I think I just launched my funds management business. I suppose I'd better do an economics and finance short course. Are my math-IT and psychology stats good enough?
Now I just need to put $5000 in it for phase 2. The performance is consistent. (The one loss listed only happened due to an ROI calculation oversight, which is now fixed. The ROI is now based upon the spread floor, not the general optimistic ROI, and it's switchable between the two.)
The burn rate for AI API tokens is the same whether I am trading out of a $50 tiddler pool or a $5million whale pool. It's the exchange fees that are the main caution, but they're fixed at percentage so provided the ROIs are healthy they're not a serious problem (unless an exchange gouges, and then we just move to another exchange.)
At some point hitting the ASX directly will be necessary. That's phase 2-3.
Proof of concept and prototype are done.
If you're interested in buying this codebase, the price is $USD5000 white label with no guarantees (you'd be buying a phase 1-2 fully working proof of concept with comprehensive strategy and technical documentation. You should be able to point your high grade LLM at it and get what you want). I will retain rights to use and develop it myself. No exclusive sale will be made unless a meaningful offer for full contractual ownership (>USD$250000) is made.
You could point your LLM at the UI and re-engineer something. Based upon what was involved in getting it working - and upon legal impositions - I really don't recommend it.
I favour Claude Opus 4.6 for most of the updates now, but Gemini 3.1 Pro (high setting) does a fine job most of the time. Watching the Clarkean technomagic AI work is inspiring and terrifying. I love SpaceX rockets and neuralink, but this thing is epic.
While I was writing this post I had the LLM working on an update allowing me to use other model gardens. It was finished before I started typing these two sentences. (And it had added a couple of mechanical toggles for switching off buying or selling to boot.)
