Delta‑Rice - Granular access to derivative for the masses
Inspiration
- Derivatives are powerful yet opaque; most tools highlight price, not structure.
- We wanted a clean “quant tape” that surfaces basis, funding, and regime shifts — then explains them in plain English.
What it does
- Real‑time multi‑asset dashboard for BTC, ETH, SOL, BNB, XRP.
- Computes spot–futures basis yields (perp/quarterly), deltas (bps), and Z‑scores with anomaly highlights.
- Detects macro funding/basis regimes over 90/180/365d and shows current position vs. mean/±σ.
- AI copilot (AWS Bedrock + Qwen) turns stats into concise narratives and Q&A.
How we built it
- Frontend: Next.js 16 + React 19 + TypeScript; Tailwind CSS v4 and shadcn/ui primitives.
- Data: SWR polling Binance (spot) and OKX (futures, funding, history).
- Analytics: On‑the‑fly basis, Z‑scores, regime rails; lightweight client/server compute.
- AI: AWS Bedrock Runtime with Qwen models for summaries and scenario prompts.
- Deploy: Vercel (Edge Runtime).
Challenges we ran into
- Normalizing Binance vs. OKX payloads and aligning timestamps for stable time‑series math.
- Calibrating Z‑score thresholds to reduce noise without hiding signal.
- Keeping the terminal‑style UI readable under constant updates.
- Managing LLM latency/cost while keeping responses contextual.
Accomplishments we’re proud of
- A cohesive basis/regime view that traders can act on quickly.
- Minimalist, responsive terminal UI that still feels premium.
- An AI copilot that explains anomalies with grounded context.
What we learned
- Small visual choices (spacing, color ramps) dramatically aid comprehension in dense UIs.
- Z‑scores work well but need asset‑specific calibration and periodic review.
- LLMs shine when grounded with fresh, structured features.
What’s next for Delta‑Rice
- Options surface (IV, skew, term structure) and cross‑asset spreads.
- Richer anomaly taxonomy with severity and alerts.
- Strategy blocks: simple basis/arbitrage playbooks with backtests.
- Team features: RBAC views and exportable reports.
Built With
- amazon-dynamodb
- bedrock
- next
- v0
Log in or sign up for Devpost to join the conversation.