I built an open source AI hedge fund 🤯
A panel of AI analysts + quant models that research tickers and form real market views
Then a portfolio layer turns those views into actual positions
Drew inspiration from @virattt's AI Hedge Fund repo - I recommend checking it out ✨
Whats inside:
◆ AI analyst panel
◆ Quant research models
◆ Portfolio construction
◆ Full audit trail
◆ Backtesting
◆ Risk controls
Tech stack:
🌐 App: @nextjs
☁️ Hosting: @vercel
🎨 UI: @shadcn
🛢️ DB: @neondatabase
I have tried to use @emilkowalski motion skills for this and im pleased at the state that we are in right now
Right now its running on mock values, but im using a this myself daily on the Norwegian stock market 🇳🇴
Demo and Github repo in the comment.
Built this as a weekend project. Really want to make it into something serious ✨
I sold two SaaS companies in one year while still closing real estate deals in Norway ✅
No funding round.
No Goldman analyst background.
Just a broker job by day and a laptop by night.
Here are the 3 things that actually moved the needle:
1. Your day job is reps, not a distraction ✅
Every phone call, every negotiation, every “no” in real estate taught me how to sell software nobody asked for yet.
I didn’t separate “real work” from “side project work.” It was the same muscle.
→ Close a deal in the morning
→ Ship a feature at night
→ Repeat for years
The compounding isn’t in the startup. It’s in the reps you stack outside it too.
2. Leverage beats hours worked ✅
Our software didn’t sell because I worked more hours than everyone else (but i worked alot). It sold because we built something that ran without me having to be in every conversation.
3. Talk to users before you finish building ✅
My last software didn’t start as a finished plan. It became a real exit because I talked to the people who would actually use it before I kept building.
Most people build for two years and then ask if anyone wants it. I did the opposite.
→ Talk to someone who has the problem
→ Build the smallest version
→ Ask again before you build more
Your product will never be better than the conversations you had before you wrote the code.
Share the process, not just the outcome. That’s what actually gets you into the rooms that matter.
This was my 50cents for today ✨
So i found the problem for me to ship more content. Its not the lack of ideas and video (I got a lot of video content)
The problem is that i cant edit. So we need to fix that.
I need to do a developer move on this one ✨
@elie2222@emilkowalski Used Claude design, but i feel like its always afterwards involving a lot of transition from CSS -> Shadcn componetns and the typography is all over the place.
I feel it works 10/10 - The prototype is to good tbh
My new workflow
/prototype from @emilkowalski and iterate on this until we are ready. This gives you 3 examples, you pick one and do a new prototype on that.
/ui-review what we have done, this goes over the screen and fixes the the paddings and margins.
/ui-polish fixes everything. This was one is crazy tbh.
Makes me a new site or feature in 10 mins.
What a crazy time we live in.
My prediction on the new way of getting views on social media after a deep dive in the algo this weekend.
And i predict that "Interest media" has taken over for Sosial media
Your feed doesn’t care who you follow anymore.
It only cares if the post is relevant right now.
Same post.
15M followers → 300k views.
Zero followers → 9.4M views.
The algorithm judges the post. Not the poster.
So if you’ve been holding back because you didn’t want to spam your audience… stop.
In interest media the boring stuff simply never shows up.
Only the good stuff finds the people who actually want it.
Post the idea.
The algorithm does the rest 📷
New mini-project coming along. This time i did something different.
Been studying UI and what feels «right» And i want those app to be polished before the release.
The new skills from @emilkowalski has been a big help on this ✨