Skip to content

TODO.md

Latest commit

 

History

History
59 lines (47 loc) · 2.59 KB

TODO.md

File metadata and controls

59 lines (47 loc) · 2.59 KB
Raw

WheelHouse - Feature Backlog & TODO

🚀 Planned Features

Model Auto-Setup Wizard (First-Run Experience)

Priority: Medium
Added: 2026-01-23

First-time users should get a guided setup for AI models:

  1. Detection: Check /api/ai/status on app load → detect missing Ollama or models
  2. Setup Modal: Show wizard with:
    • Ollama installation status + install guide link
    • DeepSeek-R1 32B download button with progress bar
    • Alternative: Grok cloud API key setup
    • "Skip for now" option
  3. Backend: Add POST /api/ai/pull endpoint that streams ollama pull progress via SSE
  4. Persistence: Store "setup complete" flag in localStorage to skip on future loads

UI Mockup:

┌─────────────────────────────────────────────────────┐
│  🧠 AI Setup Required                               │
├─────────────────────────────────────────────────────┤
│                                                     │
│  WheelHouse uses local AI for trade analysis.       │
│                                                     │
│  ☐ Ollama (not installed)     [Install Guide]      │
│  ☐ DeepSeek-R1 32B (0/19GB)   [Download]           │
│                                                     │
│  Or use cloud AI instead:                           │
│  ○ Grok-3 (requires API key)  [Setup Grok]         │
│                                                     │
│            [Skip for now]  [Start Setup]           │
└─────────────────────────────────────────────────────┘

✅ Completed (Recent)

2026-01-23

  • Switch default AI to DeepSeek-R1-Distill-Qwen-32B (better quant reasoning)
  • Fix hardcoded r=0.05 → use state.rate for consistency
  • Roll tips now reference real CBOE chain via "Suggest Roll" button
  • Chain-aware Win Rate Dashboard (counts chains, not legs)
  • Portfolio Summary aligned with Positions footer (Net Premium, Capital at Risk)
  • Removed redundant Performance panel

💡 Ideas (Not Yet Planned)

  • Dividend yield support (q parameter) for high-yield stocks
  • Dynamic VRAM detection for model recommendations
  • Export trade history to CSV/Excel
  • Mobile-responsive layout
  • Dark/light theme toggle