Vibe Trader

Day 7 Had some technical issues getting trades to fire properly, so our automation missed a couple trades including a 3.7R win 😫. All good though, here are a couple of wins from Thursday's Asia session. The bot is up $800+ on a funded and $2K+ on an eval. Not bad for 1 week. We've also been adding more strategies to the portfolio, currently sitting at 7 strategies, but only 5 are live right now. Our quant factory is doing WORK. Coming along nicely.

@ashen_one Codex 5.4 on Hermes you’re welcome

I WILL FIND YOU GUYS A REPLACEMENT TO CLAUDE OPENCLAW EVEN IF TAKES 500 MODELS OF TESTING To start, we tested GLM 5.1 (came out this week) + Minimax 2.7 Next week, we're testing Gemma, Gemini, Mimo, and Qwen What other models do you guys wanna see in Hermes/Openclaw?

@ziwenxu_ Just switch to codex. It feels like peak Opus right now. Unless you’re doing a lot of frontend heavy stuff, then it’s not great.

@VibeTradr Exactly I've been noticing this as well, sucks to see those happen But I will downgrade to 4.5 and see what's up.

Anthropic is secretly nerfing Opus 4.6 and hoping you won't notice. I have proof: Evidence is stacking up that 4.6 is getting brutally quantized to handle demand, while 4.5 stays pristine. Fresh benchmarks reveal a devastating gap: > Opus 4.6: Logic failures on repeat. > Opus 4.5: Nails it every single time. One dev now runs a "Quantization Canary" a diagnostic prompt fired at the start of every session. He just watched five 4.6 windows fail back-to-back. His verdict: "Switched to 4.5 and it felt like I finally got my brain back." If the model feels dumber lately, trust your gut. You're being throttled so they can save on compute. Switch to 4.5. The difference is night and day.

@Sherlockwhale @atlasfutures

I found a pattern on Bitcoin that has a 100% win rate across 8.5 years of data. 18 out of 18 with zero exceptions. Every week, Bitcoin opens at a price on Monday and if by Thursday the lowest price of the entire week has not dipped more than 0.25% below that Monday opening price, the week closes green. Every single time. 18 for 18 since 2017. The probability of this happening by random chance is 1 in 131,072. The average return when this pattern triggers is +10.98%. Not 2% or 3% but almost 11% in a single week. The smallest green was +1.9%. The largest was +28.8%. Even the worst outcome was solidly profitable. It works in every market condition: 2018 bear market: +28.8% 2022 bear market: +8.6% 2023 recovery: +21.9% 2024 bull market: +11.8% March 2026, three weeks ago: +10.4% Bull market, bear market or sideways, it does not matter. If sellers cannot push price even a quarter of a percent below the weekly open in 4 full days, buyers are in complete control. The rest of the week is just momentum playing out. Every Thursday, check one thing. What was the lowest price of the week so far? If it never dropped more than 0.25% below Monday's open, the week closes green and there will be no further dump, you can stay comfortable in your longs. Bookmark this and check this by Thursday close (today). If price has not dipped 0.25% below 66,000 (Monday open), there will be no further dump.

@soundslikecanoe This is literally on my todo list, nice work

Super nerdy but if you're vibe coding software you may like this. It's about to change my life I created a "Feature Request" Chat on my app, powered by Sonnet Anybody on my team can use this chat to request a new feature or report a bug in the app. It asks a few simple questions and then turns that into a thorough spec. Then the spec gets posted to a backlog where I manually review it and (1) add notes if I have them, (2) Move it to Spec Ready Then I have a Scheduled Claude Code task that picks up items in Spec Ready every night and implements the spec + moves it to PR Ready with a link to the PR. Then I have a separate scheduled task (fresh context!) pick up those items and do a thorough code review and testing on each. Claude also posts comments to the comments box in the card every time it makes an update / has feedback. ALL of this stuff I used to do manually. It was getting to be way too much this took forever to set up and I'm still tweaking it. I know many of you have cooler agent setups but this has been a lightbulb moment for me, so thought it was worth sharing.

@elgermerlo @clawra_official

As solo founder how do you deal with loneliness?

@Amank1412 Skill issue. I can fix it for you for an additional $5K. You’ll save more in the long run

SaaS is dead. We replaced $750/month in SaaS with $4,570 in LLM tokens. Now the team spends half their time debugging vibe-coded chaos instead of shipping. But hey… we “own the stack.”

@1a1n1d1y We should start a riot

if you're the engineer that trained claude code to put 3 fallbacks in everything, seriously fuck you

@phuctm97 Yes absolutely

Is Context7 MCP still necessary these days?

@Thomasslabbers 😂 but also true 😭

I used to work 12 hours a day. But thanks to AI, I now work 16.

@blackroomsec @Mitman93 @TheChiefNerd Sounds like a skill issue. You haven’t learned to deal with hallucinations or inconsistent prompting.

I was absolutely pumped about AI as well as my timeline will show but over time using it everyday for various things that I would need to actually have it do, it has gotten progressively worse. In the last 2 months though it is the worst it has ever been. It cannot remember files which it has been given after three prompts. They have to be re-uploaded. It has to be handheld significantly and that evidence alone that I am verifying with my own eyes tells me it cannot replace a human worker. And it's certainly is in sentient which is just the most preposterous thing I've ever read in my life.

🚨 Anthropic CEO Dario Amodei: “We are so close to these models reaching the level of human intelligence, and yet there doesn't seem to be a wider recognition in society of what's about to happen … There hasn't been a public awareness of the risks.”

@PrajwalTomar_ Have you even tried Stitch? It’s ok, it’s definitely not as amazing as you’re making it sound.

STOP asking Claude Code to design your apps. It’s great for logic, not design. Google Stitch 2.0 + Claude Code via MCP is the workflow that finally fixes the AI slop problem. If you want to build apps that actually look good, read this.

@katibmoe One

Introducing One. The simplest way to connect and monitor AI agents to hundreds of apps. And we’re open-sourcing the world’s largest integration database powering it: 47,000 agentic actions across 250+ apps. RT + comment “One” for access & 1M free API requests/month.

@markoilico I'll check it out

If you're now designing or redesigning a website, this will help you a lot. I recently curated the best hero sections, footers, social proof and other website parts because I got tired of having 15+ tabs open (even with Mobbin). Giving it away 100% free. Comment on this post, and I'll send a Figma link to your inbox!

@anvisha Moda

We raised $7.5M to kill AI slop. Introducing Moda: the world's first design agent with taste. RT+ comment “Moda” and we’ll design your brand for FREE.

@_not_a_fish Bro what kind of CPU do you have?

Claude Code deployed 27 agents to simultaneously analyze 300+ photos of annotated chart data I’m teaching “Tiny Fish” my intuitive edge

@sweatystartup A-players embrace change and make opportunities happen no matter what the environment is. If your A-players can’t figure out how to make AI work for them maybe they’re actually B-players.

I don’t know a single A player who is gung ho about AI. It’s the opportunity of a generation for C players to try and hide how much they suck.

@maxxmalist 🐐

We saved thousands of dollars and weeks (maybe even months) of testing work by doing this. Agentic coding is a beautiful thing.

Step 3 Hook up Codex or Claude Code or even OpenClaw (whatever tf you use to code) to your MCP. Congratulations, your coding agent can now navigate your app like any other user and can use that feedback loop to test (and fix) everything on your app.

On Friday we were ready to spend thousands of dollars hiring a user testing agency to test our app. Now it's Wednesday and we're 50% done just using Claude Code. Definitely not hiring that agency now. If you vibecode anything, here's how how to speed up your testing by 100X 👇