How to read every page, where the numbers come from, and what they mean for you.
Written by 0xLazard · last updated
HyperSonnar reads public Hyperliquid data — prices, order books, positions, trades, funding — and official sources for macro and company data (FRED, BLS, SEC). No private access is needed to analyze a wallet or a market.
Copy trading is the home page: the wallet ranking. Two tabs: Top copyable (sorted by copiability score, or by 1D, 1W, 1M performance) and Today's winners (the day's best perp performers).
Each card shows the score, the trader's style, win rate, trade count, perp equity, the 1D / 1W / 1M windows of the Hyperliquid perp portfolio, the 7d perp ROI, the median hold and all-time perp PnL. These numbers are recomputed on every analysis pass, about once a day, ranked wallets first. A "—" doesn't mean zero: this wallet hasn't been analyzed yet.
The 7d and 30d perp ROI is computed by HyperSonnar on the PERP side of the account only — the only side a copier replicates — not on the whole portfolio (perp, spot, vaults) as the Hyperliquid leaderboard does: the two numbers differ, on purpose. It is a time-weighted return: between two samples, the gain or loss is measured against the capital actually at work, so a deposit or a withdrawal counts neither as a gain nor as a loss. But Hyperliquid's samples are hours apart, and when an account keeps depositing and withdrawing, the exact timing of the flows is unknown: we then compute the bracket the return must lie in, and if it is too wide — over 5 points on 7 days, 10 on 30 — we show "±" rather than a number that does not exist. There is no all-time ROI: over years that uncertainty becomes huge for everyone; the all-time PnL in dollars, however, is exact and stays on the curve.
The product is geometric: one weak component drags the whole score down, none can be compensated by the others, and a wallet with a component that can't be computed isn't ranked. Quality and consistency are shrunk by the Wilson interval — put simply: a win rate is only believed as far as the trade count proves it. 90% wins over 10 trades weighs far less than 90% over 1,000. It's also why nobody scores 100: a short history cannot reach a high score, whatever its win rate.
Before any computation, a wallet must clear every one of these gates: positive all-time perp PnL; at least 5 trades on at least 3 coins; a trade in the last 30 days; a Hyperliquid leaderboard account value between $1,000 and $10M; at least $100 of perp equity; a median hold (median time a position stays open) between 30 minutes and 14 days. Under 30 minutes: scalping, too fast to follow. Beyond 14 days: positions opened and forgotten, with no entry or exit to copy within a reasonable window.
The score is a heuristic, not a guaranteed prediction: it summarizes public history — it promises nothing about future trades.
At the top: the address and the trader's style — Net-loser if all-time PnL is negative, otherwise Position trader (median hold ≥ 7 days), Swing trader (≥ 24h) or Intraday. Then six indicators: all-time perp PnL, PnL / net deposits, win rate (wins · losses), median hold, perp account value and unrealized PnL of open positions. "PnL / net deposits" stays a dash when the transfer history is partial or net deposits aren't positive: it's the number that exposes a fake winner, and it never lies.
Market brings the whole Hyperliquid market onto one page (the Analytics page was merged into it on 23 September 2026). At the top, on one line: whole-market total open interest and 24h volume, then the biggest mover (the biggest loser on “Losers”). Then a table of every market: price, change over the chosen interval, open interest, 1h and 24h Δ OI, hourly funding, 24h volume and trend. The table refreshes every 15 s, every numeric column sorts, and clicking a row opens the coin page.
Trade puts a market's chart, order book and ticket on one screen. The ticker at the top left (or the "/" key) opens the list of every market, by category — Crypto, Stocks, Indices, Commodities, FX, ETF, Other — with search and sorting by volume, 24h change, funding or name. The market bar shows the live mid price, 24h volume, open interest, hourly (and annualized) funding and the next funding.
Every market on the Market page opens its own page; the "Browse all Hyperliquid coins" index, under the Market table, lists them all, grouped by dex. A coin page shows, in this order:
Three connected pages: the Economic calendar (US statistics — CPI, jobs, GDP…), Earnings (company releases and the company directory) and Company pages (accounts filed with the SEC). One shared rule: no in-house estimates — a value only appears once its source has published it, and anything computed rather than published says so ("reconstructed", "detailed filing pending").
Yes — public data read directly from Hyperliquid. Nothing is invented, and when coverage is partial, the page says so.
A dash means "unknown": not analyzed yet, source unavailable, or not computable. A zero would assert a value. The "±" on perp ROI means something else: the return exists, but the samples don't allow an honest number.
Prices, candles and order book: live. A wallet's positions: read from Hyperliquid when its page opens; a wallet read before shows its last read, with its age, while it refreshes. Copy trading ranking: one analysis pass, about once a day. Calendar and companies: when the source publishes.
Not yet for everyone: the Trade page ticket and copying place real orders on the Hyperliquid mainnet, with your money, but only for authorized accounts logged in with a wallet; for everyone else, the "Copy" button only records your interest. You can lose money, and no gain is promised. HyperSonnar is mostly for finding, scoring and studying wallets and markets.
No. HyperSonnar analyzes public history. A good copiability score describes the past — it isn't a promise about the future.
Write to us at HyperSonnarContact@proton.me — access, correction or account deletion, or any other question.