VC Deal Flow Signal

Track 350+ venture-backed startup GitHub organizations across 15 sectors using engineering-velocity data. Open methodology, live dashboard, MCP server.

The finding: engineering acceleration plus low contributor concentration preceded Series A announcements by 21 to 47 days (3.4x lift, n = 219 confirmed funding rounds). The full methodology, cleaning protocol (bot contamination, survivorship bias), and the pre-registered prediction ledger are published and auditable.

What is here

• Signal engine configuration and sector taxonomy
• The public prediction ledger, including a 0-for-10 first cohort published anyway
• MCP server wiring for agent access to the deal flow data API
• Benchmark data behind the sector benchmark pages

Live surfaces

• VC Deal Flow Signal - the signal dashboard (weekly refresh)
• GitDealFlow - product home
• commit-velocity methodology - how the signal is built
• the public ledger - prediction track record, misses included

Citation

The dataset paper: SSRN 6606558, DOI 10.2139/ssrn.6606558. Zenodo dataset: 10.5281/zenodo.19650920 (CC BY 4.0). Cite the data, not this repo.

Resources

• MCP server: npx -y @gitdealflow/mcp-signal
• JSON API: https://signals.gitdealflow.com/api/signals.json
• Checklist companion: vc-due-diligence-checklist