The Data Nerd, social mascot bible

twitter

Signal of the week: ShadcnUI's 14d/90d commit-velocity ratio is 1.62, third consecutive week above 1.5.

Public-data only. github.com/shadcn-ui/ui → Insights → Pulse → 1 month, then 3 months. Compute (weekly × 4) ÷ monthly.

Composite is 5/6. signals.gitdealflow.com/predicted?utm_source=social&utm_medium=mascot&utm_campaign=batch-2026-05

linkedin

**Signal of the week:** ShadcnUI's commit-velocity ratio is at 1.62, the third consecutive week above the 1.5 acceleration threshold.

What that means in plain English: the team is shipping ~62% more in the last 14 days than their 90-day baseline. Sustained acceleration over 14+ days is the cleanest leading indicator we track.

The procedure to verify yourself (5 minutes):
1. Open github.com/shadcn-ui/ui
2. Click Insights → Pulse → 1 month. Note weekly + monthly commits.
3. Click 3 months. Note quarterly commits.
4. Compute (weekly × 4) ÷ monthly. If > 1.3, accelerating. If > 1.5, sharply accelerating.

The full 7-signal composite scores them at 5/6. The only miss: their dependents graph isn't formally exposed because they're a UI library distributed via copy-paste rather than npm. That's a known framework edge case, not a real negative signal.

Backtest: composite of 5/6 with sustained 14d acceleration → ~38% chance of a venture round closing within 47 days. Roughly 5× the base rate.

Read the live ranking: signals.gitdealflow.com/predicted?utm_source=social&utm_medium=mascot&utm_campaign=batch-2026-05

Methodology paper (CC BY 4.0): ssrn.com/abstract=6606558

instagram

Signal of the week 📊

ShadcnUI, commit velocity 1.62
3rd week running above the 1.5 threshold.

Composite: 5/6
Read it yourself: insights → pulse → 1 month
github.com/shadcn-ui/ui

Link in bio for the live ranking.

#GitDealFlow #DataNerd #CodeSideSourcing #VentureCapital #AngelInvesting #GitHubSignals #StartupSourcing #AlternativeData

facebook

Signal of the week: ShadcnUI is showing a 14d/90d commit-velocity ratio of 1.62, the third consecutive week above the 1.5 acceleration threshold.

The framework backtest: composite of 5/6 with sustained 14-day acceleration → ~38% chance of a venture round closing within 47 days.

You can verify the velocity number yourself in 5 minutes using only github.com (no API, no tool). The procedure is in the free 30-day course at signals.gitdealflow.com/challenge?utm_source=social&utm_medium=mascot&utm_campaign=batch-2026-05.

Read the live ranking: signals.gitdealflow.com/predicted

threads

Signal of the week: ShadcnUI 14d/90d commit-velocity = 1.62. Third week above 1.5.

Public-data only. github.com/shadcn-ui/ui → Insights → Pulse.

Composite: 5/6. signals.gitdealflow.com/predicted

bluesky

Signal of the week: ShadcnUI 14d/90d = 1.62. Third week above 1.5.

Verify on Insights → Pulse.

Composite 5/6. signals.gitdealflow.com/predicted

instagram

How to read a dependents graph in 60 seconds 🔍

Most investors don't know GitHub exposes this page.

→ Open the org's flagship repo
→ Insights → Dependency graph → Dependents
→ Count external dependents (not the org's own repos)

The cheapest external-adoption proxy that exists. A few hundred external dependents on a dev-tools startup = strong PMF signal regardless of revenue.

Edge case: enterprise/private package registries don't show. That's a sign of paid distribution, not weakness.

Save this. Run it on the next dev-tools startup that pitches you.

#GitDealFlow #DataNerd #CodeSideSourcing #VentureCapital #AngelInvesting #GitHubSignals #StartupSourcing #AlternativeData

twitter

How to read a dependents graph in 60 seconds, most investors don't know GitHub exposes this page.

→ Repo home → Insights → Dependency graph → Dependents
→ Count external dependents (excl. the org's own repos)

100+ external = strong PMF signal regardless of revenue.

linkedin

**How to read a dependents graph in 60 seconds.**

Most investors I talk to don't know GitHub exposes this page. It's tucked under Insights → Dependency graph → Dependents and it shows you every public repo that depends on this code.

It's the cheapest external-adoption proxy that exists.

The procedure:
1. Open the org's flagship repo (the most-starred one, or the one in the README).
2. Click Insights → Dependency graph → Dependents.
3. Count the external dependents, repos that aren't part of the same org.
4. For a dev-tools startup, 50+ external dependents is meaningful. 200+ is strong.

What the framework filters for: external usage that the founder doesn't have to talk about for the data to exist. Strong PMF signal regardless of revenue.

The edge case worth knowing: enterprise teams using private npm or pip registries won't show up. The dependents page being empty isn't a no-signal. It can be a "private distribution" signal, and that's a different kind of moat.

The full procedure for all seven signals is in the free 30-day course: signals.gitdealflow.com/challenge?utm_source=social&utm_medium=mascot&utm_campaign=batch-2026-05

facebook

Framework explainer: how to read a dependents graph in 60 seconds.

Most investors don't know GitHub exposes this page. Insights → Dependency graph → Dependents shows every public repo that depends on this code.

It's the cheapest external-adoption proxy that exists. 100+ external dependents on a dev-tools startup = strong product-market-fit signal regardless of revenue.

The full 7-signal procedure (free, no card): signals.gitdealflow.com/challenge?utm_source=social&utm_medium=mascot&utm_campaign=batch-2026-05

twitter

Data point from the panel:

Startups that closed: contributor-diversity Gini = 0.34 at month -3.
Startups that did not close: 0.61.

The shape of the codebase matters more than the size of it. SSRN id 6606558.

linkedin

**Data point from the 219-round panel.**

Startups that closed a venture round: contributor-diversity Gini coefficient of 0.34 at month -3 before the announce.

Startups that did not close: 0.61.

Translation: more concentrated codebases (one or two committers doing everything) close fewer rounds than distributed codebases.

The mechanism is straightforward. A 4-person codebase with no single committer dominating is funding a real engineering team. A 1-person codebase is funding a salary. The contract value, the dilution math, and the diligence story are all different, even if the headline metrics look similar.

This is signal #2 of the seven we publish. The procedure to read it yourself runs in 5 minutes from any GitHub org's Insights → Contributors page.

Methodology paper: ssrn.com/abstract=6606558 (CC BY 4.0)
Free 30-day course: signals.gitdealflow.com/challenge?utm_source=social&utm_medium=mascot&utm_campaign=batch-2026-05

instagram

📈 Data point from the panel:

Closed rounds → Gini 0.34
Didn't close → Gini 0.61

The shape of the codebase matters more than the size of it.

(SSRN id 6606558)

#GitDealFlow #DataNerd #CodeSideSourcing #VentureCapital #AngelInvesting #GitHubSignals #StartupSourcing #AlternativeData

threads

Data point from the panel:

Closed: Gini 0.34
Didn't close: Gini 0.61

Shape > size. SSRN 6606558

bluesky

Closed rounds: Gini 0.34 at month -3.
Didn't close: 0.61.

Shape of the codebase > size of it.

twitter

Try this on one startup you almost-invested in:

1. Gut prediction first, score 0-6 from memory.
2. Then run the 5-min composite.
3. Reply with the delta.

Most useful artifact you'll produce this week. Procedure: signals.gitdealflow.com/challenge?utm_source=social&utm_medium=mascot&utm_campaign=batch-2026-05

linkedin

**One operator prompt for your weekend.**

Pick a startup you almost-invested in last quarter. Doesn't have to be high-stakes, just one where you remember saying "I'll think about it" and then didn't.

Step 1: write down a gut prediction. Score 0-6 from memory, no looking. What's your guess?

Step 2: run the 5-minute composite on their GitHub org. Insights → Pulse for commit velocity, Insights → Contributors for diversity, Insights → Dependency graph for external adoption, README history for freshness, Repositories → Newest for platform buildout, Issues + PRs for ratio.

Step 3: compute the delta between your gut and the composite.

The artifact you produce, a one-line note with org + gut + composite + delta, is the most useful sourcing-process exercise you'll do this week. It calibrates your gut against the framework, in both directions. Sometimes the framework is wrong; sometimes your gut was. Both are useful.

If you reply with the delta (DM is fine), we're collecting cases for the next iteration of the curriculum.

Procedure: signals.gitdealflow.com/challenge?utm_source=social&utm_medium=mascot&utm_campaign=batch-2026-05

threads

Try this:

1. Pick a startup you almost-invested in.
2. Gut prediction (0-6) first.
3. Run the 5-min composite.
4. Note the delta.

Most useful artifact you'll produce this week.

bluesky

One weekend prompt:

1. Pick a startup you almost-invested in.
2. Gut prediction first (0-6).
3. Run 5-min composite.
4. Note the delta.

linkedin

**Calibration case, a recently-funded round, scored at month -3.**

Vercel announced their Series E in 2026-04. We backtested their composite as it would have read 90 days before the announce.

Score at month -3: **5/6.**

→ Commit velocity ratio: 1.41 (above 1.3 threshold) ✅
→ Contributor diversity: top contributor 38% of commits, 7+ active contributors ✅
→ Dependents graph: 280+ external dependents on Next.js alone ✅
→ README freshness: substantive diff at -42 days ✅
→ New repo creation: 4 platform repos in last 30 days ✅
→ Issue-to-PR ratio: 0.9, below the 1.5 threshold (the only miss)

The miss on issue-to-PR is interesting because it suggests the team was fielding inbound faster than they could ship, the kind of "we need the round to hire" moment that often precedes a growth-stage close.

Composite framework would have flagged this round before the headline. ~5× lift over the base rate at month -3.

Calibration runs against known rounds are how you build trust in the framework. Run one a week and the score becomes more useful than the headline.

Methodology: ssrn.com/abstract=6606558
Free 30-day course: signals.gitdealflow.com/challenge?utm_source=social&utm_medium=mascot&utm_campaign=batch-2026-05

twitter

Calibration case: Vercel Series E 2026-04.

Composite at month -3: 5/6.
Only miss: issue-to-PR (0.9 vs 1.5 threshold).

Framework would have flagged this round before the headline. ~5× base rate.

instagram

🎯 Calibration case

Vercel Series E (April 2026)
Backtested at month -3: 5/6

Only miss: issue-to-PR ratio (suggested inbound > shipping speed)

Framework would have flagged this round 90 days early.

#GitDealFlow #DataNerd #CodeSideSourcing #VentureCapital #AngelInvesting #GitHubSignals #StartupSourcing #AlternativeData

facebook

Calibration case, Vercel's Series E announced April 2026, backtested at month -3.

The composite read 5/6 ninety days before the announce. The only miss was the issue-to-PR ratio (0.9 vs the 1.5 healthy threshold) which suggested inbound was outpacing shipping, the kind of "we need the round to hire" moment that often precedes a growth-stage close.

Calibration runs are how you build trust in the framework. Run one a week. SSRN: ssrn.com/abstract=6606558