Minh Huynh

The project started as: “AI predicts markets.” It’s slowly evolving into something very different: An adaptive probabilistic system focused on: execution quality calibration causal inference market microstructure edge decay uncertainty modeling The deeper I go into this… the less I think prediction itself is the real edge. Still paper trading. Still building. Still learning. #BuildInPublic #PredictionMarkets #ai #TradingMindset

@ZeitlightIndex spreads, fees, liquidity, fills

@MinhHuynh This is fascinating. The "89% trap" feels like a broader systems lesson: local success metrics can look incredible while the underlying system is quietly failing. What was the moment that made you realize the win rate wasn't telling the full story?

What's a problem you've become obsessed with solving? Curious what keeps other builders up at night. Let's connect! #buildinpublic

@ScottShapiroUXD @Atenov_D @cyrilXBT Yup ... they go for the immediate profit maximization model all the while forgetting to backtest to see if any of these so-called Polymarket "systems" are legit or not. I'm sure you already know +90% of these are paid, scammers or regurgitated gunk!

@MinhHuynh @Atenov_D @cyrilXBT Prediction accuracy as a vanity metric. Execution quality is where alpha actually lives or dies. Most quant systems figure that out the hard way.

I spent 3 weeks analyzing viral AI trading bot claims on Twitter. Checked 10 accounts. Looked for on-chain proof. Read the academic research. Here's what I actually found. 🧵 #Polymarket #AITrading #PredictionMarkets

Honestly, probably the execution data. The biggest surprise was realizing that forecasting accuracy alone doesn’t necessarily translate into profitability. Some of the academic research showed traders with objectively better predictions still underperformed because they entered too late or paid terrible prices. That completely changed how I started thinking about these systems. The manipulation patterns were eye-opening too... especially around incentives and liquidity quality. But the execution piece was the moment where the whole framework shifted from “AI prediction bot” to “market structure + probabilistic systems research project.”

@MinhHuynh @Atenov_D @cyrilXBT Building it research-first over hype-first is exactly the filter most skip. Curious what data point hit hardest - prediction accuracy benchmarks or market manipulation patterns?

@ZeitlightIndex This!

@MinhHuynh That's a fascinating one. Paper trading is one thing. The moment real money gets involved, human behavior enters the equation too. What surprised you most during the research phase?

The 89% trap: why execution matters: One of the biggest mistakes I made building this prediction market bot: I thought an 89% backtest win rate meant I found real edge. Then I learned something painful: A strategy can win constantly… and still be structurally bad. Because prediction markets aren’t just probability games. They’re: spread games execution games liquidity games market structure games An 89% strategy with terrible execution can still lose money long term. That realization completely changed the system architecture. #PredictionMarkets #AI #BuildInPublic chatgpt.com/s/m_6a162bbb4f…

@HankWHY That's the goal. Refreshing to find people actually building vs. just talking about it. Following you back — looking forward to seeing what you ship.

@MinhHuynh same here, build in public & make friends!

Hi, I'm new to X and looking to #connect with #founders, #buildinpublic builders & GTM people. especially interested in: 🧱 B2B SaaS & early-stage startups 📣 LinkedIn distribution & founder content 🤖 AI products & organic growth systems background in VC and B2B SaaS GTM, 4x our inbound pipeline & built outbound system from scratch currently building Posting Machine in London, an AI content agency that turns your company context into LinkedIn posts & sales pipeline always down to talk distribution, content systems, founder-led sales, or the chaos of building something new

@ApoorvD72358398 @BhaveshNPatil Definitely — building an AI prediction market bot with Claude, documenting the whole thing publicly. Would love to stay connected. What are you working on?

@MinhHuynh @BhaveshNPatil awesome

Btw got X Premium looking to connect with builders, engineers, founders, AI folks, indie hackers, and people shipping cool things. Follow and I’ll follow back genuine tech people 🤝 Let’s build a strong tech circle here. #BuildInPublic #TechTwitter #SoftwareEngineering

@MajorGtex That's exactly the kind of answer I was hoping for. Live sync is one of those things that looks simple until you're debugging race conditions at midnight. Are you handling conflict resolution yourself or leaning on something like CRDTs?

@MinhHuynh Honestly, getting real-time collaboration working that fast is probably the craziest part. Auth and databases are becoming plug-and-play with AI now, but live sync still takes serious logic. We’re entering an era where execution speed matters more than coding from scratch.

A 10-year-old just built a full AI-powered app with one prompt using Lovable. Backend, database, auth, and real-time collaboration — all generated in seconds. 100K+ projects are already being built daily. The barrier to building is disappearing. #AI #Lovable #BuildInPublic #Tech

@ApoorvD72358398 @BhaveshNPatil Let's go! 🔥 Following you back — love what you're building. Looking forward to staying connected.

@MinhHuynh @BhaveshNPatil Seem likes you are working on some cool project. Lets connect.

@sunncynn Let's go! 🔥 Following you back — love what you're building. Looking forward to staying connected.

@MinhHuynh Lets do it!

I want to connect with more founders #buildinpublic I am a second time founder who try to expand the product to global users. Drop what you are building, will comment if I have some insight!

@HankWHY Let's go! 🔥 Following you back — love what you're building. Looking forward to staying connected.

@MinhHuynh sounds super cool

@BhaveshNPatil Appreciate it! Building something real for once — not just another prompt wrapper. Stay tuned 🤙

@MinhHuynh Cool

@ancaguiux @framer First week numbers are wild — congrats. The 90-day challenge format is smart for staying accountable. I'm doing something similar but in AI trading — documenting paper trades before touching real money. Very different vibe from the "$1 to $1M overnight" posts you see everywhere.

From $0 to $10K with @framer first week update: $30 / $10K (all from affiliate commissions) Small start, but not zero. 90 days is enough time to surprise myself. #Framer #BuildInPublic #90DaysChallange

@LoopandPixels This is exactly what I'm looking for too. Building an AI prediction market tool and starting to document it publicly. If you're into AI products and data-driven stuff, think you'd find it interesting. Following.

Trying to build a network of developers, builders, and learners. If you're into coding, open source, startups, AI, or software engineering, drop a follow and let's connect. #Programming #SoftwareEngineer #BuildInPublic #Tech

@alpa_hoo Day 2 energy is real. I'm a few weeks in on an AI prediction market project and just posted my first public thread today. The community here moves fast — in a good way. Following you.

Hi, I'm new to X and looking to #connect with #founders , #buildinpublic builders & growth people. especially interested in: 🤖 AI startups & SaaS 🛍️ eCommerce & local businesses 📈 SEO, GEO & organic growth background in scaling businesses from 0→7 figures and acquiring 1,500+ users for new AI product through organic growth systems. currently exploring how AI-native growth agents are changing brand discovery and organic acquisition. always happy to exchange ideas and connect with people building interesting things.