- The exchange health score (0–100) summarises how sound an exchange is AS AN ARBITRAGE VENUE — liquidity, listing breadth, withdrawal availability, spread stability and funding sanity — not its trustworthiness or solvency.
- It is the average of the AVAILABLE sub-scores. A signal we cannot measure (e.g. withdrawal networks without an API key) is shown as “not measured” and excluded — never silently counted as good.
- It is built only from data this service already computes. It is a heuristic, not financial advice or a safety guarantee.
Each exchange dashboard shows a health score from 0 to 100. It answers one question: how sound is this exchange as a place to run arbitrage right now? It is deliberately about the venue, not about a single trade — per-trade danger is the risk score, and profitability lives in the arbitrage cards themselves.
What does the score look at?
Five sub-scores, each 0–100. The overall score is the average of the ones we can actually measure for that exchange.
| Signal | What it measures |
|---|---|
| Liquidity | Aggregate 24h spot volume — how easily you can fill size on this venue. |
| Breadth | How many spot coins are listed — more pairs = more potential spreads. |
| Networks | Share of listed coins you can actually withdraw on at least one chain. Only measurable where we have network data (public, or via a platform API key). |
| Spread stability | Inverse of the stuck-spread rate: large spreads involving this exchange that never converged in 24h. A high stuck rate flags frozen transfers / delistings. |
| Funding sanity | Inverse of the share of perps with extreme funding — a crowded, stressed derivatives market scores lower. |
Why is a signal sometimes “not measured”?
Network and withdrawal data is only available for exchanges with public currency endpoints or a connected platform API key. Without it, the withdrawal signal is marked “not measured” and left out of the average — absence of data is not evidence of health, the same principle as the risk score. The dashboard also shows whether the data came from a public source or an API key.
How are liquidity and breadth scaled?
Against fixed reference points (not relative to the other exchanges), so a venue’s score doesn’t move just because a different exchange changed. Liquidity is scaled on a log of 24h volume; breadth on the number of listed spot coins. The funding component reuses the same annualised-rate math as the funding carry mode.
The health score is a comparison heuristic across exchanges, built from market data — not a statement about an exchange’s custody, solvency or trustworthiness. Always check withdrawal availability and the order books before sizing a trade.