RepoWorth by YC batch
Compare totals, medians, velocity, concentration, and technical health at the same number of days after cohort start or Demo Day.
Methodology and data release in reviewCodeValuation.com Research
Software evidence worth citing.
Understand how code value, technical health, contributors, and momentum change across repositories, cohorts, ecosystems, and places. Every published result is tied to versioned measurements, not a generated claim.
Index program
Compare like with like.
Public indexes enter publication only after membership, deduplication, exclusions, model consistency, minimum samples, and coverage pass review.
Cities and countries creating value
Keep company headquarters and contributor-declared locations separate, publish coverage, and require safe aggregation thresholds.
Methodology and data release in reviewWhere open source momentum is building
Measure RepoWorth, active contributors, velocity, health, and release activity by language without double-counting mixed repositories.
Methodology and data release in reviewValuable, new, and overlooked projects
Surface qualifying public repositories through reproducible age, value, growth, quality, and funding evidence instead of editorial guesswork.
Methodology and data release in reviewPublication standard
A number is not research until someone else can test it.
CodeValuation.com separates repository movement, formula movement, coverage changes, and cohort membership changes. Historical snapshots remain immutable when a model improves.
The canonical metric name, unit, aggregation, population, exclusions, and methodology version.
The exact point in time through which repository and GitHub evidence was collected.
The RepoWorth formula and price version used for the calculation.
How much qualifying evidence was available and which limitations remain.
The sourced, dated, and reviewable rule that placed an entity in an index or cohort.
A frozen query definition, supporting rows, source links, and required CodeValuation.com attribution.
For analysts and journalists
Ask a public question. Inspect the rows behind the answer.
The public research interface will be rate limited and restricted to approved public views. It will never query private repositories, private portfolios, raw emails, precise personal locations, unpublished scans, or private security details.
Research access is opening in stages.
We are validating dataset quality and publication guardrails before enabling self-serve questions. Tell us the cohort, ecosystem, geography, or repository question you want to investigate.
Discuss a research use casePrivate portfolio benchmarking
Compare private evidence without making it public.
Approved monitoring workspaces can align a private portfolio with a qualified public cohort by company age, investment date, cohort start, cohort event, or current date. The workspace can inspect totals, medians, velocity, technical health, concentration, sample size, and coverage while company and repository identities stay private.
This capability is in controlled qualification. It activates only after portfolio membership, comparison definitions, source quality, and the workspace privacy boundary pass review.
Questions
What makes the research usable.
The result should carry enough definition, coverage, and provenance for another reader to test it.
What must be published with a research result?
Each released result includes its metric definition, source cutoff, model version, coverage, sample size, membership rules, exclusions, warnings, supporting rows, and reproducible query identifier.
Can results from different model versions be compared directly?
Not without adjustment. Research separates repository movement from formula movement, coverage changes, and cohort membership changes. Frozen historical results keep their original model identity so a reader can see when the basis changed.
Does public research include private repositories or portfolios?
No. Public datasets do not query private repositories, private company rows, unpublished scans, raw emails, or private security details. Approved workspaces may compare private aggregates with a qualified public cohort without contributing those private rows to the public index.
Why might an index or comparison be withheld?
A result is not published when membership, deduplication, source coverage, sample size, model consistency, privacy thresholds, or reproducibility requirements do not meet the documented publication standard.
How should published research be attributed?
Use CodeValuation.com as the source and identify the dataset or index, data-through date, and RepoWorth model version shown with the result. The release includes the required attribution format.