Samuel Teboul

@nodi_macro you forgot parenthesis to be more accurate: BTCUSD/(USDT.D+USDC.D)

green zone for $BTC is loading 🧘‍♂️

@tSamoss sure

I lost $132k on perps last year. Not because my thesis was wrong. I shorted BTC from 116k. It's 69k now. I lost it because I couldn't stop touching the screen. So I built something to stop me. Here's the story. The numbers were embarrassing. 120 trades a day. Average hold time: 11 minutes. I'd nail the entry, watch it go green, panic on a wick, close, then re-enter 30 seconds later at a worse price. Rinse and repeat. All day. Every day. I knew I had a problem but I didn't know HOW BAD until I fed my entire trade history into Claude and asked it to be brutally honest. "Find every pattern that's costing me money." What came back was a murder scene. The AI found that my re-entries alone — closing a position and opening the same one minutes later — accounted for most of my losses. Not bad reads. Not wrong direction. Just me, fighting myself, paying spread and fees to do it. My thesis was printing. My hands were burning it. Here's what I realized: I don't have a strategy problem. I have an execution problem. And execution problems don't get fixed with more discipline. They get fixed with systems. So I asked myself: what if I literally couldn't overtrade? I built a trading bot in one afternoon. Not an algo. Not a signal generator. A discipline engine. It connects to Hyperliquid through Telegram. I send it my thesis: coin, direction, size, stop loss, take profit. That's it. That's all I'm allowed to do. The bot has a "Guardian Mode" that enforces rules I set for myself: - Max 3 trades per day - Mandatory stop loss on every position (no naked trades) - Forced 2-hour cooldown after consecutive losses - Can't increase size after a loss - No rage re-entries. Physically impossible. The beautiful part? I can't override it. If I try to open a 4th trade, it just says "GUARDIAN BLOCKED." If I panic and close via the exchange UI, the bot detects it within 30 seconds, logs it as a "manual override," and it still counts against my daily limit. There is no escape hatch. First trade through the bot today. Sent one message. Got a confirmation with entry, SL, TP, risk/reward ratio. Then I put my phone down. That's it. That's the whole workflow. The bot watches the position, trails the stop if I want, and pings me when it closes. I don't touch anything. The irony is not lost on me. I spent months trying to be more disciplined. Reading trading psychology books. Setting rules I'd break by lunch. Then I spent one afternoon with an AI and built a cage around my worst impulses. The cage works better than the books. Now there's only one question left: Is my macro thesis actually good, or was I just lucky once? The emotions are out of the equation. The overtrading is gone. The revenge trading is gone. All that's left is the thesis. Pure signal. If I'm right, the bot prints. If I'm wrong, the stop loss catches it and I live to trade tomorrow. The lesson isn't "build a bot." The lesson is: if you already know what you're doing wrong, stop trying to fix yourself. Fix the system around you instead. Automate your discipline. Remove yourself from the equation. Your edge isn't your entries. It's your ability to not destroy them.

Built a simple landing page with Base44. Got Stripe, Three.js, PDF generators, rich text editors, drag & drop libs… Didn’t ask for a SaaS starter kit but ok. Also: everything is JSX. Wild.

Expecting juniors not to rely on AI is unrealistic. It’s too good and moving too fast. The real risk isn’t juniors — it’s senior devs with 10–20 years of experience who don’t learn to use AI properly. AI raises the floor for juniors, but it raises the ceiling even more for seniors who adapt.

If I were a junior developer today, I would speedrun learning the fundamentals. Things like SOLID, clean code, clean architecture, and a bit of algo/puzzle solving for logical thinking (leetcode) Then I would maximize using AI, but actually read the generated code and understand it. Talk to AI or a senior dev if you have any questions. Learning from open source is also a great way to upgrade your skills quickly. I think AI accelerates the learning process and raises the level of what a junior-level engineer can do. The expectations for a junior dev role will be much higher than they were 3 years ago. But devs are smart, they can learn anything. Only people who are too comfortable with their level and stopped learning will risk getting replaced and forced to learn the new way.

AI won’t replace software engineers. But it will expose bad engineering decisions faster than ever. AI is great at making things look finished: • UI works • Demos impress • Features ship fast What it doesn’t guarantee: • Security • Scalability • Maintainability AI is a force multiplier, not a decision-maker. The winners won’t replace engineers with AI — they’ll combine human judgment with machine speed. That’s the real edge.

@CryptoVikings07 You got a point lol I remember the only "revolutionary" they did is backing their UST with BTC 🤡

@tSamoss Lol LUNA was never a revolutionary, it didn't had any revenue of its own, team and VC owned shit. All it was is a ponzi which lasted long before the collapse

A bet on revolutionary project is always a win Look how hyperliquid revolutionized the perps industry and pushed for transparency when even the top exchanges were scamming openly. It kept its community at top and we can slowly see the impact of many upcoming projects keeping community on the top. No matter where the price goes, hyperliquid will always be a gem and large contributor in crypto industry Another revolutionary project could be lighter with its zero fees model and by keeping the community at the top. Institutions who missed $HYPE before won't dare to fade this one. Also it is pushing other upcoming dex to follow their zero fees model They got everything to be revolutionary - Solid execution - Good liquidity - Fastest growth - Chad founder Let's see what happens next 🕯️

We will be joined by the @fantasy_top_ team for a stream this Friday. Tune in as we discuss speculation markets, fantasy in web3, and much more. 🗓️ 29 Aug, 10 AM UTC, set your reminders.

FDF has seen immense growth over these first two weeks. Now it’s time to go global! We’re looking for passionate volunteers to lead the FDF communities in their local country/region. Keep reading for details 🧵

@AzmaniMouhieddi Fake

@alexwestco Thanks for your post and being honest. It feels like it goes up and right for everyone on X. But in reality, as you said, business is tough nowadays. Keep it up 💪

2024 recap: - was fixated on removing myself 100% from my business, failed - hired an (amazing) operator, failed - was fixated on hitting $1m/yr instead of product/service, failed - ended up with a mediocre service - revenue dropped by 50% after that - took months to recover revenue-wise - industry got incredibly hard, wanted to pivot - bunch of experiments failed, a couple succeeded - too much stress - too much traveling - did more of the things i don't enjoy, like working all day alone and stressed - did less of the things that i do enjoy, like hobbies, writing and youtube - lost my natural excitement about the future and became more pessimistic first "down-year" mentally and financially in a very long time. since i launched cyberleads in 2020. i was used to things always going up and to the right. which is not how life works.

Come meet us at @token2049 !!

Narval Armory Stack can now be operated seamlessly using Rabby wallet 🚀 See the demo in the link below 👇🏼

check it out 👇🏼

@gakonst 🚀 for @Narvalgmi

which are your favorite ethereum devtools these days that more people should use? open source & free required, ideally maintainer that cares for community looking to highlight some ecosystem efforts on what's good

Breaking News!!! @Narvalgmi Armory Stack is officially open source 🚀

Next generation of access management for web3 is here ⏱⏳ @GregJessner

Those are very interesting points! As the blockchain industry evolves, it's clear that OAuth tailored for #Web3 is essential. This is exactly the challenge we are tackling at @Narvalgmi. Stay tuned for exciting updates! 🚀

While OAuth, the open authorization standard in web2, offers a framework for access delegation (ie letting a system access and use a protected resource/data), it's not fit for web3. Web3 has specific Auth needs: ✴ Wallets require trust models & signing schemes removing trust from even trusted operators ✴ Transactions need web3-aware policy rules around transfer details & contract calls ✴ Integrations with the blockchain ecosystem requires interfacing with decentralized systems and less-traditional signing curves To tackle these challenges, most wallet technologies have bundled their own auth solutions with the rest of their stack - but many problems and use-cases remain unresolved. 👀 Time for OAuth 3.0, tailored for web3? 👀