Arnaud Jeulin

The tools traders actually need in 2026: Not: another chart platform (there are too many) Not: a copy-trading app Not: a "signals" bot But: fast, reliable analytics that help you understand market structure. Currency correlations + volatility + currency strength = context for better decisions. That's mataf.net. #Forex #TradingTools #BuildInPublic

Week 1 recap, what I shipped vs. what I planned: Planned: Data pipeline + UI mockups Shipped: Data pipeline ✅ + started currency correlation engine Where I'm behind: UI mockups (pushed to week 2) Real talk: rebuilding is slower than greenfield. Every legacy assumption has to be questioned. Sharing the journey here every week. 👀 #BuildInPublic #FinTech

The hardest part of building in public? Sharing things before they're ready. But the feedback is worth it. Already getting great ideas from alpha sign-ups about what traders actually need vs. what I assumed they needed. Keep them coming! 💬 Join the alpha: forms.gle/azQZLjpz8TcCkJ… #BuildInPublic #Feedback

Why subscription > ads for a tools platform: Ads incentivize: more page views, more clicks, keep users scrolling. Subscriptions incentivize: users actually succeeding with the tool. mataf.net is going subscription in 2026. The product has to actually work. Or people cancel. That's a much healthier accountability structure. #BuildInPublic #ProductStrategy #FinTech

One thing I learned rebuilding a legacy app: The original architecture tells you a lot about what you valued at the time. Mataf 2003: optimize for page views (ads). Mataf 2026: optimize for user outcomes (subscriptions). Different goals → completely different product. What matters most to you in a trading tool? #BuildInPublic #ProductThinking

Weekend thought: Most forex tools were built for desktop, in the 2000s, for professional traders. The retail trader of 2026 is on mobile, wants instant insights, and has way more options. The tools haven't caught up yet. That's the gap mataf.net is trying to fill. #Forex #FinTech #TradingTools

End of week 1. Building in public update: ✅ New architecture designed ✅ Data pipeline v1 working ✅ Alpha sign-ups coming in 🔄 Currency correlation engine: in progress If you want early access to the new mataf.net, join the waitlist: → forms.gle/azQZLjpz8TcCkJ… Have a good weekend! 🙌 #BuildInPublic #FinTech

AI in fintech is overhyped. There. I said it. BUT... there are specific places where ML genuinely adds value: → Anomaly detection in price feeds → Volatility pattern recognition → Correlation stability scoring These are the AI features I'm building into the new Mataf. No buzzwords. Just useful. #BuildInPublic #AIinFinance #FinTech

The feature I'm most excited to ship: A Currency Strength Index that shows you, in real time, which currencies are gaining momentum vs losing it. Not just a chart. A tool for actual trading decisions. → Based on correlation matrices → Updated on live market data → Visualized so traders can act on it fast Alpha testers wanted → forms.gle/azQZLjpz8TcCkJ… #BuildInPublic #Forex #CurrencyIndex

Forex data is HARD to work with. → Markets run 24/5 (with gaps, not 24/7) → Prices come from multiple providers (which one do you trust?) → Historical data has inconsistencies across sources → Real-time feeds have latency you need to manage Building reliable currency tools means solving these problems first. Working on it. 🔨 #BuildInPublic #FinTech

The first architectural decision for the new Mataf: Ditch the monolith → microservices for data feeds Real-time updates → WebSocket instead of polling Caching strategy → Redis for forex data (changes every tick) Boring? Maybe. But boring = stable for traders who depend on it. What's your take on microservices for fintech? #BuildInPublic #SoftwareArchitecture

Why did mataf.net go silent for 3 years? Honest answer: I got lost in other projects. The site kept running on autopilot... until it didn't. High latency. Random crashes. Users frustrated. The wake-up call I needed. Building it back better. Day by day. In public. 👷 #BuildInPublic #FounderJourney #FinTech

@tymofii I'll go full Deno/Typescript in a monorepo (Postgres+Redis). Easier to share types between backend and frontend.

@dauran For a high-traffic fintech app, I'd go with Go or Java for the core backend, Postgres with strong partitioning, and a React/TypeScript frontend. For AI features, a separate Python service using FastAPI and PyTorch is the cleanest way to manage scale and complexity. Good luck with

Fun fact: the original mataf.net codebase is 20+ years old. When I looked at it recently, I barely recognized my own code 😅 Starting fresh. New stack. Better architecture. AI-powered features. What tech stack would YOU choose for a high-traffic fintech app in 2026? A) Next.js + Vercel B) FastAPI + React C) Go + Vue D) Something else 👇 #BuildInPublic #FinTech

Something big is coming back to life. 🔄 I've been quietly rebuilding @mataf from scratch for the past months. The old site? It was my first major project back in 2003. It's also been crashing too much lately. Time for a full rewrite. Thread on what I'm building (and why it matters) 👇 #BuildInPublic #FinTech #ForexDev

@ChaseMc67 @fchollet So we’re back at what it was 15 years ago

@fchollet We’re doing things with 6 people that would have taken at least 20 only a few years ago. Shipping stuff in a day that would have taken a week is the norm now

We need a "where are they now" series about the developers who claimed in February-March 2023 that they had become "100x" more productive thanks to LLMs. Surely by now they must have released multiple new apps in production and multiplied their earnings by a large factor?

@tdpauw @AlpesCraft It was great

Made a button 🙃 lab.hakim.se/clickmeifyoucan

This is a solid visual summary of how quantum computing works: ig.ft.com/quantum-comput…

- Il y a un trou de 7 mois dans votre CV ? - J'ai eu Chronopost, ils m'ont dit qu'ils m'envoyaient un colis. Fallait que je sois chez moi.