Skip to content

feat: add bench 117/118/119 - gains showcase trio - #1715

Merged
Flotapponnier merged 1 commit into
devfrom
feat/gains-bench-trio
Aug 2, 2026
Merged

feat: add bench 117/118/119 - gains showcase trio#1715
Flotapponnier merged 1 commit into
devfrom
feat/gains-bench-trio

Conversation

@Flotapponnier

Copy link
Copy Markdown
Collaborator

Summary

  • Bench 117 - perp-cost-slope: derives cost slope ratio ($1M / $1k all-in bps) from existing perp-fees data via PromQL, no new harness needed. Oracle-priced venues (gains, GMX) always score 1.00; orderbook venues > 1.00
  • Bench 118 - perp-mark-price-lag: new Go harness sampling mark price vs Binance spot mid every 60s across 6 venues x 3 assets (ETH/BTC/SOL). Pyth feed used as oracle proxy for gains/gmx
  • Bench 119 - perp-protocol-longevity: embedded incident registry (rekt.news sourced), computes live "days clean" counter. gains: 1705 days clean since Dec 2021; GMX last incident Jul 2025 (~$42M)

Harnesses deployed to VPS (ocb-par-main), Prometheus updated and reloading.

Test plan

  • Verify bench 118 container healthy: docker logs ocb-perp-mark-price-lag
  • Verify bench 119 container healthy: docker logs ocb-perp-protocol-longevity
  • Check Prometheus scraping both targets at prom.openchainbench.com
  • Validate all 3 spec YAMLs pass Zod schema on staging
  • Confirm bench 117 queries return data (from existing perp-fees metrics)

…eviation, protocol longevity)

bench 117 perp-cost-slope: derives cost slope from existing perp-fees data, no new harness
bench 118 perp-mark-price-lag: new Go harness sampling mark vs Binance ref every 60s for 6 venues x 3 assets
bench 119 perp-protocol-longevity: embedded incident registry, 1705 days clean for gains vs 389 for gmx
@Flotapponnier
Flotapponnier merged commit a56d89e into dev Aug 2, 2026
1 check failed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant