Market Pulse
“What's happening today?” answered as a structured cross-asset read — risk-on or off, dispersion, gold, dollar, VIX — with every asset class pulled in one pass and compared, not a list of tickers that are up.
Reads the market. Finds alpha. Runs your strategy 24/7. Fixes your leaks.
Five read layers run against the whole Hyperliquid board at once — price, funding, open interest, who is winning right now and where the proven money is leaning — across crypto, tokenized equities, commodities, indices and FX. Not a price screen. A read.
“What's happening today?” answered as a structured cross-asset read — risk-on or off, dispersion, gold, dollar, VIX — with every asset class pulled in one pass and compared, not a list of tickers that are up.
One sweep of the entire universe — funding, open interest, the proven cohort's positioning against the 4h crowd, momentum events and cross-asset flows — ranked through two lenses: a trade lens for actionable edge and a news lens for what's surprising.
Where the most profitable wallets on Hyperliquid are positioned, where they diverge from the crowd, and the near-term flow — the divergence is the signal.
Rank the best track records and vet any trader before you mirror them — historical PnL, win rate, drawdown, gain-to-pain, consistency and risk labels, and what's driving their streak right now.
What that trader has actually been doing, read as a strategy and critiqued. A quant score across six explained dimensions. The regime they're trading in and how they behave in each. Their live book with a protection audit. Every leak priced as counterfactual dollars. Their book set against the proven cohort — top traders by all-time realized P&L, ≥ $1M — and the hot 30-day cohort: side, headcount, when they moved, what they hold that this trader doesn't. Works on your own wallet or anyone else's.
Start from one of 114 templates or from a sentence. Either way it becomes your strategy — forked under your name, walked through lever by lever, and proven to run before a cent moves. There is no paper-trading mode: senpi validate plus a $10-floor live run is the test, in real market conditions.
An analyst-style picker: “what should I trade”, “I have a view on the world — trade it”, “run a hedge-fund book”. 19 starter templates for a first deployment, 95 advanced. Every template deploys as your own fork — purplefrog-starling, or a name you choose. Open a category to see every strategy in it. Every template's source is in Senpi Skills on GitHub.
Describe the idea; your quant asks one decision at a time and writes the package — a strategy.yaml, a runtime.yaml per instance, and a Python scan(inputs, ctx). Editing a live strategy — leverage, sizing, the stop ladder — is the same conversation. Writing from scratch is a peer route, never a downsell.
A strategy declares the gates a candidate must clear — trend, volume, open-interest confirmation, smart-money alignment, session windows — and refuses to fire on one signal alone. An optional model-in-the-loop can still block the final execution.
Declared in YAML, evaluated live before every open, and fail-closed: if the risk state can't be read, nothing opens. Halts latch and survive restarts and updates.
| Gate | Trips when | Result |
|---|---|---|
| Daily loss halt | Today's loss breaches a $ or % limit | closed |
| Drawdown halt | PnL falls a set % from peak — immune to deposits | closed |
| Consecutive-loss cooldown | Last N trades all lost | cooldown |
| Per-asset cooldown | That asset was closed within the window | cooldown |
| Max entries / day | Cap reached — optional bypass if the day is profitable | closed |
Strategy code only ever returns a signal. The runtime owns everything after that — validation, sizing, the guard rails, the order, and the exit — on your own wallet, in its own isolated sub-wallet per strategy, every trade public and verifiable onchain.
Supervises every strategy package in-process: calls scan() on its interval, validates each signal, sizes the order, enforces the rails, places it, then hands the position to the stop engine. Zero model cost per tick — monitoring is the runtime's job, not a chat turn.
Per action and per exit, choose between a market order and a post-only maker order that rests for the fill and pays the lower fee — with a taker fallback if it doesn't fill inside the window, so urgency is never traded away for a rebate.
“Long HYPE 10×”, “close my ETH”, “copy this wallet” — one decision at a time, with protection optional: bare, a static stop, or a profit-locking ladder. Mirror a specific trader with a multiplier and a budget; sizing checks available funds before the call, never after.
Every strategy runs on its own Hyperliquid sub-wallet with its own margin. One can't liquidate another, budgets stay honest, and closing a strategy returns exactly what's left to your funding wallet.
Money enters through the funding card — buy USDC with a card or deposit from Base, Arbitrum, Ethereum, BNB Chain, Polygon and more — and leaves only by your own withdrawal. Two hard rails: only you move your funds, and only with a fresh yes.
Every trade, top-up, withdrawal, deploy or stop change is proposed, then placed only when you answer in a new message — even when the order is exactly what you asked for. Helps first. Acts with permission.
Senpi watches, you don't. Every position a strategy opens is handed to a two-phase stop engine that defends first and then locks profit — the floor is a resting stop on the exchange, so it fires even if nothing else is awake.
| Phase | Mechanism | Field |
|---|---|---|
| 1 · Defend | Absolute loss floor from entry plus a trailing retrace; a breach must repeat across consecutive ticks before it closes, so one wick can't take you out | max_loss_pctretrace_thresholdconsecutive_breaches |
| 2 · Lock | Arms at the first profit tier. Each tier locks a percentage of high-water ROE as a price floor; the floor ratchets and never loosens. The floor lives as a resting exchange stop | tiers[].trigger_pcttiers[].lock_hw_pct |
| Time cuts | Optional, either phase: hard timeout; cut a position that peaked weakly; cut dead weight that never moved | hard_timeoutweak_peak_cutdead_weight_cut |
Profit locked, stop raised, a daily-loss or drawdown halt tripped, a decision that failed, a scanner that went quiet. Each event notifies once, in the app and by push. Position open and close are announced too. No noise.
Every strategy your agent runs is watching its positions overnight with the stop engine, the ratchet tiers, and stops resting onchain. Ask “are my positions protected?” and get the answer per position, from a live read — never from memory.
Every decision the runtime makes is on the record — which gate said what, why a signal fired or didn't, how each exit ended. Your quant reads that record back to you as coaching, with every leak priced in dollars.
“Did I sell too early?” “What did I miss this week?” “Am I getting shaken out?” A retrospective against the market and against the best wallets — exits versus hold, what your own limits blocked, maker versus taker fees, and the single lever that would have paid the most. Counterfactuals never sum; the desk names the one that matters.
A chronological record per strategy: every signal, every gate verdict with its reason and data source, every order, every exit with a categorized reason — stop hit, adverse close, profit banked, time cut, closed by you. Searchable, and read by the coaching above.
Strategies, positions and balances across every wallet — main and sub-wallets, deployed and idle — read live. Each strategy is judged against its own mandate: is it doing the job it was built for? Positions are the evidence, not the headline.
Your Senpi points and rank, loyalty tier and the fee it earns you, referral rewards, and the AI-credits meter — what a strategy costs to run, why credits move, and what's free. Strategies run on the runtime at zero model cost; only conversations spend credits.
Connect the agent you already use — Claude, Claude Code, ChatGPT, Codex, Grok, Muse, or any app that speaks MCP — and it runs the whole Senpi harness. Your agent asks; Senpi reads the market, builds and runs the strategy, and protects every position — on your wallet, under your rules. Nothing on this page is held back from a connected app.
Your app gets a typed MCP connection tied to your Senpi login. No API keys to paste, nothing to install on your side.
“What's moving?”, “run the quant desk on this wallet”, “deploy Hawk with $200”. Your agent calls Senpi; Senpi does the work and answers with the real result.
Only you move your funds, and money-moving tools still need your explicit approval. Revoke any app in one click from Connectors.
Your agent is yours — its own workspace, its own memory of how you trade, its own conversation history. Talk to it wherever you are.
The same agent in the Senpi app on web and mobile: onboarding, funding, your first small trade, and every alert. Or skip the app entirely and drive Senpi from Claude, ChatGPT, Gemini or Cursor with a Connector.
The in-house model tuned for this job — behind every market read, every strategy your quant writes, and every coaching answer. Not a general chat model wearing a trading skin.
Every Senpi trade lands on Hyperliquid under a wallet you can inspect. Nothing is simulated, nothing is hidden, and the only person who can move your funds is you.
Everything above runs on the Senpi MCP — one typed interface your agent, the runtime and your own tooling all speak. Open, documented, and the same surface whether a human asked or a strategy did.
Create a new copy-trading strategy against a specific trader.
Create a custom strategy with specific assets, leverage, and risk parameters.
List all strategies owned by the current authenticated user.
Fetch full details for a specific strategy: config, status, wallet balance.
Update parameters of a running strategy without closing positions.
Pause a strategy — stops new trades but holds existing positions.
Close a strategy, liquidate all positions, return capital to the user.
Close all open positions without terminating the strategy itself.
Deposit additional capital into a running strategy wallet.
Withdraw available balance from a strategy wallet.
Bridge funds from Hyperliquid back to an EVM-compatible address.
Per-asset position size limits and trading constraints for a strategy.
Full on-chain margin state: equity, leverage, margin used, position health.
All pending limit and stop orders currently on the book.
Historical PnL curve and account value over a specified range.
Attach a ratcheting trailing stop with tier thresholds and lock percentages.
Modify tier levels or lock percentages on an existing ratchet stop.
Remove a ratchet stop from a position entirely.
Current ratchet state: active tier, locked PnL level, high-water mark.
All active ratchet stops across every position in a strategy.
Stream of ratchet trigger events: tier upgrades, activations, stop-outs.
Current mark price and index price for one or more assets.
All tradeable instruments on the main and XYZ DEX (equities, metals, indices) with specs and live context.
Asset snapshot: OI, 24h volume, funding rate, liquidity depth.
Historical 8-hour funding rate series for an asset.
Classify current funding regime: extreme positive, neutral, extreme negative.
Cross-asset capital flow signals — which assets are gaining or losing attention.
Top traders in the current 4-hour rolling momentum window.
Smart Money market-concentration scores across all Hyperliquid assets.
Large position opens and closes by top traders, streamed in real time.
Full leaderboard profile for a specific trader address.
All open positions currently held by a specific leaderboard trader.
Current leaderboard snapshot: rank table, scores, asset distribution.
Top-performing traders by PnL, ROE, or win rate across multiple timeframes.
Strategy leaderboard ranked by totalPnl across all users.
Full historical trade and position record for a specific trader address.
Snapshot of a trader's current open positions and recent activity.
Realized PnL on a position that is still open — partial-close accounting.
Current authenticated user profile, linked wallets, account settings.
Pending and previously claimed referral rewards for the current user.
Claim accumulated referral rewards to the connected wallet.
Current Senpi points balance, active multipliers, and tier status.
Points leaderboard: top users ranked by accumulated Senpi points.
Open a new perpetual position at market or limit price.
Modify size, leverage, take-profit, or stop-loss on an existing position.
Fully close a specific open position at market or limit price.
Cancel a pending limit or stop order by order ID.
Estimate fees, slippage, and required margin before committing.
Full details of a live position: entry price, size, unrealized PnL, margin.
Post-close breakdown: realized PnL, hold duration, exit price, fees.
Order status: filled, partially filled, pending, or cancelled.
Consolidated view: all strategies, balances, open positions, unrealized PnL.
Historical account-level performance stats and cumulative metrics.
Dry-run estimation: cost, margin impact, and slippage before opening.
List all available Senpi guide resources: URI, name, description.
Fetch a Senpi guide by URI — parameter definitions, methodology, workflow examples.
Loyalty tier definitions: point thresholds, fee discounts, tier benefits.
Current Arena standings: rank, ROE, prize tier qualification.
Total prize pool size for the current active Arena week.
Prize breakdown by rank tier for the current Arena competition.
ROE time-series chart data for a trader's Arena performance.
Move spot balance into the Hyperliquid perpetuals margin account.
The path is improve → protect → automate. Start by asking what's moving; end with a strategy that runs while you sleep, on your own wallet, with its stops onchain.