MrMochi (@baronvonspread) header
MrMochi (@baronvonspread) profile picture

Run the market. Not the other way around.

right time, right place
Joined January 2018
Photo by MrMochi (@baronvonspread)Photo by MrMochi (@baronvonspread)Photo by MrMochi (@baronvonspread)Photo by MrMochi (@baronvonspread)Photo by MrMochi (@baronvonspread)Photo by MrMochi (@baronvonspread)Photo by MrMochi (@baronvonspread)Photo by MrMochi (@baronvonspread)Photo by MrMochi (@baronvonspread)Photo by MrMochi (@baronvonspread)
Google is on a tear. When using an interface that is platform agnostic you can just switch or delegate to whatever provider is offering the best service at that moment. Easy switching makes the space very competitive and much more enjoyable in building your own pipelines.
Gemini 4 Neon: frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows. Begins rolling out today in the Gemini API in @GoogleAIStudio, Gemini for business, for paid Google AI Pro and Ultra subscribers in Google @Antigravity, and for Google AI Pro and Ultra subscribers in the @GeminiApp.
9
I don't think it really matters whether Jensen will be back at $NVDA but it probably happens at some point. The cat is out of the bag and it will be hard to navigate for any CEO through the coming months especially with the changes coming to the house and senate.
1
1
50
For Nvidia that means demand will subside with China and neoclouds. Especially China was most likely a big portion of the demand. With the power of balance shifting again in the US, that means there might be more strict export controls as well as regulation for data centers.
1
9
Add on top that there has been significant progress in platform agnostic solutions that remove the moat from CUDA. I think the cost of AI has to go down a lot more for it to make it next leg up. Bringing down the cost of compute and power will be a big part of this.
Deer in the headlights when the pool somehow vaporizes
29
Druckenmiller & Elon bullposting into equities reporting weaker earnings and Binance getting hacked. The week isn't even over yet.
60
Yields will continue to rise until policy improves
BREAKING: President Trump is preparing a 90-day ban on US diesel exports in an effort to bring down record diesel prices, per Politico. Any halt to shipments would be the first such restriction since 2015, when President Obama lifted a long-time ban on oil exports. This comes as diesel prices have surged over +100% from their lows seen just nine months ago. Diesel futures are down -7% on the news.
49
"we tested GLM-5.3-Flash anonymously as ox-alpha on OpenCode and OpenRouter to gather user feedback. It quickly became the most popular model of the week — with all of this traffic served on Chinese AI chips." Chipping away at the hardware + software moat.
Introducing GLM-5.3-Flash - Leading capabilities at a highly competitive price - Natively multimodal with a 1M-token context window - A 320B-A18B model released under the MIT License - Previously previewed as Ox Alpha, running entirely on Chinese AI chips Blog: z.ai/blog/glm-5.3-flash Available now across all official platforms: Weights: huggingface.co/zai-org/GLM-5… API: docs.z.ai/guides/llm/glm-5.3… Coding Plan: z.ai/subscribe ZCode: zcode.z.ai/en Chat: chat.z.ai AutoClaw: autoclaw.z.ai
104
Sir, Bond Bandits have struck T-Bill Harbour.
1
60
Token dev is bull posting. Bessent has to be absolutely desperate at this point with all the Yen and growth grandstanding. Every day the debt ticker goes up, the US loses more allies, and the bond selling continues. I wonder what trade he'd take if he was not in government.
🇺🇸 Treasury Secretary Bessent says "we can grow our way out" of $40 trillion US national debt with tariffs and economic growth.
3
78
The synchronous economy will have to lay in wait for the time being. Synchronicity is inevitable!
It is with a heavy heart that we are pausing development on Synchronicity at this time. This is due to a variety of factors, including lack of funding, less than hoped for adoption, and an increasingly competitive environment for AI-assisted trading products. We are incredibly grateful for your participation on this journey. Here is what to expect moving forward. The testnet has been taken down. On Monday the 24th at 21:00 UTC, we will transition the mainnet to either paper trading mode, or demo mode. We had been developing a paper trading mode which, if it is able to work in production, we will activate. Otherwise, we will keep the existing mainnet live, but will not be making any commitments about maintenance, correctness or reliability after Monday the 24th at 21:00 UTC. We will keep it live as we consider opportunities to sell the product and want to have it accessible to potential buyers. If you have an open order, or open position with a SL or TP, it will be active on HL until Aug 24th at 21:00 UTC. If you want to cancel it, you can do so on our site before then. Otherwise, we will close and cancel everything at approximately that time. If you have a strategy that hasn’t opened a position, it will remain active until approximately that time. At approximately that time we will cancel all strategies. You can cancel things before then if you prefer. We encourage you to close positions and cancel strategies and orders through our interface, because if you do so on HL our system will not register it and will still close them (which in practice may open orders or positions in the other direction if you have closed through HL). If you have any questions, please don’t hesitate to reach out in our Discord, and we will respond in a timely fashion. We will be announcing the winners to the final competition shortly. It has been a great pleasure to build this product for you, and to be a part of this lovely community. It will be missed. Thank you, truly.
1
2
158
🌞 It's open weight summer 🌞
Introducing GLM-5.3: Built to Code. Ready for Cyber Defense. - Top-tier coding and agentic capabilities, achieved through post-training on the 743B base model - A major leap in cybersecurity, setting a new standard among open models Tech Blog: z.ai/blog/glm-5.3
1
158
Society after Qwen 3.8 27B release
📢Meet Qwen3.8-Max — our most capable model to date. Next week, the open weights of Qwen3.8-Max will be released, and Qwen3.8-27B is also going open-weights to meet you all!🎉 Qwen3.8-Max, a new bar for coding and cowork at 2.4T parameters: - Autonomous coding: 10+ days of self-evolving development, from empty folder to production without hand-holding, complete project trace in the GitHub:github.com/qwen-code-dev-bot… - Real work, real results: Production-quality deliverables across hundreds of professions. - Long-horizon mastery: System-level autonomous planning with closed-loop adaptive learning, driving 500+ turns of chip design optimization and 365 days of e-commerce strategy. - Native multimodal intelligence: Vision isn't just input — it's a continuous feedback loop for planning, execution, and self-correction. 💰Pricing: Input: $2.0 / M tokens Output: $6.0 / M tokens Implicit Caching: $0.25 / M tokens Start building with Qwen3.8-Max! 🚀 📖 Blog: qwen.ai/blog?id=qwen3.8 ✅ Qwen Studio: chat.qwen.ai/?models=qwen3.8… ⚡ API: qwencloud.com/models/qwen3.8…
243
There's a real danger in tuning your trading setup to backtesting data. As you tune it closer to mimic the data from the past, it will be prone to fit to the noise it carries. Testing out-of-sample can help mitigate some of this by backtesting against one slice of data, then testing it against new unseen data. Synchronicity makes this easy to achieve with our integrated Backtesting Suite. Come and have a look, we'd love to have your feedback! :)
A backtest can look great for the wrong reason. Tune a strategy long enough and it starts memorizing the past instead of learning a rule that holds up over time. Our strategy optimisation tools guards against this in two steps: tune the strategy on one stretch of history, then test it on a later stretch it has never seen. If the results hold up on the new data, you can have more confidence your strategy learned something that lasts. If they collapse, it likely memorized the old data, and you found out before real money was involved. Join the waitlist at synchronicity.xyz/l/SHeKtg Our Discord will help you through your first backtesting run.
66
Open weights have been on a tear lately. I'm very excited to try out more of the newest models in the 27B and 35B range. Both from @Alibaba_Qwen and smaller teams. Being able to have Sonnet like performance running privately at home is a testament to how quickly things have been evolving. . I've been running auto researchers and building personalized systems for everyday use. Not everything requires a trillion parameter model. I haven't tried out BTL-4 but I'm hoping to be pleasantly surprised.
Releasing BTL-4 35B and Macaw 2.7B a frontier agentic reasoning model and an on-device Mac agent. Both open weights, both available now. • 73.5% BFCL v4 (AST) — +4.3 over the base model, paired: identical harness, identical decoding, only the weights differ, all 1240 cases • 78.4% SWE-bench Verified • 66.1% LiveCodeBench v6 — 99.1% easy · 86.7% medium, full 442-problem window • 262K native context, and it uses it: raising the output budget 16K → 32K moved LiveCodeBench +5.2 points on its own. Long problems need room to finish. Macaw 2.7B — the assistant your Mac should have shipped with. 97 verified macOS tools, chained from one sentence. • 1.5 GB on disk, ~2 GB running, ~2s per request on a base M2 • Reads your screen and your documents — no vision model(uses esp as vision), no cloud • 100% local. No account, no telemetry, no server to send anything to Open weights, deploy anywhere:🤗 huggingface.co/badtheorylabs… 🤗 huggingface.co/badtheorylabs… 💻 github.com/Badtheorylabs/Mac… 🔗 badtheorylabs.com/macaw vLLM and MLX supported day one.
1
164
This is unironically good for crypto
JUST IN: X head of product, Nikita Bier, steps down
1
110
Describe it like a trader. Deploy it like a quant. Synchronicity built systematic trading infrastructure built so you don't have to. You can spend your time where you instinct matter: trading.
A trade idea usually feels complete until you try writing down its rules. Describe the setup to Signac in the words you'd use with a trading friend. It'll ask what you left out (what invalidates this, how big, what closes it) and draw the strategy on the canvas as nodes you can read. Go through it together and tighten the rules until it makes the trade idea you have ready for prime time. Nothing runs until you approve it. Get on the waitlist at synchronicity.xyz/l/kdjWts Our community shares referral codes in our Discord so you can jump straight in.
1
64
A while back I overheard a customer at my barber's explain he'd been buying Bitcoin. The barber asked whether a crash worried him. "Bitcoin always goes back up". A few weeks later Bitcoin topped. He may even be right of course. But that belief and behaviour was learned somewhere, and I think it was learned in two decades of cheap money. The ECB alone spent eight years below zero. Every dip got bought, betting more kept paying, and regular people with regular jobs drew the obvious lesson. Not just in Bitcoin either. I've heard the same conviction about stocks, ETFs, whatever people can get leverage on. Some investment vehicles may be forever up and to the right, if you have a long enough time horizon. Personally I don't believe markets have priced in that the cheap money era may be over for a longer periode of time. Rates might sit above 5% for a while yet, geopolitics has pushed the world in a certain direction, and in time I'd expect a far more risk averse market to come back. Clearly not today though. From Seoul to New York, traders are hunting for fresh capital to bet bigger. I could be mistaken. But usually when the barber shop stops worrying about draw-downs, it's a good time to start re-evaluating my own risk appetite.
Situational Awareness, the $20bn hedge fund founded by former OpenAI employee Leopold Aschenbrenner, has sought to raise fresh capital from investors after suffering heavy losses during the recent rout in AI stocks ft.trib.al/SwCjjE6
79