The timing of this release seems interesting? ๐ค๐
In less than 48 hours, GTA VI finally dropsโand if you own an S&P 500 or total-market index fund in your 401(k), you probably own a tiny piece of it.
โParent company Take-Two Interactive ($TTWO) is forecasting $3.35B to $3.5B in bookings just for the OctโDec launch quarter (Stock Titan), with Newzoo projecting the game alone lifts global console full-game spending by 17.5% (Newzoo).
โHere's who actually owns the biggest slices of $TTWO (Investing, Boardroom Alpha):
โข Vanguard: ~11% combined (incl. index funds like VTI)
โข BlackRock: ~9.8%
โข Saudi PIF / Savvy Games Group: 6.1%
โข State Street: 5.8%
โข Plus the platform toll collectors: Sony (PS5) and Microsoft (Xbox) take a cut of every digital/console copy sold at launchโwith no PC release dated yet (Xbox Wire)
โAre you buying GTA VI at midnight Thursdayโor watching $TTWO stock instead?
โ#GTA6 #GTAVI #TakeTwo #StockMarket #Gaming #Investing
23
Jensen or no Jensen. $NVDA IS STILL A BEAST ๐
๐จ NVIDIA Q3 EARNINGS JUST DROPPED โ AND THE NUMBERS ARE UNREAL ๐จ
In its first report since Jensen Huangโs shock exit, $NVDA just proved the AI engine hasn't slowed down a single inch:
โข Revenue: $115.8B (+103% YoY) vs $109.1B est ๐ข
โข EPS: $2.68 (+111% YoY) vs $2.47 est ๐ข
โข Data Center: $107.6B (+110% YoY) ๐
โข Q4 Guidance: $130.0B (smashing $125B est) ๐ฅ
THE CATCH WALL STREET IS WATCHING: ๐
1๏ธโฃ Margin Squeeze: Q4 gross margins are guided down to 71%โ72% (from 74.9%) as HBM4 memory costs spike and next-gen Vera Rubin racks ramp.
2๏ธโฃ Massive Supply Bet: Supply commitments surged 40% QoQ to a staggering $390 BILLION to lock in capacity through 2028.
3๏ธโฃ The Overhang: Despite 100%+ growth, $NVDA is down ~35% from highs and trading under 20x trailing earnings amid the CEO search, SEC probe, and China export scrutiny.
THE BOTTOM LINE:
Interim CEO Colette Kress confirmed next-gen Vera Rubin is shipping and off to an "explosive start" with Meta, OpenAI, and Anthropic.
With $227.3B left in share buybacks, is $NVDA the biggest bargain in techโor are shrinking margins a warning sign? ๐๐
#NVDA #NVIDIA #AI #StockMarket #Earnings
138
Phoenix 5D๐ฅ retweeted
An estimated 100 million American homes are running on a router that nobody has logged into since the day it was plugged in.
Not once. Not to change a password, not to check for updates, not to look at a single setting.
And those same homes are paying $70, $80, even $100 a month for "fast" internet.
Then the video call freezes in the middle of a work meeting. Netflix buffers on a Friday night. The kids' games lag. The Wi-Fi "just feels slow."
So what does everyone do? They call the provider and yell. They upgrade the plan. They buy a $140 Wi-Fi extender. One household I know replaced their router twice in 18 months and spent more than $400 doing it.
None of it fixed the problem.
Not because the internet was too slow. Not because the new router was bad. Because every router ships on factory settings built for an "average home on an average day," crowded onto the same Wi-Fi channel as every neighbor, steering devices onto the wrong band, using the slowest default DNS, running firmware that is months or years out of date, and treating your work video call the same as someone's 40GB game download.
Then I sat down for 10 minutes in the admin panel. I didn't change the plan. I didn't buy new hardware. I didn't call the provider.
I changed 6 settings: channel, band steering, DNS, firmware, QoS, and one buried security toggle most people never touch.
Same router. Same plan. Same house. [Speeds nearly doubled / insert your real before and after speed test numbers].
"Your router was never too slow. It was never told how to manage your home. And nobody, not your provider, not the store, not the box it came in, was ever going to tell you the fix was free."
Here are the 6 settings I changed (and exactly how to do it) ๐งต
Phoenix 5D๐ฅ retweeted
๐จ BTC BREAKDOWN: The $65K Trap & The Liquidity Sweep
1๏ธโฃ ๐ฏ Executive Conviction & Strategic Bias
Bitcoin is currently caught in a violent "liquidity vacuum." Price has broken below critical structural support, triggering a cascade of long liquidations. However, the tape is flashing a massive Bullish Spring (SMC Liquidity Sweep) with 86% confidence. We are seeing classic institutional accumulation behavior: price is being pushed down to flush out weak retail longs before a violent re-claim of the range. The bias is Neutral-to-Bullish on the dip, but strictly waiting for a reclaim of the $66,000 zone to confirm the sweep.
2๏ธโฃ ๐ข Fundamental Health & Valuation Matrix
* Market Cap $1.31 Trillion
* Asset Class: Decentralized Digital Gold / Smart Contract Platform
* Supply Dynamics: Fixed 21M cap. Post-halving supply shock is currently absorbing institutional ETF inflows.
* Macro Catalyst: Spot ETFs (IBIT, etc.) continue to act as the primary bid, but corporate treasury demand (MSTR) is currently paused due to price weakness.
3๏ธโฃ ๐ฌ Market Physics & Technical Structure
* Spot Price: $64,852.85 (-2.75% today)
* 20-Day MA: $75,825.57 (Overhead Resistance โ +16.9% above spot)
* 50-Day MA: $80,151.06 (Major Overhead Resistance โ +23.6% above spot)
* RSI (14): 21.6 (Deeply Oversold โ Extreme selling pressure)
* **RVOL**: 0.94x (Volume is actually *below* average, indicating a lack of aggressive new selling, just passive liquidation flow)
* ATR (14-day): $2,821.25 (Expected daily swing range is ~$2.8k)
* Bollinger Latent Energy: 35.0% Compression (Normal volatility, coiling for expansion)
* Key Support: $61,946.17 (Calculated Technical Support)
* Key Resistance: $81,955.00 (Calculated Technical Resistance)
4๏ธโฃ ๐ Smart Money Flow & Institutional Tape
* Dealer Gamma Exposure (GEX): **Zero-GEX Flip @ $39.84** (Note: This is relative to the $40 IBIT ETF price, not BTC spot. For BTC, dealer hedging is currently amplifying the downside as price falls).*
* Coinbase Premium Index: -0.0571% (MILD_BEARISH). US spot demand is slightly lagging global perps.
* Hyperliquid Perps OI: $6.92B. High open interest means the liquidation cascade is still active. If OI drops rapidly while price stabilizes, the flush is complete.
2026 YTD CRYPTO PERFORMANCE LEADERBOARD (As of Nov 14)
The narrative shifted. AI Agents & DePIN ate the world. Memes died in Q1. L1s flatlined. The capital rotated into revenue-generating protocols.
#1: VIRTUAL (Virtuals Protocol) โ +1,420% YTD
The "Shopify for AI Agents" thesis played out perfectly.
Jan 1 Price: $0.42 -> Current: $6.38
Market Cap: $4.2B (Top 25)
Why it won: First-mover on Base for tokenized AI agents. Every major KOL/brand launched an agent here. Revenue share model (buyback & burn) created a flywheel TradFi understands.
Key Metric: $180M annualized protocol revenue (up from $2M Jan 1). P/E ratio ~23x โ cheapest major in crypto.
#2: RENDER (RNDR) โ +380% YTD
The "Nvidia of Crypto" narrative finally monetized.
Jan 1: $4.80 -> Current: $23.10
Catalyst: Apple Vision Pro enterprise adoption + Hollywood strike resolution drove GPU demand parabolic. Migration to Solana cut fees 99%.
Moat: Actual paying customers (Disney, EA, architecture firms). Not speculation โ invoices.
#3: HELIUM (MOBILE/IOT) โ +310% YTD
DePIN proved it prints cash.
Jan 1: $3.20 -> Current: $13.15
Catalyst: T-Mobile & Telefonica offload partnerships went live. 120k+ 5G hotspots deployed. Subscribers > 2M.
Killer Stat: $85M ARR from carrier offload fees. First DePIN with GAAP revenue > token emissions.
THE GRAVEYARD (Narratives that failed)
SOL: -12% โ "Ethereum Killer" fatigue. MEV extraction > user value. Capital rotated to Base/Arbitrum L2s.
ETH: +18% โ "Ultrasound Money" broke. Staking yield < T-bills. L2s captured all fees. ETH became a settlement layer, not an investment.
DOGE/SHIB/PEPE: -65% avg โ Retail liquidity evaporated. No new buyers. AI agents replaced memes as the "casino."
AVAX/NEAR/APT: -30% avg โ VC unlocks crushed bid. Zero sticky users.
THE 2026 LESSON
Price follows Fee Revenue, not Narrative.
The winners weren't "better tech." They were businesses with:
1. Real customers (Web2 enterprises, telcos, brands)
2. Token accrual mechanics (Buybacks, burns, revenue share)
3. Regulatory clarity (DePIN/AI infra != Securities)
WHAT'S NEXT (Q4 2026 Watchlist)
1. AERO (Aerodrome) โ Base's liquidity hub. $40M fees/quarter. Ve(3,3) flywheel spinning.
2. HIVE MAPPER (HONEY) โ Google Maps killer. 15M km mapped. Uber/DoorDash pilot deals signed.
3. AKASH (AKT) โ Permissionless cloud. AWS price parity achieved. Enterprise onboarding Q4.
Bottom line: If your 2027 bag doesn't have a P&L statement attached to the token, you're gambling. Not investing.
Which 2026 winner did you catch? Which narrative are you fading for 2027?
#Crypto #DePIN #AIAgents #Virtuals #Render #Helium #2026Review
1
207
We know what happens next. Sweep lower liquidity than short squeeze ?
๐ค๐ฐBitcoin just saw $1.6 BILLION in positions wiped out in 24 hours. 90% were longs betting price would rise. Here's what liquidation data actually tells you โ and the 'magnet' effect, explained simply. ๐งต
First, the facts. BTC traded ~$65,800 this morning (Coinbase), down from ~$80K ten days ago. Friday: ~$1.6B liquidated crypto-wide. The single largest: one $31M long force-closed at $72,250.
What IS a liquidation? A forced sale. A trader uses leverage to bet price rises. If price falls far enough, the exchange closes the position automatically. That forced selling pushes price lower โ liquidating the NEXT trader. That's a cascade.
Now the 'magnet.' Exchanges don't publish where positions sit. Models estimate it from open interest + typical leverage. Where many positions would be force-closed at one price, that's a cluster. Price tends to drift toward clusters. That's the magnet.
This morning the nearest magnet BELOW price was ~$65,450 (longs). The nearest ABOVE was ~$67,900 (shorts), 4x further away. And accounts are still ~74% long. The closest magnet usually pulls first. This morning, it was underneath.
Plot twist: the BIGGER fuel pool sits ABOVE โ ~$1B in shorts vs ~$320M in longs nearby. If price ever reclaims ~$68K, those shorts get force-closed = forced BUYING. That's a short squeeze. Distance decides which magnet pulls first.
One more lesson: RSI hit 24.7 โ deeply 'oversold.' Oversold does NOT mean buy. It means the move is stretched. In a cascade, RSI can stay oversold for days. Forced selling ending is not the same as willing buying returning. US BTC ETFs: -$185M Friday.
Takeaway: liquidation maps show where LEVERAGE sits, not where price WILL go. They explain the last move better than they predict the next one. Watch the magnets, respect the trend. A flush ending is not a bottom.
Sources: Coinbase (price + candles), CoinGlass (liquidation totals), SigBTC / ByKaranteli / AlphaFinance (heatmap models โ estimates, not orders). Data as of ~10:00 ET, Nov 14, 2026. Heatmaps reprice every few minutes. Educational only โ not financial advice.
44
Stop burning your frontier model quota on routine file edits. ๐
โGoogle Antigravity just dropped Gemini 4 Hydrogen + Orchestratorโand it solves the biggest bottleneck in agentic coding:
โThe problem:
Frontier models (Carbon / Argon) are brilliant architects, but running them for every file edit and test loop is slow and expensive. Fast models are cheap, but lose the plot on complex repos.
โHow Orchestrator fixes it (The Planner-Worker Loop):
1๏ธโฃ Read-Only Planning: Carbon or Argon maps your repo, resolves edge cases, and writes a step-by-step blueprint.
2๏ธโฃ High-Speed Execution: Gemini 4 Hydrogen takes the wheelโbatching parallel file reads, writing code, and running tests 2.8x faster with up to 23% fewer tokens.
3๏ธโฃ Auto-Escalation: If Hydrogen hits a wall (like a test failing twice), it hands control back to Carbon/Argon to debug the plan before resuming.
โWhy Hydrogen doesn't break your codebase:
โข 12x fewer misaligned tool shortcuts
โข 81% less agentic overreach
โข 1M context window at $1.50 / $7.50 per 1M tokens
โHow to try it right now in Antigravity CLI:
agy --model "Orchestrator (Carbon)"
โIntelligence for the architecture. Speed for the grunt work. โก
โ#AIAgents #Gemini4 #GoogleAntigravity #Coding #BuildInPublic
๐ค Made with AI
1
114
Phoenix 5D๐ฅ retweeted
You can basically build a tiny AI podcast studio on your laptop now with just this stack:
1. Whisper
github.com/openai/whisper
2. Demucs
github.com/facebookresearch/โฆ
3. Chatterbox
github.com/resemble-ai/chattโฆ
4. AudioCraft
github.com/facebookresearch/โฆ
5. Auto-Editor
github.com/WyattBlue/auto-edโฆ
Save this and set it up when you have time
Best Uncensored AI models now :
Overall : Qwen3.8-27B Abliterated
Small GPU : Gemma 4 12B Abliterated
Fastest : Gemma 4 26B-A4B Abliterated
Agentic : Qwen3.6-35B-A3B Uncensored
Writing : Mistral Small 3.2 Abliterated
No surgery : Hermes 4.3-36B
Image : FLUX.1-dev
Video : Wan 2.2
Ceiling : GLM-5.1 Abliterated (754B, not for home)
Where to get them :
all LLMs โ huggingface.co/huihui-ai (search the name, grab the GGUF)
Hermes โ huggingface.co/NousResearch
FLUX / Wan โ ComfyUI + the model's HF page
Setup :
ollama pull hf.co/<repo>:Q4_K_M
ollama run <model>
/set nothink
that's it.
no API key, no account, no card, just few clicks setup
more on how to setup below:
Here's 4 steps to run an uncensored AI on your own GPU for $0:
I run a local abliterated Gemma 4 for anything I don't want leaving my machine
By the end, you'll know how to:
- run a model which has no refuses
- plug it into Claude Code
- do it all for $0, no API key, no account
So, let's go step by step:
1. Install Ollama
curl -fsSL ollama.com/install.sh | sh
you want a recent version, older builds don't speak the Anthropic API and don't support Gemma 4
2. Pull the model
ollama pull igorls/gemma-4-12B-it-heretic-GGUF
the biggest mistake people make here is grabbing a random "uncensored" tag with no quality numbers
this one is abliterated with Heretic. refusals went 99/100 โ 0/100 at a KL divergence of 0.028 from the original, so you keep almost all of the base model's capability
one thing: it's still Google's Gemma license underneath, fine for personal use, but read the terms before you build a product on it
3. Turn off thinking (the part every other guide skips)
Gemma 4 is a hybrid thinking model and the abliteration targets non-thinking mode
leave thinking on and the reasoning block eats your token budget and the answer gets cut off
ollama run igorls/gemma-4-12B-it-heretic-GGUF
/set nothink
if you're calling the API instead, add "think": false to the request
4. Plug it into Claude Code (OPTIONAL)
ollama launch claude --model igorls/gemma-4-12B-it-heretic-GGUF
or manually:
export ANTHROPIC_BASE_URL=http://localhost:11434
export ANTHROPIC_AUTH_TOKEN=ollama
claude --model igorls/gemma-4-12B-it-heretic-GGUF
two gotchas:
- bump the context first. Ollama defaults to 4k and Claude Code's system prompt alone is bigger than that, so the model never sees its tool instructions. restart the server with OLLAMA_CONTEXT_LENGTH=32768 ollama serve (setting it in your shell does nothing if Ollama is already running as a service). this needs more VRAM than the 8GB below
- Gemma 4 tool calling is still rough, you should expect some failed edits, imo for serious agentic work use qwen3.5 or glm-4.7-flash instead
What you need:
- 8-12GB VRAM for Q4 (12B)
- ~8GB of disk for Q4, ~13GB for Q8
- more of both if you want a big context
honest take:
- "abliterated" means refusals removed, not quality added. it's still a 12B
- fewer refusals โ smarter. for real coding work I still use cloud models
- you own the weights, your prompts never leave the box. that's the actual win
*IMPORTANT: don't use it for anything illegal or harmful. that's on you, not the model
now i'm collecting a list of every uncensored model i'm adding to my local stack and which tasks each one replaces for me
also planning a big "Uncensored 101" article, still figuring out all the angles
save it so you don't lose it
Phoenix 5D๐ฅ retweeted
NVIDIA Open-Sourced Auto Ontology, a Semantic Layer for AI Agents!
Auto Ontology is a Python framework for building semantic, graph-based reasoning layers over data.
AI agents can already query databases and retrieve documents.
The harder problem is making sure they understand what the data actually means.
A metric like โrevenueโ might be defined differently across two systems. The same customer can appear under different identifiers. A generated SQL query can be syntactically correct and still use the wrong business definition.
Auto Ontology is built for that layer in between.
It can build a semantic catalog over enterprise metadata so agents can reason about entities, metrics, relationships, lineage, and where trusted definitions come from.
The workflow is roughly:
Enterprise data + metadata
โ Auto Ontology
โ Semantic catalog / knowledge graph
โ AI agent
โ Grounded query + provenance
It can generate ontologies from existing metadata, connect to semantic models you already have, reconcile inconsistent definitions across systems, and expose that shared context through MCP, CLI, or API.
For structured data, it can also translate natural-language questions into SQL while validating the query against the ontology and certified definitions before returning the result.
So instead of an agent simply asking:
โHow much revenue did we generate last month?โ
it can understand which revenue definition to use, which tables contain the right data, how those entities relate, and where the final answer came from.
Auto Ontology currently supports relational sources including Databricks, Postgres, Snowflake, and DuckDB.
It also exposes an MCP server, which means agents can query the same semantic layer directly.
For enterprise agents, retrieval alone is not enough.
They also need a shared understanding of what the underlying data actually means.
100% open source.
Iโve shared the GitHub repo in the comments!
NVIDIA Open-Sources Auto Ontology, a Semantic Layer for AI Agents
Your boss asks the AI agent: "What was revenue last quarter?"
It answers in 3 seconds. Clean number. Nice chart.
But it used the wrong "revenue" column. ๐ก๐ผ๐ฏ๐ผ๐ฑ๐ ๐ป๐ผ๐๐ถ๐ฐ๐ฒ๐.
That's the problem NVIDIA's new open-source library, Auto Ontology, goes after. Its README puts it plainly:
"โฆ๐ข ๐ณ๐ฆ๐ท๐ฆ๐ฏ๐ถ๐ฆ ๐ฎ๐ฆ๐ต๐ณ๐ช๐ค ๐ข๐ฏ๐ด๐ธ๐ฆ๐ณ๐ฆ๐ฅ ๐ง๐ข๐ด๐ต๐ฆ๐ณ ๐ช๐ด ๐ธ๐ฐ๐ณ๐ต๐ฉ ๐ฏ๐ฐ๐ต๐ฉ๐ช๐ฏ๐จ ๐ช๐ง ๐ต๐ฉ๐ฆ ๐ฏ๐ถ๐ฎ๐ฃ๐ฆ๐ณ ๐ช๐ด ๐ธ๐ณ๐ฐ๐ฏ๐จ."
๐ช๐ต๐ฎ๐ ๐ถ๐ ๐ถ๐:
A "context layer" that sits between an AI agent and your database. It tells the agent where the data lives, what each field means, and whether to use it.
๐๐ผ๐ ๐ถ๐ ๐๐ผ๐ฟ๐ธ๐, ๐๐๐ฒ๐ฝ ๐ฏ๐ ๐๐๐ฒ๐ฝ:
โข ๐๐ถ๐๐ฐ๐ผ๐๐ฒ๐ฟ: it reads your database metadata and drafts an ontology, a map of business terms like "active customer"
โข ๐๐ฟ๐ผ๐๐ป๐ฑ: agents use those terms instead of guessing from raw table names
โข ๐ฉ๐ฎ๐น๐ถ๐ฑ๐ฎ๐๐ฒ: it checks generated SQL against certified definitions
โข ๐๐
๐ฝ๐ผ๐๐ฒ: agents reach it through MCP, a CLI or an API
โข ๐๐ฒ๐ฎ๐ฟ๐ป: certified corrections go back into the map
๐๐ผ๐ฟ ๐๐๐ฒ๐ฟ๐:
Ask a question, get ๐๐ต๐ฒ ๐ฎ๐ป๐๐๐ฒ๐ฟ, ๐๐ต๐ฒ ๐ฆ๐ค๐ ๐ถ๐ ๐ฟ๐ฎ๐ป, ๐ฎ๐ป๐ฑ ๐๐ต๐ฒ ๐ฟ๐ผ๐๐. You can check the work, not just trust it.
๐๐ผ๐ฟ ๐ฏ๐๐ถ๐น๐ฑ๐ฒ๐ฟ๐:
โข Apache 2.0, runs with Docker Compose or Helm
โข Works with Databricks, Postgres, Snowflake and DuckDB
โข Defaults to NVIDIA's Nemotron models, but ๐ฎ๐ป๐ ๐ข๐ฝ๐ฒ๐ป๐๐-๐ฐ๐ผ๐บ๐ฝ๐ฎ๐๐ถ๐ฏ๐น๐ฒ ๐ฒ๐ป๐ฑ๐ฝ๐ผ๐ถ๐ป๐ ๐๐ผ๐ฟ๐ธ๐
โข The MCP server has ๐ต ๐ฟ๐ฒ๐ฎ๐ฑ-๐ผ๐ป๐น๐ ๐๐ผ๐ผ๐น๐. There is deliberately no tool to browse raw tables, so agents must go through the business terms
๐๐ ๐ฏ๐๐ถ๐น๐ฑ๐ ๐ผ๐ป ๐ฒ๐ฎ๐ฟ๐น๐ถ๐ฒ๐ฟ ๐๐ผ๐ฟ๐ธ:
โข Cube and dbt MetricFlow: semantic layers you write by hand in code
โข Wren AI: an open semantic layer for AI agents, with a context layer built partly by agents
โข Snowflake's Semantic View Autopilot: also drafts semantic models with AI, but only for Snowflake
What's different here: ๐ผ๐ฝ๐ฒ๐ป ๐๐ผ๐๐ฟ๐ฐ๐ฒ, ๐๐ผ๐ฟ๐ธ๐ ๐ฎ๐ฐ๐ฟ๐ผ๐๐ ๐ณ๐ผ๐๐ฟ ๐ฑ๐ฎ๐๐ฎ๐ฏ๐ฎ๐๐ฒ๐, ๐ฎ๐ป๐ฑ ๐ฏ๐๐ถ๐น๐ ๐ณ๐ผ๐ฟ ๐ฎ๐ด๐ฒ๐ป๐๐ ๐๐ต๐ฟ๐ผ๐๐ด๐ต ๐ ๐๐ฃ.
๐ง๐ต๐ฒ ๐ต๐ผ๐ป๐ฒ๐๐ ๐ฝ๐ฎ๐ฟ๐:
โข ๐ง๐ต๐ฒ "๐ต๐ฌ% ๐ณ๐ฎ๐๐๐ฒ๐ฟ" ๐ฐ๐น๐ฎ๐ถ๐บ ๐ถ๐ ๐ก๐ฉ๐๐๐๐'๐ ๐ผ๐๐ป ๐๐ฒ๐๐๐ถ๐ป๐ด, with no baseline published
โข ๐ก๐ผ ๐ฎ๐ฐ๐ฐ๐๐ฟ๐ฎ๐ฐ๐ ๐๐ฐ๐ผ๐ฟ๐ฒ๐ ๐ฎ๐ฟ๐ฒ ๐ฝ๐๐ฏ๐น๐ถ๐๐ต๐ฒ๐ฑ. NVIDIA ships an eval harness (BIRD, 1,534 questions) but no results
โข That harness was tested with ๐๐น๐ฎ๐๐ฑ๐ฒ ๐ข๐ฝ๐๐ ๐ฐ.๐ด, ๐ป๐ผ๐ ๐ก๐ฒ๐บ๐ผ๐๐ฟ๐ผ๐ป
โข ๐ช๐ถ๐๐ต๐ผ๐๐ ๐๐ฎ๐๐ต๐ถ๐๐ผ๐ฟ๐ฝ ๐ฉ๐ฎ๐๐น๐, ๐ฑ๐ฎ๐๐ฎ๐ฏ๐ฎ๐๐ฒ ๐ฝ๐ฎ๐๐๐๐ผ๐ฟ๐ฑ๐ ๐ฎ๐ฟ๐ฒ ๐๐๐ผ๐ฟ๐ฒ๐ฑ ๐๐ป๐ฒ๐ป๐ฐ๐ฟ๐๐ฝ๐๐ฒ๐ฑ in Postgres (per the README)
โข It's young: ๐ฎ๐ฏ๐ผ๐๐ ๐ณ๐ฑ๐ฌ ๐๐ถ๐๐๐๐ฏ ๐๐๐ฎ๐ฟ๐, and NVIDIA isn't accepting outside contributions
๐ง๐ต๐ฒ ๐ฏ๐ถ๐ด๐ด๐ฒ๐ฟ ๐ฝ๐ผ๐ถ๐ป๐: the hard part of AI analytics isn't writing SQL. ๐๐'๐ ๐ฎ๐ด๐ฟ๐ฒ๐ฒ๐ถ๐ป๐ด ๐ผ๐ป ๐๐ต๐ฎ๐ ๐๐ต๐ฒ ๐๐ผ๐ฟ๐ฑ๐ ๐บ๐ฒ๐ฎ๐ป.
GitHub: github.com/NVIDIA/auto-ontolโฆ
Links in reply ๐
Binance told customers in France, Spain, Italy and Poland to withdraw their crypto before their accounts close Dec 31, 2026 โ it still holds no MiCA licence to operate across the EU.
Only 2 out of the top 10 dollar stablecoins fully comply with upcoming U.S. safety rules. โ ๏ธโ ๏ธโ ๏ธ
According to a recent report, Circleโs USDC and Paxosโs PYUSD meet the strict Federal Reserve standards taking effect in Jan 2027 under the GENIUS Act.
The major fails? Prominent onshore stablecoins like World Libertyโs USD1 and Rippleโs RLUSD do not pass the test. The primary roadblock is their reserve backingโboth hold longer-dated Treasuries and agency debt instead of the ultra-short-term, low-risk assets required by regulators.
As direct federal oversight approaches, issuers face a hard choice: aggressively restructure their reserves or risk being squeezed out of the compliant U.S. ecosystem.
#Stablecoins #CryptoRegulation #FederalReserve #Web3
43
Phoenix 5D๐ฅ retweeted
holy f*ck, NO HUMAN WEB DESIGNER WANTS TO COMPETE WITH THIS IN 2027.
youโre watching AI agents build websites at a speed most humans canโt match.
and the scary part? these tools are available to almost anyone.
7 GitHub repos you NEED to learn before your clients find someone who already has:
1. Bolt.diy - BUILD ENTIRE WEBSITES WITH AI
generate, edit and run full-stack web apps with AI.
github.com/stackblitz-labs/bโฆ
2. shadcn/ui - PROFESSIONAL UI IN MINUTES
beautiful, customizable components for modern websites.
github.com/shadcn-ui/ui
3. Magic MCP - YOUR AI FRONTEND DESIGNER
generate polished React components directly from your AI coding workflow.
github.com/21st-dev/magic-mcโฆ
4. Playwright MCP - AI THAT TESTS ITS OWN WORK
let AI agents open websites, click buttons and catch UI bugs.
github.com/microsoft/playwriโฆ
5. Chrome DevTools MCP - FIX BROKEN WEBSITES FASTER
give AI access to browser debugging, performance and network diagnostics.
github.com/ChromeDevTools/chโฆ
6. Vercel Agent Skills - MAKE AI CODE PRODUCTION-READY
React optimization, frontend best practices and web design reviews.
github.com/vercel-labs/agentโฆ
7. Webstudio - BUILD VISUALLY, KEEP CONTROL
an open-source visual website builder with reusable components and styling.
github.com/webstudio-is/websโฆ
the real threat isnโt AI replacing every designer.
itโs ONE person using AI to deliver what used to take an entire team.
2027 is getting closer.
learn these tools before your competitors do.
LIKE + BOOKMARK + REPOST this so your designer friends see it.
and be honest in the comments:
do you know anyone whoโs already LOST THEIR JOB because of AI?
While crypto headlines debate retail tokens, Wall Street has quietly built a $7.5 Trillion-a-month tokenized collateral network.
โBroadridgeโs Distributed Ledger Repo (DLR) platform processed an average of $359 Billion in daily volume (Coincub):
โโข Overnight Repo vs. Intraday Repo: Traditional repurchase agreements lock up government debt collateral overnight. Distributed ledgers turn Treasuries into fluid intraday collateral (Broadridge)
โข Sub-Hour Settlement: Primary dealer banks (using J.P. Morgan's Kinexys and tokenized commercial deposits) borrow cash against tokenized bonds for 2 to 4 hours, returning the collateral same-day (J.P. Morgan Kinexys)
โข Liquidity Buffer Relief: Shifting from overnight locks to continuous intraday velocity frees billions in idle regulatory liquidity buffers
โTokenization isn't coming to institutional finance in 2030โit is already processing hundreds of billions every morning.
โ#Tokenization #RepoMarket #Broadridge #JPMorgan #CapitalMarkets #Fintech
๐ค Made with AI
50
You need to see this!!
๐๐27 Signals Tested, 2 Survived โ Full Results ๐
I backtested 27 trading signals with walk-forward analysis. 25 failed. Here's every result, the methodology, and the 2 that survived. No cherry-picking. ๐งต
METHODOLOGY (so you can check my work):
โข 5yr Schwab daily data (2021โ2026)
โข SPY, QQQ, IWM
โข Walk-forward, 10 folds, strictly causal (closed bars, next-open entry)
โข Pre-OOS purging to kill lookahead leakage
โข 0.02R friction per trade
THE BAR (all four required to pass):
โข Avg return > +0.15R
โข Profit factor > 1.3
โข n โฅ 10 trades
โข No single ticker worse than โ0.2R
Most published backtests would fail this bar. That's the point.
THE 2 THAT PASSED โ
Confluence v3 composite
โข +0.43R avg, PF 1.84, n=252
โข 62.7% win rate
โข Layers: Supertrend + volume profile + RSI/MACD + flow + ADX regime + VWAP + weekly trend
#2: Trend leg (Confluence v3 trend component + โ1.5R hard stop)
โข +0.75R avg, PF 2.67, n=19
โข 63.2% win rate
โข The hard stop was the key โ without it, one โ3.89R tail trade on IWM killed it
PLOT TWIST: the two "different" signals produce byte-identical entries (27/27 match on 2y SPY). They were never two signals โ one signal, two universes, two exits. The research proved the stack was already correct.
THE GRAVEYARD (all failed) ๐
โข Supertrend alone: โ0.25R
โข MACD histogram: โ0.30R
โข RSI divergence: โ0.32R
โข Stochastic(9,3,3): โ0.21R
โข Hull MA(21): โ0.19R
โข TTM Squeeze: โ0.48R (worst)
โข Turtle System 2: โ0.19R
More failures:
โข Donchian 20/10: โ0.23R
โข Pivot Trend: โ0.20R
โข CVD divergence: โ0.12R
โข ADX/DI cross: +0.08R (regressed on 5y, parked)
โข DeMark 9/13: flat to negative
โข Flag patterns: โ0.88R, 0 wins in 7
Interesting near-misses:
โข Supertrend+MACD+RSI combined: โ0.05R (closest to breakeven โ 3 filters beat 1, but still negative)
โข Signal Charge: one config "passed" but 8/8 neighbors failed = overfit, rejected
โข Dual Momentum (Antonacci): beat SPY by 0.5pp CAGR with better drawdown โ but it's a portfolio overlay, not a signal
THE FINDING: In the 2021โ2026 window, NO single-indicator signal works on index ETFs. Not one. The only thing that clears the bar is a heavily filtered confluence stack โ 7+ conditions that must ALL agree.
Why most "proven" indicators fail here: the 2024โ2026 market is a choppy bull market. Breakouts whipsaw. Mean-reversion catches falling knives. Published backtests from 1996โ2006 don't transfer โ different regime.
What I'd tell any trader: stop hunting for the magic indicator. The edge isn't in the signal โ it's in the FILTERING. Every failed signal above works in theory. They all die on false positives. Stack filters until only high-conviction setups survive.
Full methodology, all 27 results with per-ticker breakdowns, and the reviewer checklist I used to audit my own work โ happy to share the details. Backtest before you trust.๐งต
Do you want the pinescript?
Like, share and comment tested.
1
40
Phoenix 5D๐ฅ retweeted
holy f*ck, I just sold this single AI-generated 3D neuron for $2,000 to a private biotech lab.
a private lab saw my visuals on X and reached out. they just ordered a prototype for their modeling work.
designers who can't do this will be looking for a job in 2027. save this before its too late:
7 design repos that make Claude Code feel like a whole design team:
1. Awesome DESIGN.md
84 design systems pulled from real sites, each one a single markdown file. drop it in your project, say "build me a page that looks like this" and it actually does
github.com/VoltAgent/awesomeโฆ
2. Hallmark
a design skill from Together AI that refuses to look AI-generated. 50 themes, builds the page around what the visitor came to do, then checks it in a real browser. /hallmark redesign rebuilds any old page
github.com/Nutlope/hallmark
3. Huashu Design
type one sentence, get a launch animation, a clickable app prototype, an editable PPT or an infographic. it shows you 6 directions first, you pick 3, it builds them for real
github.com/alchaincyf/huashuโฆ
4. video-shotcraft
tell Claude "make a promo for my product". it screenshots your real pages, storyboards, animates, syncs to the beat, adds sound and checks every frame. you get an MP4 you can keep editing
github.com/Vincentwei1021/viโฆ
5. OpenPencil
an open-source Figma alternative that opens real .fig files. describe what you want in chat and the AI builds the design, and Claude Code can drive it directly
github.com/open-pencil/open-โฆ
6. Motion2Code
show it a motion design video and it rebuilds it as clean SVG + GSAP code, checked frame by frame against the original. a 10-second animation ends up around 24 KB
github.com/lovisdotio/motionโฆ
7. Blender Agent Skills
22 pro Blender workflows for Claude Code: modeling, materials, rigging, animation, lighting, rendering. the full 3D pipeline, not just "make a cube"
github.com/EL4CTEO/blender-sโฆ
the neuron in this video?
I built it myself with Claude Code and a few 3D and motion tools and sold it for 2K. no studio, no team.
want to make something like this?
like + save this post, comment "NEURON" or DM me and I'll send you the exact prompt.































