ROSS Index Methodology

Our index detects the top trending open-source startups via open data analysis. Here, we explain how exactly it works.

The Runa Open Source Startup (ROSS) Index collects data on all repos with 1,000+ GitHub stars. We publish the top trending open-source startups that are affiliated with these repos on a quarterly or annual basis.  The purpose of this index is to highlight new, trending open-source startups, using a simple and transparent base. 

Why do we only use GitHub star growth?

We chose only one basic metric as, otherwise, it would be hard to verify the index results externally. Any composite and complex derivative metrics overcomplicate the picture and are better consumed internally. 

We chose “stars” because it allows for a reasonable comparison between various repos using the same scale while showing trending open-source products. There is no perfect business metric for open-source software — all have their flaws. For instance, “contributors” are good for comparing “apples to apples” (e.g. databases), but can barely be used for different categories of software. Stars seem the best single metric for index purposes.

We measure the relative growth of stars (and not absolute) because it shifts focus from established open-source champions to interesting newcomers in the developer community. GitHub has network effects, so it is easier to acquire stars for already well-starred repos. But this index is about startups!

Calculation approach

For the quarterly index, we sort repositories by their max relative growth of stars observed in any periods adhering to simple rules:

  1. The period length — 90 days
  2. The period ends strictly within the target quarter of the ROSS index
  3. The repo had 1,000 or more stars at the beginning of the period 

Relative growth = Stars at the end / Stars at the beginning of the period.

Finally, our investment team only selects the repos belonging to startups and publishes the top-20 in the ranking. If a startup has several top repos, we will only mention the repo with the best growth rate.

Startup definition

For purposes of the index, we define a “startup” as a product-focused commercial organization that

  1. was founded fewer than ten years ago
  2. raised less than $100M in total known funding
  3. has a product reasonably connected to its open-source repos
  4. was not acquired or went public before the end of the target period

This definition does not cover, for instance, side projects. For example, when a founder has another full-time job. Nor does it include projects without any associated commercial product-focused entity or those developed by large tech corporations. The definition also excludes businesses that are mostly focused on the provision of professional services.

Open source definition

As VC investors, we stick to the commercial perception of “open-source”. So we would consider e.g. Elastic and MongoDB as commercial open-source companies, although their Server Side Public Licenses are not approved by the OSI. The same applies to many participants of the ROSS Index.

Data sources

The index publications are based only on public data from GitHub, Crunchbase, mass media, and other sources. We strive to show cities and countries connected with the origin of companies, not just formal HQ locations. We are purely focused on companies using English descriptions, so some local startups (e.g. Chinese) could be missed.

Given the open-source nature of data, some inaccuracies are possible. In any case, Runa Capital does not use any private information received from startups for the index. For instance, we may know that a company raised a new VC round but we will only publish publicly available numbers like those on Crunchbase.

Old calculation approach

In 2020-2023, we have been calculating the relative growth of stars within the fixed period (quarter or year) for all repos with at least 1,000 stars at the start of the period. For that rankings, we used the annualized growth rate which was calculated as AGR = (value now / value 1 quarter ago)⁴ – 1.

Since Q1 2024, we slightly updated the approach with sliding “90-day windows” to detect fast-growing repos even if their growth spike happened between calendar period.