The git-card rank
F to SSS.
Every profile gets a rank from F to SSS: an estimate of where it sits among active GitHub developers, built from seven public numbers. S and above are the super elite, the top 1%.
Tiers
Nine tiers, by percentile.
- SSSLegendaryTop 0.01%0.01% of developers
- SSMythicTop 0.1%0.09% of developers
- SEliteTop 1%0.9% of developers
- AExpertTop 5%4% of developers
- BAdvancedTop 15%10% of developers
- CSkilledTop 35%20% of developers
- DActiveTop 60%25% of developers
- EGetting startedTop 85%25% of developers
- FDormantEveryone else15% of developers
The method
How the score is computed.
- Seven public metrics, grouped into impact, activity and collaboration. Private repositories never count.
- Each is scored on a log scale against what a typical active developer has, because GitHub numbers are skewed: a few people have a hundred thousand stars, most have a handful. The score is
(ln(value + 1) - ln(typical + 1)) / spread, so 0 is typical and every spread above is a big step up. - The scores are averaged by weight, then rescaled for how much the metrics move together. Strong developers tend to be strong at several things at once, so the model assumes a correlation of 0.5 between any two metrics. Without this, being good at everything would look far rarer than it is.
- One exceptional category gets a bonus. If impact, activity or collaboration is more than 3 spreads above typical, every spread beyond that adds 0.15. A legend in one dimension is not dragged down for being quiet in another.
- The result is read off a normal curve as an estimated top percentage, and that picks the tier.
The numbers
Every weight, in the open.
| Metric | Category | Weight | Typical | Spread | Top 1% alone | Window |
|---|---|---|---|---|---|---|
| Stars earned | Impact | 24% | 5 | 2.2 | 1k | all time |
| Followers | Impact | 12% | 10 | 1.7 | 573 | all time |
| Contributions | Activity | 16% | 150 | 1.3 | 3.1k | last 12 months |
| Active days | Activity | 10% | 60 | 0.7 | 310 days | last 12 months |
| Pull requests | Collaboration | 16% | 20 | 1.6 | 867 | all time |
| Code reviews | Collaboration | 12% | 3 | 1.8 | 262 | last 12 months |
| Orgs contributed to | Collaboration | 10% | 1 | 1 | 19 | last 12 months |
"Top 1% alone" is what one metric needs on its own to sit in the top 1% for that metric. The rank itself needs a strong combination.
Fine print
It is an estimate.
- The typical values and spreads are informed estimates of the active GitHub population, not fitted to a sampled dataset. Treat the percentage as a well-reasoned guess.
- Only public data counts. Contributions to private repositories show up only as anonymous calendar counts, and only if the person chose to show them on their profile.
- Activity metrics cover the last 12 months, so the rank moves as people ship, or stop shipping.
- The model is open source. Read src/lib/rank.ts, and if you think a weight is wrong, open a PR with your reasoning.