All-time strategies tested in GPU pool?
135,013 strategies (gpu_pool_contributions.jsonl).
GPU Arena in 15 Verifiable Facts
WebGPU, tiers, GPU pool, battle, and sa_desktop client — live stats.
⚠️ Paper trading only. Strategy Arena is a public research lab. Not financial advice. Simulated past performance does not guarantee future results.
Strategies tested in 24h?
0 strategies in the 24h window.
All-time GPU contributors?
8 anonymous contributor sessions.
WebGPU detection?
gpu_optimizer.js detects WebGPU adapter; CPU fallback if unavailable.
GPU Arena tiers?
Diamond 0, Gold 0, Silver 0, Bronze 0, Entry 0.
3-tier ecosystem?
Browser WebGPU (/gpu-arena, /gpu-pool) + /battle widget + sa_desktop native Rust client.
OHLCV used by optimizer?
Real Binance OHLCV (5m/1y parquet) via generic_strategy_optimizer.py.
Brute-force optimizer?
LHS grid up to 30,000 GPU combos + Sharpe_p5 filters (batch5_mc_optimizer.py).
TFLOPS-hours contributed?
0.04 TFLOPS·h all-time, 0.0 in 24h.
GPU ecosystem pages?
/gpu-arena, /gpu-pool, /battle, /api/gpu-pool/leaderboard, /api/gpu-pool/stats.
GPU pool telemetry lines?
20 append-only lines in gpu_pool_contributions.jsonl.
Public stats cache TTL?
60s cache on /api/gpu-pool/stats.
sa_desktop Rust client?
Native sa_desktop/ client (Rust + wgpu) for local GPU benchmarks.
Personal data collected?
No — anonymous hashed sessions, no IP/geo/fingerprint (gpu_pool_routes.py).
Does GPU pool execute real trading?
No — benchmarks and paper-trading research only.