Comment by saratsai | Hacker News Reader

Comment by saratsai | Hacker News Reader

I built Neuberg because I’d always wanted a Bloomberg Terminal, but as an independent trader, I just couldn’t justify paying $24,000 a year for it. So I decided to build my own with Claude Code.

It’s a real-time trading terminal covering fixed income, derivatives, commodities, equities, credit, macro, and alternative assets, with 516 panels in total. You can freely drag, drop, and arrange them however you like.

What it does: Scrapes financial news and uses AI for sentiment analysis, including first- and second-order impact detection Offers 516 market data panels, such as Treasury auctions, option surfaces, CLO analytics, GDP nowcasting, and dark pool data Lets you trade directly on Hyperliquid (49 stock perpetuals) and Polymarket (prediction markets) Supports stock trading through Alpaca, both paper and live Includes TradingView-style charting with indicators like RSI, MACD, Bollinger Bands, ATR, and VWAP Features a world map showing geo-located news events and conflict-zone heatmaps

A few honest caveats: Most of the 486 newly added panels currently use seeded or simulated data. The UI is functional, but not all of them are connected to live data feeds yet. The core 30 panels — including news, charting, trading, heatmaps, calendar, options flow, and insider trading — use real data. This is a solo project, so there are still some rough edges. It’s licensed under BSL 1.1, which means it’s free for non-commercial use.

In about three weeks, I expanded the platform from 30 panels to 516, mostly by vibe-coding with Claude Code. Around 90% of the panel code was written by AI, while I reviewed and integrated everything. I’d really love feedback on the architecture, UI, or panel coverage, and I’m happy to answer any questions.

Source code: GitHub - KoNananachan/Neuberg: Real-time trading news terminal with AI analysis, prediction markets

If I understand correctly, you had 30 working panels for the core functionality. You then vibe-coded an additional 486 new panels that use fake data? That's not quite how you presented the project in the title.

What do you mean? It says used AI in the title. That already implies it doesn't actually work.

Strong disagree on AI implying it doesn’t actually work. Staff engineers at FAANG are vibe coding these days and their stuff absolutely works.

> Staff engineers at FAANG are vibe coding these days and their stuff absolutely works.

60% of the time it works 100% of the time

You really want to claim that with the amount of outages lately?

We are trending towards zero nines (not 100) uptime.

I use Gemini CLI and given the might of all their engineers and AI tools, the features they ship are constantly buggy and broken.

It's so easy to pump out borderline useless functionality with AI that nobody is actually testing or designing it.

Most of the value of the Bloomberg terminal is the data, not the number of panels

this.

Nonsense, when he'll have one panel per screen pixel, he'll be able to see over 8 million Fibonacci retracements, 40 heatmaps and real-time market sentiment headlines at once on a 4k monitor, then you'll see.

... and speed. They publish billions of ticker items in near real time simultaneously to all their users... the tech is pretty crazy!

it's all about the bas

as someone who worked at TR (now Refinitiv) and then Bloomberg, it's not even the data. it's IB. After IB it's the data. :)

As someone that worked for the spin out from GS, then got acquired by TR, then sold off to Blackstone, I wonder how much of our tech debt is still there for you to deal with 10 years later. :)

As someone who never worked anywhere: what’s IB?

Chat.

https://professional.bloomberg.com/products/bloomberg-termin...

I haven't been at TR in 10+ years as well. AT TR, I mostly dealt with Eikon's pipeline for estimates, actuals and the like (IBES data) ingestion, querying and analytics. So can't say much, though some of my old colleagues still seem to be there.

I'm not sure if this is a coincidence or not, but today my github account was removed from the organisation... after 10 years.

The thing here is I don’t think you’re going to be able to get your hands on the data that the Bloomberg terminal provides especially not near real time live streaming data which is the reason people pay $25,000 a year. You’re not really paying for the interface, that is easy.

Can you please get rid of all the fake data panels? I will use it but only if I can trust that the data I am looking at is real.

UI

I’m curious about the UI development. What sort of prompts did you give? Ie look like this or look like that? Or did you specify the qualities?

This is great work! Where are you getting your data from? Finding quality data sources is difficult

As far as I know, finding quality data sources is virtually impossible unless you already have so much money that your involvement in the stock market is either because you have insider knowledge or severe boredom owing to your never needing to work a day in your life.

I remember some startup selling a stock market data API as a subscription. I don’t think they exist anymore. So anyone who spent weeks, months of their free time building an app around that API is now completely shit out of luck.

I suspect the real APIs are still running the same code they ran in the 90s and if you have to ask how much they cost, you can’t afford them.

You can get real-time market data through Databento. Been doing it for years. Top of book suffices for most people.

You can buy historical data at least from some reputable vendors, although you're still responsible to understand their collection process (sampling frequency, timestamp conventions, corporate action adjustments etc etc), as even 'obvious' things like how daily stock levels are reconstructed based on intraday data can mess up your analytics really bad. I don't think it generalizes to real-time data.

I open the site and nothing is loading and all I see is a grid

Reply on news.ycombinator.com