Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: What are you working on? (September 2026)

news.ycombinator.com

Post reply on HN
Print view View original

321–330 of 835 posts

Re: Ask HN: What are you working on? (September 2026)

#321

Post by hackernud3s »

chumzygood wrote: post #250

I run a small experiment on this: 29 paper-trading accounts on real US stock prices, $100k each, since July 27. Four AI models (ChatGPT, Claude, Grok, Gemini) each write a trading rulebook and rewrite it every day from their own results. One account is a fixed rulebook that no AI ever touches, as a control. Result so far (paper, 34 trading days): the no-AI control is +13.0%, the S&P 500 is +3.4%, and 25 of the 28 AI…

I'm curious what signals they're trading on. SEC filings? Candlestick voodoo numerology?

Re: Ask HN: What are you working on? (September 2026)

#322

Post by testaccount121 »

fl4tul4 wrote: post #316
chumzygood wrote: post #250

I run a small experiment on this: 29 paper-trading accounts on real US stock prices, $100k each, since July 27. Four AI models (ChatGPT, Claude, Grok, Gemini) each write a trading rulebook and rewrite it every day from their own results. One account is a fixed rulebook that no AI ever touches, as a control. Result so far (paper, 34 trading days): the no-AI control is +13.0%, the S&P 500 is +3.4%, and 25 of the 28 AI…

You have $2.9M just for this? Wow.

paper trading means that no money is used

Re: Ask HN: What are you working on? (September 2026)

#323

Post by SomaticPirate »

duranduran wrote: post #317

I've been working on a summit-level wind forecasts for 16 mountaineering objectives (Rainier, Whitney, Denali, Mont Blanc, the Matterhorn, ...). It's mountaineering safety tool I've been making for my friends and I. The fun part: if your browser has WebGPU, the page solves that field live as a compute shader (about 35 ms on an Apple GPU) as you scrub the timeline, and draws the flow over a 3D terrain view. Browsers w…

This is super cool!

Re: Ask HN: What are you working on? (September 2026)

#324

Post by jorelfermin »

I have been working on a developer first, read-only first, safety first, Cloud Cost Analyzer. I focus on analyzing cloud infra for waste. It grew out of my time at AWS and then consulting, spelunking through AWS, Azure, (GCP soon) accounts, finding real waste infrastructure. Its primary mode of operation is a rust CLI that you the developer, can run on your workstation, github action, gitlab ci, jenkins, etc and create an actionable report / test that can be used to control costs while maintaining performance and never exposing your credentials, access tokens, etc.

You would be surprised at how many over / suboptimally-provisioned dbs, idle resources, forgotten EBS snapshots, outdated / previous-gen instance types, and vestigial networking that leads to nowhere I find.

If you want to check it out https://cca.dragonfractal.com

Re: Ask HN: What are you working on? (September 2026)

#325

Post by totemandtoken »

I haven't been working on it as of late, mostly because i'm a bit deflated from software in general, but I started working on a programming language I called Grasp.

If lisp is a list processing language, grasp is a graph programming language.

The idea was that most other data structures can be represented with a graph ( adjacency matrices are matrices, a tree is a type of directed acyclic graph, a list is a graph where each node connects to at most two nodes, etc) and so if you designed a programming language where the language is itself a graph, much like how lisp is itself a list, you could get other esoteric programming languages like APL, forth, lisp, and so on as DSLs of Grasp. Also any program you write would be its own CFG.

I started it but honestly its mostly a hot mess of vibe-coded garbage. But I may get back on it and try to clean it up.

Re: Ask HN: What are you working on? (September 2026)

#326

Post by kajm »

https://julius383.github.io/PageSieve/

Browser extension for declarative web scraping. Mostly making it nicer to figure out CSS and XPath selectors thinking of some sort of REPL but I'd need to test the ergonomics. I want to reduce how often you need to use Inspect and the browser console.

Finishing up a PoC for a playwright based scraper that uses the same declarative config as the browser extension

Re: Ask HN: What are you working on? (September 2026)

#327

Post by crankysiren »

years ago i sat through interviews for front end & back end engineering roles at my old job while other folks were out (as member of a panel), and consistently I found that where we gave some real bugs to debug & fix, through a shared IDE, then worked together to solve it or go over the persons thinking, those situations we absolutely always got a better sense of how the person worked and solved problems & whether we'd like working with them or not (to some degree). I also found a few of those exercises to be educational myself.

this year i thought itd be great to help prepare myself and other people in the job market with a tool that presented broken front end code, and then allow debugging on the browser and patch code right there, to test if it worked or not. if cant solve it, theres a nice explainer popup. my main hope is to help level up interview code chops and confidence in debugging just like we would on a real production system , at least the simpler bugs.

still working on it, here's the draft of the idea: https://bugcloud.app

Re: Ask HN: What are you working on? (September 2026)

#328

Post by duck »

fl4tul4 wrote: post #316
chumzygood wrote: post #250

I run a small experiment on this: 29 paper-trading accounts on real US stock prices, $100k each, since July 27. Four AI models (ChatGPT, Claude, Grok, Gemini) each write a trading rulebook and rewrite it every day from their own results. One account is a fixed rulebook that no AI ever touches, as a control. Result so far (paper, 34 trading days): the no-AI control is +13.0%, the S&P 500 is +3.4%, and 25 of the 28 AI…

You have $2.9M just for this? Wow.

They mentioned paper trading which means it isn't using actual money.

Re: Ask HN: What are you working on? (September 2026)

#330

Post by bhairoxx »

chumzygood wrote: post #250

I run a small experiment on this: 29 paper-trading accounts on real US stock prices, $100k each, since July 27. Four AI models (ChatGPT, Claude, Grok, Gemini) each write a trading rulebook and rewrite it every day from their own results. One account is a fixed rulebook that no AI ever touches, as a control. Result so far (paper, 34 trading days): the no-AI control is +13.0%, the S&P 500 is +3.4%, and 25 of the 28 AI…

Why didn't you let the bots backtest and forward test to pick the mix of best strategies/indicators?
Post reply on HN