Three weeks ago I started turning the methodology behind GitDealFlow into a proper trade book. It shipped today. 104 pages, free PDF and EPUB, also fully readable on the open web.
Here is what is in it and why I wrote it.
The premise in one paragraph
Public GitHub data — commit logs, contributor graphs, dependency trees, infrastructure repos — fires three to six weeks before a typical Series A announcement. The signals are computable from a single free REST endpoint, by anyone, on a $0 budget. The seven-signal stack in the book has a 68% hit rate at a 33-day median lead time on the SSRN-indexed panel of 219 Series-A-bound startups. None of it requires a private network, a paid data licence, or a relationship with the founder.
The seven signals
- Commit-velocity acceleration (14-day window, two-period confirmation, +200%).
- Contributor influx (4+ new humans in 14 days, 120-day look-back, bot-filtered).
- Infrastructure repository buildout (Terraform / Helm / runbook / proto / internal-tools).
- Star-velocity detachment (stars accelerating 3× while commits stay flat — orchestrated attention).
- Issue closure cadence (median time-to-close, sharply tightening).
- Downstream dependency adoption (Libraries.io aggregation across npm / PyPI / Maven / crates.io).
- Founding-team public visibility (engineering blog cadence, conference talks, OSS maintenance).
Each signal chapter has the formal definition, threshold guidance, false-positive patterns, a worked example, and exercises.
What's in the appendix
A 90-minute replication walkthrough. From a fresh laptop, a personal access token, and pip install requests to a verified Scout Score against the live leaderboard. About a hundred lines of Python. Every primitive (paginated commit fetcher, contributor classifier, repo metadata cache) is reusable.
If you finish the appendix, the methodology is yours — replicable on a $0 budget, indefinitely, no dependence on my servers.
Free or €0.99
The book is free in PDF, EPUB, Markdown, and plain text on signals.gitdealflow.com/book. The €0.99 Kindle copy adds three bonus emails (a worked walkthrough of the most recent Series A catch, the unedited interviews with two developer-investors who use the workflow daily, and a 30-day direct line to me for methodology questions). Same content, different bonus stack.
ISBN 979-8-9876543-1-7 · CC-BY-4.0 license. The methodology paper is on SSRN at abstract 6606558. The dataset is mirrored on Zenodo, Hugging Face, and Kaggle.
Why free
Free distribution is the point. Every reader who finds a false-positive pattern reports it back, and the next edition gets better. Readers who get value from the book are the ones who eventually subscribe to the €9.97/mo Dashboard — a book that closes that loop pays for itself in three subscribers.
Read it on a flight. Skip to the appendix. Run the script. The Monday-morning workflow looks completely different inside a quarter.
If you have a methodology question, comment here or reply to any email from signal@gitdealflow.com. The next edition folds in every reader correction with attribution.
