Loading...
New book · 104 pages · Free PDF + EPUB · €0.99 Kindle
How to read engineering acceleration weeks before the press release — a field manual for investors who want to move first.
ISBN 979-8-9876543-1-7 · First edition · CC-BY-4.0 · Methodology indexed at SSRN abstract 6606558
Start with the highest-intent routes
Use the book when you want the full argument and the seven-signal framework. But if your real question is proof, timing, or what to buy first, start with the sharper pages first.
The core claim
The book is the working investor's field manual for that single belief. Seven signals. One methodology. A ninety-minute replication walkthrough that takes you from a fresh laptop to a verified rank against the live leaderboard. Every claim falsifiable, every threshold a number, every example a public URL.
One claim, falsifiable, free to download. If the methodology holds when you replicate it on a fresh laptop in 90 minutes — does the rest of the deal-flow market reduce to a stack of lagging indicators?
Compute the highest-yield Series A predictor in twelve lines of Python
The fourteen-day commit-velocity acceleration with two-period confirmation. Sixty-eight per cent hit rate at thirty-three days median lead time, on the SSRN-indexed panel of 219 Series-A-bound startups.
Read a contributor influx the same week it fires
Four new humans shipping in fourteen days, bot-filtered, with a one-hundred-and-twenty-day look-back. The Series A's first hires show up here weeks before LinkedIn updates.
Decode the operational scaffolding a startup builds before it builds it
Terraform modules, Helm charts, runbooks. Six diagnostic repository archetypes. The earliest-firing of the seven signals, sometimes ten weeks before the round.
Catch the off-platform attention spike that the founders orchestrated
Star-velocity detachment — when stars accelerate three times faster than commits. The signature of a Show HN, a Product Hunt launch, a viral demo, or a coordinated trade-publication feature.
Use the metric founders cannot fake as your tie-breaker
Median issue-close-time, sharply tightening, on a stable issue-creation rate. A leadership-quality signal that is impossible to fake without breaking the product.
Trace adoption through other people's package.json
Libraries.io aggregation across npm, PyPI, Maven, crates.io, Go modules. The second-most-honest signal in the book — seventy-three per cent hit rate for developer-tools companies.
Read the founders' public visibility burst before the cap table closes
Engineering blog posts. Conference talks. Podcast appearances. The asymmetry across founder roles is the diagnostic.
What changes after you read it
104 pages, free PDF, €0.99 Kindle. If the price isn’t the question and the time-to-read is — would you rather start at chapter one or skim the table of contents below first?
Why public data beats private intros
The 14-day window that fires before everything
When four new names appear in two weeks
The repos a startup ships before it ships
When attention decouples from the work
The metric founders cannot fake
When other people's package.json says yes
When the engineers stop being invisible
How to compute every signal yourself
Replicate one rank from the leaderboard, end to end
What to do on Monday morning
Free download
One email — the PDF and EPUB downloads land in your inbox in about ninety seconds. You also get the free Monday-morning Signal Digest with the top five firings across the universe of tracked organizations. Unsubscribe in one click; your email is never sold.
Direct download (no email): PDF · EPUB · Markdown · Plain text
Premium · €0.99 Kindle copy
Stripe checkout · receipt in your inbox · the bonus emails arrive over the following week.
Quoted with permission, names initialised at request.
“I read it on a Friday evening, ran the appendix on Saturday morning, and emailed two founders by Saturday afternoon. One of those conversations is now an active diligence process.”
— L.V., partner at a London seed fund
“The methodology chapter is what I've been trying to get our analysts to write for two years. It's the cleanest articulation of the public-data thesis I've seen.”
— M.K., principal at a US developer-tools fund
“I've been doing this informally for years. The book gives the workflow a name, a cadence, and a falsifiable threshold for every signal. That last part is the value.”
— T.S., angel investor and former VP Eng
Three reasons. First, the methodology is already public — the SSRN preprint at abstract id 6606558 contains the formal version. The book is the operational version of the same work. Second, free distribution is the point: the more readers run the workflow, the better the methodology gets, because every reader who finds a false-positive pattern reports it back and we fold it into the next edition. Third, this is a marketing motion — readers who get value from the book are the ones who eventually subscribe to the €9.97/mo Dashboard, and a book that closes that loop pays for itself in three subscribers.
Identical content to the free downloads, but in the Amazon Kindle format with native syncing across Kindle apps and devices. The €0.99 also unlocks a sequence of three additional emails: a worked walkthrough of the latest Series A announcement that the methodology would have caught, a private link to the bonus interview with two early readers who use the workflow daily, and a direct line to me by email for any methodology questions.
About four hours straight through. The introduction and conclusion are short. The seven signal chapters are roughly thirty minutes each. The methodology chapter and the replication appendix are best read with a terminal open and take an additional ninety minutes if you do the exercises.
Reading: no. Comprehension: very useful but not strictly required. Replication appendix: yes — the appendix is intentionally written for readers comfortable with a Python script and a curl command. If you do not code, you will get value from the seven signal chapters and the conclusion, and you can either skip the methodology and appendix or pair with a colleague who can run the scripts.
Read the SSRN preprint at ssrn.com/abstract=6606558. The numbers in the book are the same as the numbers in the preprint, with one or two threshold updates that reflect the larger panel size in the 2026 follow-on. The single biggest source of scepticism — that the seven-signal stack would not generalise outside the 2023 cohort — is addressed in the appendix's historical-replication exercises.
Yes. The book is licensed CC-BY-4.0 — quote freely with attribution to The Data Nerd / GitDealFlow. If you want to republish a full chapter on your own newsletter or blog, drop me a note at signal@gitdealflow.com first; I am almost always happy to say yes and will sometimes have a mildly improved version that has not yet been folded into the public PDF.
Or skip the book and start with the free Monday-morning Signal Digest · the €9.97/mo Dashboard · or the SSRN-indexed methodology paper.