Ratios Describe. Frameworks Evaluate.

The short version: ccrank

ccrank is a token leaderboard that tracks Claude, Codex, Cursor, Kimi, Grok, GLM, Pi, and OpenCode — powered by ccusage, with per-user profiles, history, and analytics. What makes it notable: it does not stop at volume. The board can be re-ranked by Tokens, Cost, Output/$, Cache Rate, or Output Ratio — which means somebody else independently arrived at the idea that raw volume alone is insufficient. That is the correct instinct, and ccrank deserves credit for shipping it.

SigRank starts from the same premise and formalizes it. Instead of separate ratio views, it defines one composite efficiency score — Υ = cache_read × output / input² — plus the surrounding framework: Leverage, SNR, Velocity, behavioral archetypes, class tiers, and ed25519-signed telemetry. ccrank exposes the dials; SigRank defines what the dials mean.

Feature comparison

Why ccrank is the interesting one

Most token boards converge on the same formula: collect local telemetry, rank volume or cost, add profiles and time filters. ccrank adds something different — derived ratios that treat the telemetry as a signal source rather than a scoreboard. Ranking by Output/$ instead of raw spend is the difference between asking "who spent the most" and asking "who got the most out." That is the question the whole category should be asking.

Where the approaches diverge is integration. ccrank's ratios are independent views — you can sort by Cache Rate, but Cache Rate alone conflates hoarding with reuse. SigRank's Υ Yield is a single defined formula where the four token pillars compose: cache_read and output amplify, input² penalizes waste quadratically. Leverage, SNR, and Velocity describe the cascade's shape around that number. An Output/$ leaderboard tells you who is cheap; a Υ-ranked leaderboard tells you whose cascade is compounding.

From ratios to a framework

If ccrank already shows you Cache Rate and Output/$, you have the raw ingredients. SigRank composes them into a score you can sign and publish:

npm install -g sigrank
sigrank enroll      # create your operator identity
sigrank submit      # reads logs, scores, signs, publishes

Prefer to inspect before you submit? Run sigrank me --dry-run to see your scored payload locally, or paste your token counts into the /score calculator to compute your Υ Yield, class tier, and compression ratio instantly — no account, no submission, just the numbers.

Frequently asked questions

Is ccrank a SigRank alternative?

ccrank is the closest thing to one in the category — and that is worth saying plainly. Most token leaderboards rank raw volume only. ccrank independently reached the conclusion that volume alone is insufficient and exposes derived ratios: Output/$, Cache Rate, and Output Ratio, switchable across time windows and tools. SigRank goes one step further: instead of separate descriptive ratios, it defines a single composite efficiency score (Υ Yield) plus a full metric stack — Leverage, SNR, Velocity, archetypes, and class tiers — with cryptographically signed telemetry. If you are comparing the two, you are already past the 'tokens = skill' stage, and both tools are on the right side of that line.

What does ccrank measure that other leaderboards do not?

Efficiency-adjacent ratios. Where most boards rank total tokens or total spend, ccrank lets you re-rank the same field by Output per dollar, Cache Rate, and Output Ratio. That is a real conceptual step — it treats token telemetry as something to derive signals from, not just sum. SigRank builds on the same insight but formalizes it: Υ Yield is a defined formula (cache_read × output / input²), not a view toggle, and it sits inside a governed evaluation framework with operator archetypes and class tiers.

What is the difference between a ratio and a framework?

A ratio answers one question on one axis: Output/$ tells you cost efficiency, Cache Rate tells you reuse. A framework defines how the pieces compose into a judgment. SigRank's Υ Yield combines cache reuse and output into a single efficiency measure; Leverage, SNR, and Velocity describe the shape of the cascade around it; archetypes classify the operating pattern; class tiers calibrate it against scale. ccrank gives you several good instruments. SigRank gives you the instrument panel plus the definition of what 'good' means.

Can I use both ccrank and SigRank?

Yes — they read the same local telemetry. ccrank's ratio views are a great quick read on cost efficiency and cache behavior. SigRank adds the composite score, the signed submission, the class tier, and the cross-window ranking. If you are on ccrank you already believe efficiency matters; SigRank is where that belief becomes a number you can defend.

Which is better for proving operator skill?

For a quick ratio read, ccrank's Output/$ and Cache Rate views are genuinely useful. For a defensible claim — a single defined score, signed telemetry, archetype classification, and cohort-relative ranking across time windows — that is what SigRank was built for. Ratios describe; frameworks evaluate.