Frequently Asked Questions

Get answers to common questions

What models are used for chart analysis?

We use custom YOLOv5 and ResNet models for candlestick detection, liquidity sweep identification, and structure recognition. All models are trained on synthetic datasets and exported to ONNX format for efficient inference.

Is my data safe with OpenSignal?

Yes! OpenSignal is decentralized and Nostr-native. Your private key (nsec) never leaves your device. Screenshots are processed locally, and signals are published directly to Nostr relays.

How accurate are the AI predictions?

Our models achieve 90%+ accuracy on synthetic test sets. Real-world performance depends on chart quality, timeframe, and market conditions. We recommend using signals alongside your own analysis.

Can I use this on desktop?

Yes! OpenSignal is built with Kotlin Multiplatform, so we provide both Android and Desktop versions. The desktop app has the same features and UI.

Is there an API for integrations?

The FastAPI backend can be deployed separately. See our API documentation for endpoints, authentication, and example requests.

How do I publish signals to Nostr?

After analyzing a chart, tap "Publish Signal" to generate a Nostr event and send it to your configured relays. You need a Nostr account (public key) to publish.

Technical Stack

Built on proven, modern technologies

Frontend

  • Kotlin Multiplatform
  • Jetpack Compose
  • Android + Desktop

Backend

  • FastAPI (Python)
  • ONNX Runtime
  • OpenCV

Protocols

  • Nostr Protocol
  • NIP-96 (Blossom)
  • NIP-07 (Auth)