I can rebuild the agentic workflows I built for Claude Code in a more platform-independent manner.