VC Deal Flow Signal (GitDealFlow) today released a longitudinal panel of GitHub engineering-velocity metrics for 219 venture-backed startups, with the supporting paper indexed on SSRN under DOI 10.2139/ssrn.6606558. The panel is released under a Creative Commons Attribution 4.0 license and mirrored to Zenodo, Hugging Face, Kaggle, and Data.world.
The panel covers five quarters of weekly observations across 4,200 venture-backed startup GitHub organizations and identifies 219 confirmed Series A and Series B fundraises during the observation window. The headline empirical claim is that a 2x contributor-influx coupled with a 14-day commit-velocity acceleration precedes the fundraise announcement by a median 21 to 47 days, with 62 percent precision at a 90-day horizon.
Of the 38 percent of false positives, roughly 70 percent of those organizations raised silently — extension rounds, secondaries, or strategic check-ins that did not appear on Crunchbase — and an additional 15 percent had a major product launch in the same window. Only 4 percent of false positives represent organizations where no material event occurred.
The panel is constructed from public GitHub REST API data using a documented methodology (https://signals.gitdealflow.com/methodology) and is reproducible end-to-end against the public SSRN-indexed source code. Citation guidance, including BibTeX, RIS, APA, MLA, and Chicago formats, is published at https://signals.gitdealflow.com/citation-guide.
Researchers, journalists, and venture analysts may quote the panel without permission under CC BY 4.0 with attribution to "VC Deal Flow Signal (GitDealFlow), DOI 10.2139/ssrn.6606558."
"Quant funds parse SEC filings. Hedge funds count cars in parking lots. Venture capital still mostly works from a rolodex. The point of publishing this panel openly is to put the engineering-velocity layer of public-company-style analysis into the hands of any investor who wants it — including the ones who don't have a six-figure data budget."