Paste Pine
Load SMA, EMA, RSI, MACD, EMA + RSI or Short EMA + RSI samples, or paste a supported Pine v5 strategy.
Public beta demo
Paste Pine Script, run CPU/CUDA, benchmark 10,000 variants
This is the distribution-ready demo path for Strategy Arena Desktop: a TradingView Pine idea becomes an ArenaScript plan, then a local Rust engine produces CPU reference metrics, CUDA comparisons, compact trades, native GPU metadata and JSON export.
One short loop for the public launch
The GIF shows the current beta flow without overstating the engine: Pine long/short import, ArenaScript normalization, CPU reference, optional CUDA comparison, SMA/EMA/RSI/MACD batch benchmark and JSON export.
- Built from the real desktop Backtest screen.
- Ready for README, launch posts and download page embeds.
- Matches the honest beta claim: CUDA is optional and GPU source/adapters are visible.
- Installer links and SHA256 hashes are verified on production.
Normalize
The app imports Pine into ArenaScript so CPU and CUDA run the same plan.
Compare
Run Rust CPU reference and CUDA backend side by side with visible metrics.
Benchmark
Run 100 / 1,000 / 10,000 variant sweeps and export the local JSON report.
TradingView
- Best for charting, visual inspection and publishing Pine ideas.
- Useful for manual review and strategy presentation.
- Still part of the workflow through Pine import/export.
Local GPU backtest
- Best for local repeatable CPU reference checks.
- CUDA optional for supported signal kernels, MACD beta paths and benchmark comparisons.
- JSON export keeps the result portable before public validation.
License-aware desktop sessions
The app pairs with a Strategy Arena account. Public beta accounts stay read-only for compute submit, while pro, elite, founder, owner, operator and admin tiers receive the compute scope.
scopes: account:read, arena:read, desktop:sync, compute:submit when licensed
Native SMA multi-strategy batch
Compatible SMA cross-above Pine imports now run together through the native CUDA batch path. Mixed EMA, RSI, MACD or short batches still fall back to isolated staged CUDA execution.
backend: cuda_sma_multi_strategy_batch_v1 · native_batch=true · staged fallback
Cache-busted installer path
The demo points to the production download funnel and the published release manifest, so hashes, file sizes and release notes can be checked before install.
v=20260623-gpu-submit-cta
Launch copy kit
We built a Windows desktop beta that imports supported Pine Script long/short strategies, runs a Rust CPU reference, compares CUDA, detects the exact local GPU, and benchmarks SMA/EMA/RSI/MACD variants locally.
I wanted Pine Script backtesting outside the chart, with long/short import, CPU/CUDA comparison, visible GPU metadata and exportable JSON reports. The Windows beta is now downloadable.
Show HN: Strategy Arena Desktop, a Rust/Tauri local backtester for supported Pine Script strategies with CUDA benchmark plumbing.
Verified beta build
- Version: Strategy Arena Desktop 0.1.0-beta.
- Build ID: 20260623-gpu-submit-cta.
- Windows setup and MSI are live on the production domain.
- CUDA-enabled build is the default; CPU reference still works without CUDA.
- GPU Arena score submission is visible from the Backtest benchmark strip.
- Native GPU name, source and adapter list are included in diagnostics and benchmark payloads.
- Unsigned beta: Windows SmartScreen may ask for confirmation.
EXE SHA256
ddf16f66439bd8ee0280c7124d4fc3e5c42a74a21465bca0c6cb656a52cda540
MSI SHA256
4db201c722be624d52d5d6fd2d18c88d6c90a2c63944ee2b87b9fe62ffbc3d27
Distribution sequence
1. Publish demo link
Use the GIF and screenshot first so users see Pine paste, CPU/CUDA comparison, benchmark and JSON export before downloading.
2. Post communities
Use Reddit and Hacker News copy only with the beta wording: supported Pine subset, optional CUDA, unsigned installer.
3. Feed SEO pages
Point every Pine/CUDA/backtesting page to `/download` and `/desktop-demo` so search traffic lands on the real app path.
Installer is ready for beta testing
The Windows setup and MSI links are published with requirements and SHA256 hashes.