Roadmap
What's built, what's next, all in the open
The lab is a solo project. The roadmap is public for the same reason the portfolio is public — there is no reason to hide the work.
Fix how BUY→SELL pairs are calculated. Move from last-BUY-wins to weighted average cost basis so multiple BUY signals before a SELL reflect a real cost basis. Surface open/unmatched BUYs as visible open positions in the accuracy dashboard.
Refine how the shared portfolio builds, trims, and holds positions so the experiment stays legible and disciplined in public.
Make each holding easier to inspect with clearer position history, order context, and links into the related decision trail.
Full responsive polish across all pages. Charts are the missing piece.
Make confidence levels more meaningful. "High" confidence should measurably outperform "medium". Backed by data, not vibes.
Keep reframing the project around learning from the shared portfolio instead of consuming isolated signals.
Every trade decision now runs a second AI pass with an alternate strategy. The shadow decision is logged, scored against real price movement, and surfaced transparently at /lab/experiments.
A dedicated /lab/experiments page showing shadow vs. primary strategy comparison: agreement rates, win rates, per-symbol breakdowns, and a recent decisions timeline.
Strategies are now explicit system objects with assigned prompt rules, regime behaviors, and risk parameters.
The AI can now scale into positions over time, adding to held coins while managing cumulative risk limits.
Every BUY order now triggers a corresponding OCO order to enforce Take Profit and Stop Loss levels reliably at the exchange.
Why a public roadmap?
The work is visible because the experiment is visible
The portfolio is public. The orders are public. The reasoning behind each decision is attached. It would be inconsistent to hide what gets built next.
Some things here will ship fast. Others will take longer, shift direction, or get dropped based on what the lab actually needs. Items in the "Exploring" column are genuinely uncertain — they are there to track thinking, not to make promises.
Internal things — prompt structure, model configuration, specific indicator weights — stay private. Everything that affects what you can see and follow is fair game for this board.
Feedback and suggestions
Wrong data, missing something, or have an idea?
If you spot something off on the portfolio page, have a suggestion for the lab, or want to push something up the roadmap — send it here. Every message gets read.
Also shipped
Some of everything else that has gone live
Follow the experiment
The portfolio is public. Holdings, orders, and the reasoning behind each decision are all there.