HyperBot

No access to the bot yet?

Contact us

Access by subscription or license — reach out for a personalized demo.

Get access to HyperBot

Automated trading on Hyperliquid — pick the plan that fits you.

Monthly
€29/mo
  • Real-time dashboard access
  • Bot hosted on our infrastructure
  • Email support
  • No commitment
Choose
License / Purchase
Custom
  • Deployed on your own VPS
  • Custom code & strategies
  • Onboarding & training
  • You own the instance
Request a quote

Questions? [email protected] · Indicative pricing, to be adjusted.

My account

Authorizing asks your wallet to sign a consent message — no transaction, no gas. Stopping leaves any existing exchange SL/TP orders in place. Affects only your account.

Trading settings

your plan

Global defaults

applied to markets without an override

These limits guard your account only — they never affect the bot or any other account.

Strategy parameters

Fine-tune the selected strategy within your plan's limits. Each value is clamped to its allowed range; leave a field empty to use the strategy default. Parameters are stored per strategy and apply on your next order.

Trailing stop

Symbols & priority

Enable the markets you want traded and set their priority (1 = served first when the position cap is reached). Your plan caps how many you can run at once.

Total equity
Available · realized
Unrealized PnL
On open positions
PnL (24h)
realized · 24h
Total PnL
Realized since inception

Stats (7 days)

Win rate
Total trades
Profit factor
Sharpe ratio
Best trade
Worst trade
Current drawdown
Daily loss

Traded markets (active)

News

Loading…
PnL breakdown by exit reason (last 30 days)
Where your edge comes from — and where it leaks.
Loading…
Positive edge Loss source Top performer Inactive flow
Understanding the reasons

What each exit reason means

TP — Take Profit hit (includes TP triggers executed on the exchange). Planned winning exit.
SL — Stop Loss triggered at its original level (includes SL triggers executed on the exchange). Cost of being wrong (or of noise).
TRAILING — The trailing stop had ratcheted the SL, and that moved SL is what closed the position (win or loss). Isolates what the trailing stop actually captures or gives back.
SIGNAL — Exit on an opposite strategy signal (before TP/SL).
MANUAL — Manual close via Telegram or the exchange.
If your SLs outweigh your TPs, your stop is too tight or your edge is weak. If SIGNAL wins more than TP, you can widen your TPs. If TRAILING gives back more than it locks in, loosen the gap or raise its activation.
Asset matrix i
Asset Category Price 1D % Trend 7d Quality score Positioning Directional Bias
Loading…
· refreshed every 30 s Quality 0 (weak) 100 (strong)
Why bias changed i
Score history news sentiment (−1..+1) & quality (0–100) per pair, over time
Range
Scores
Price
Scale
Loading…
Wheel zoom · Drag pan · Shift+drag select a range · Double-click reset · Esc leaves the expanded view
Filters
Activity & stats (charts respect the filters above)

Stats

Net PnL
% return
Win rate
Total trades
Avg position size
Most used leverage
Maker / taker
Profit factor
Sharpe ratio
Best trade
Worst trade

PnL calendar

Strategy ranking & breakdown

🔥 Hot & ❄️ Cold

Loading…
Stats by day

Daily performance

Monthly stats — all time, the date range above does not apply

Performance by month

Fundamental impact

Every entry the news filter changed — sized down, let through, or refused

Impacted trades
Realized PnL
Win rate
Blocked orders
Trade history

Latest closed trades

Every article fetched by the pipeline — filter by pair, read the summary when available, and see the LLM rating once the model has classified the article.

Loading news…
Logs

Bot activity

API …
0 lines
Recent signals

Latest generated signals

Strategy settings
Loading…
Help & user guide

👋 Welcome to HyperBot

HyperBot trades your own Hyperliquid account for you. Your funds never leave your wallet: you grant the bot a Hyperliquid API agent that can place and cancel orders, and nothing else. It cannot withdraw, transfer or move your money.

Every account is fully isolated. Your traded pairs, your strategy, your parameters, your caps and your history are yours alone — you start from the strategy defaults, never from anyone else's settings.

The short version

  • The bot reads the market every few seconds and grades each setup from 0 to 1. Below your confidence threshold, it does nothing.
  • A news and macro layer can shrink a position, refuse it, or let one through that the trend filter would have rejected.
  • Every entry ships with a stop-loss and a take-profit placed on the exchange, so they survive even if the bot goes down.
  • This dashboard refreshes over WebSocket — no reload needed. Prices tick every second, positions on each loop.
Nothing trades until you say so. The bot only touches your account once you have linked it and pressed Authorize in My account. One click there stops it again, and stopping leaves your existing exchange stop-loss / take-profit orders in place.

Your subscription

Three plans. They differ in how many positions you can hold at once, which strategies you can run, and whether you get the full settings panel and the backtester.

Starter$14.99 / month · up to 10 pairs watched, 2 positions at a time · EMABreakout, RSIMeanReversion and ChandelierZLSMA · curated settings card in My account.
Pro$69.99 / month · unlimited watchlist, 10 positions at a time · every strategy including BBReversion · full Strategy settings panel · funding carry · backtest over 180 days, 10 runs a day.
Elite$199.99 / month · unlimited positions · everything in Pro · backtest over 365 days, 30 runs a day.

Six-month and yearly terms are discounted (roughly −15% and −25%). Your current plan, your licence expiry and an upgrade button sit at the top of this page.

What is not capped by your plan

Position size. There is no tier limit on it — the real gate is your own balance, checked with a safety buffer before each entry.

Licence vs subscription. The subscription is what you pay for; the licence is what authorises the bot to trade your account. If the licence lapses, the bot stops opening new positions on your account — existing ones and their exchange stops are untouched.

🔗 Connecting your account

Four steps, once. After that the bot runs on your account until you stop it.

1 · Create an API wallet on Hyperliquid

In the Hyperliquid app, generate an API wallet (agent) for your account. An agent can place and cancel orders on your behalf; it can never withdraw or transfer funds.

2 · Link it here

Open My account and paste your master account address and the agent private key. The key is encrypted server-side and is never stored in clear text. Until this is done, the bot has no way to trade and Authorize stays blocked.

3 · Fund the account

Your USDC stays on Hyperliquid, in your own account. The engine skips any account whose balance is below the minimum needed for your configured position size.

4 · Authorize

Press ⚡ Authorize bot (sign). Your wallet signs a consent message — no transaction, no gas. From there the bot starts trading your account on the next loop.

If the bot is not trading your account, My account tells you exactly why — expired subscription, missing link, missing licence, not authorized, or balance below the minimum. It never shows "authorized" while the engine is actually skipping you.

🧭 Navigation

  • Home — status, balance, equity, PnL, statistics, open positions and scores
  • Performance — stats by day and by month, the PnL calendar, strategy ranking and the fundamental-impact breakdown. Filterable by symbol, strategy, confidence and date.
  • Strategy — where your PnL comes from, split by exit reason (TP / SL / trailing / signal / manual)
  • Trade history — every recent closed trade, with fees and maker/taker
  • News — the headline feed the bot reads, with its impact rating
  • Fundamental — per-asset quality & news watchlist, market sentiment and the macro risk regime
  • Logs — full activity log, filterable by level
  • My account — your link, your authorization, your trading settings and your plan
  • Strategy settings — the full parameter panel and the Backtest tab (Pro and Elite)

The bottom bar carries the runtime meta: mode, active strategy, market sentiment (Fear & Greed + VIX), any announced maintenance, the local clock with world exchange hours, uptime and latency.

🟢 Bot statuses

Shown at the top of Home and via the dot next to the logo.

Active — running normally Degraded — running but with errors Stopped — the bot is not running Idle — no data yet
Your dashboard

💰 Reading the numbers

  • Available balance — uncommitted funds available to open new positions
  • Total equity — balance + market value of open positions
  • Unrealized PnL — floating gain/loss on positions still open
  • Realized PnL (session) — cumulative gain/loss since the bot started
  • Green means a gain, red a loss
Realized PnL is always the net figure read back from your Hyperliquid fills — fees included, at the real fill price. It is not an estimate from the mark price.
Your dashboard

📡 Prices & market data

  • Top ticker — every pair on the watchlist
  • "Live prices" card — only the pairs currently enabled for trading on your account, with their timeframe and, when you override it, the strategy running on that pair (BB-MR, EMA, RSI…)
  • 24h change — anchored on Hyperliquid's prevDayPx, so it matches the exchange exactly

Candles arrive over a WebSocket subscription per pair and are the same series the strategy grades. The timeframe is global to your account, not per pair.

Your dashboard

📌 Positions & trade history

An open position shows its side, size, leverage, entry, current price, unrealized PnL and both protective levels.

Tags you may see on a position

  • −50% — the fundamental filter halved the size because the news or the macro window opposed the trade. Hovering it says which of the two, and on what evidence.
  • BYPASS — the trend filter would have refused this entry; a strong, sustained news score aligned with the trade let it through.
  • MANUAL — you opened it yourself on the exchange. The bot displays it but keeps it out of its own statistics.

Closed trades

Each row carries the real exit price and net PnL read from your fills, the fees, whether entry and exit filled as maker or taker, and the exit reason. Hyperliquid's fills index can lag its clearinghouse by several seconds — when that happens the row is repaired a few minutes later rather than left with a stale price.

Your dashboard

🏁 Performance

Everything on this tab respects the filter bar: symbol, strategy, minimum confidence and date range. Two blocks deliberately ignore the dates and always cover your full history — the PnL calendar and Monthly stats.

Stats worth knowing

  • % return — net realized PnL ÷ equity at the start of the window
  • Profit factor — gross wins ÷ gross losses. Above 1 means the wins pay for the losses.
  • Sharpe ratio — per-trade mean ÷ standard deviation, scaled by √n
  • Maker / taker — the share of notional filled as a resting order. Maker costs 0.015% against 0.045% taker, so higher is cheaper.

Fundamental impact

The last block on the tab answers "what did the news layer actually change?". Filter it by the kind of impact:

  • −50% macro — halved by a macro risk-off window, both directions
  • −50% news — halved because the news opposed the trade's direction
  • Bypass — entries the trend filter would have refused, taken on the news score. This is where you see whether the bypass earns its keep.
  • Blocked — every entry the bot refused: news filter, disabled pair, risk limit or a failed order. No position, so no PnL — only a count and a reason.
Your dashboard

🔔 Notifications

The bell collects what happened while you were not looking: positions opened and closed, macro events on the calendar, and every entry the bot refused with its reason.

The feed is persisted, not held in memory, so a restart no longer wipes it. The unread dot is driven by timestamps rather than row ids, which is what makes it survive a restart correctly.

Refused entries are also counted exactly — repeated refusals on the same pair each count, where the feed itself only shows one every 15 minutes to stay readable.

Fundamentals

🛡️ Quality score

A 0–100 health rating per asset, separate from the news direction. Crypto reads free on-chain fundamentals; TradFi commodities/indices read CFTC COT positioning & trend. The enriched composite now also modulates trading — it amplifies entries aligned with strong fundamentals and trims those that oppose them.

BTC78
XYZ 10051
SpaceX ≈≈72

Data-less assets (e.g. SpaceX) get a global LLM estimate, flagged (no pillar bars), which still feeds trading by design.

Fundamentals

📊 Directional & macro scores

The bot continuously reads live news and an economic-events feed and turns them into numbers that nudge position sizing and can veto a trade. There are two of them — a per-asset directional score (the single number shown per asset) and a global macro risk-off penalty (surfaced as a separate flag when active, since it is not a direction).

📈 Directional score

The directional score (dir) is the per-asset news sentiment, from -1 (very bearish on this asset) to +1 (very bullish). Only real news about that asset moves it — the global macro mood is kept out on purpose (it lives in the Macro score), so a directional reading always points to a real side.

−1 bearish0 neutral+1 bullish

↑ Example marker at dir = +0.62BULLISH

How it is computed

  • Every headline younger than 24h is scanned for bullish / bearish keywords (rally, ETF approval, inflows… vs hack, lawsuit, sell-off…).
  • Recency weighting (6h half-life): <6h ≈ full weight, <12h ≈ 0.6, <24h ≈ 0.2, older = ignored.
  • High-impact headlines count double (×2).
  • The asset's own news counts at full weight; generic market-wide news only at 35%, so each asset is driven by its own story.
  • Blended 65% average / 35% strongest single headline (so one big story isn't diluted by noise), then squeezed through tanh into -1 … +1.

Labels

BULLISH≥ +0.5 SLIGHTLY_BULLISH+0.2 … +0.5 NEUTRAL−0.2 … +0.2 SLIGHTLY_BEARISH−0.5 … −0.2 BEARISH≤ −0.5
Example · BTC

Three fresh headlines: "Bitcoin ETF sees record inflows", "BTC rallies past resistance", "Institutional adoption climbs".

dir = +0.62 BULLISH. A LONG trades with the news (full size); a SHORT would be fighting it (see Effect on trades).

🌍 Macro score

The macro score is a global, non-directional risk-off penalty from 0 (calm) down to -1 (major event). It is the same for every asset in a given tick and never picks a side — it only makes the bot more cautious by trimming size. It is also what raises the ⚠️ Macro event card in the Notif tab.

It is driven by the scheduled economic calendar (the same releases shown on Investing.com's calendar), not by news headlines — so it reacts to real, timed events (FOMC, CPI, NFP…) and ignores analyst-rating / forecast articles. Only releases for the market-moving currencies (USD, EUR, GBP, JPY, CHF, CNY, plus cross-region events) are considered.

−1 major−0.5 moderate0 calm

Detection window

An economic event counts while it is inside the caution window: from 2h before until 1h after the scheduled time. The strongest active event wins.

Event tiers

Tiers map straight to the calendar's impact rating (the ★ stars on Investing.com). Low-impact events are ignored entirely — no penalty, no Notif card.

−1.0 Tier 1 — High impact (★★★): FOMC / Fed decisions, CPI, NFP / non-farm payrolls, GDP, rate decisions…
−0.5 Tier 2 — Medium impact (★★): PPI, retail sales, jobless claims, ISM / PMI, PCE, housing, trade balance…
Example · any asset

A "FOMC Statement" release is scheduled 90 minutes from now → inside the window, High impact = Tier 1.

macro = -1.0 for all pairs this tick. No side is favoured, but both LONG and SHORT get −50% size, and the Notif tab shows the event with its currency and scheduled time.

🧮 The score you see

The number shown per asset is simply the directional score (dir) — a single, signed reading (−1 … +1) that answers one question: which way is the news pointing? It is exactly the value the trade logic acts on for the veto and the size cut, so what you read is what drives the direction.

The macro risk-off is deliberately not folded into that number. It is not a direction (it is always ≤ 0 and only means "be cautious"), so blending it in would drag the score negative without anything being bearish. Instead it appears as a separate ⚠ Global risk-off flag next to the score, shown only while a risk-off is live.

−1 bearish0 neutral+1 bullish
Example · XYZ100 (live)

Bullish index news, no macro event: dir = +0.408.

→ the card shows +0.41 SLIGHTLY_BULLISH, with no risk-off flag. (Internally the bot still logs a blended fund_score = dir×0.65 + macro×0.35 for history, but the UI shows the clearer directional number.)

🎯 Effect on trades

Here is exactly what these scores do when a strategy fires an entry signal:

Opposition veto. An entry that fights a strong directional score is blocked entirely — a LONG when dir ≤ −0.6, or a SHORT when dir ≥ +0.6. It surfaces in Notif as "Order not executed — Fundamental opposition". Threshold: fund_opposition_threshold (default 0.6; set ≥ 1.0 to disable).
✂️ Size −50%. The position is halved when the trade opposes the directional score by ≥ 0.3 (a LONG into bearish news / a SHORT into bullish), or when the macro penalty is strong (macro ≤ −0.5). No stacking — it is a single −50%.
Entry confidence bonus (optional, off by default). A strong directional score can add confidence to that side, and on its own push it over min_confidence — i.e. a big news can trigger an entry. Enabled via fund_bonus_enabled.

Three quick scenarios

A · trade with the news

BTC dir = +0.62, no macro. A LONG signal → full size. A SHORT signal → blocked (opposition veto).

B · macro risk-off

SILVER, no asset news (dir = 0) but FOMC in 90 min (macro = −1.0). Either side may still open, but at −50% size; Notif shows the Tier-1 event.

C · mild tailwind

XYZ100 dir = +0.41, no macro. A LONG is not vetoed and keeps full size (opposition only kicks in past ±0.6).

Fundamentals

📰 News feed

The News drawer shows headlines with a real RSS image (or a generated tile fallback). Impact is shown as a coloured outline on the card, and an LLM filters each item for relevance and sentiment before it is displayed.

  • Toggle High impact only to keep the market-movers
  • All vs Fundamental tabs split generic news from asset-specific stories

Headlines are pre-filtered by keyword before any model sees them, then classified in batches — that is what keeps the reading current without an unbounded LLM bill. When no model is reachable, the bot falls back to a lexicon rather than going blind.

Fundamentals

😨 Market sentiment

Two crowd gauges, shown both in the bottom bar and in full on the Fundamental tab. They are context only — neither of them can open or refuse a trade on its own.

Crypto Fear & Greed

A 0–100 reading from alternative.me: 0 is extreme fear, 100 extreme greed. The card shows today's value with its classification plus the readings from yesterday, last week and last month, so you can see which way the crowd is moving rather than just where it stands.

VIX

The TradFi fear gauge, from Yahoo Finance: the level, the change against the previous close, and a 30-session sparkline. A rising VIX is risk-off, which is why the change is coloured red when it goes up — the opposite of a normal quote.

Where these do matter is inside the macro risk regime: the VIX is one of the measured inputs behind the quantitative half of that score.

🧩 How a trade is decided

Every entry runs the same gauntlet. Any single gate saying no means no trade — and the refusal is recorded with its reason in Blocked orders.

1 · The strategy grades the setup

Your pair's strategy scores the current candle from 0 to 1. Below your confidence threshold, nothing happens. A higher threshold means fewer but stronger trades.

2 · The spread has to be sane

A pair whose bid/ask spread is wider than max_spread_pct is skipped: the edge would be paid straight to the book.

3 · The fundamental layer weighs in

  • A strong news score opposing the trade blocks it outright.
  • A weaker opposition, or an active macro risk-off window, halves the size instead.
  • A strong, sustained aligned score can do the reverse: let an entry through that the trend filter would have refused (the bypass). The score has to have held past the threshold for the whole window — a spike is not enough.

4 · Risk has the last word

Position count against your plan's cap, daily-loss guardrail, per-pair enable switch, leverage cap and the balance check with its buffer. Then the order goes out, and the stop-loss and take-profit are placed on the exchange immediately after the fill.

The protective orders live on Hyperliquid, not in the bot. If the bot restarts, or goes down entirely, your stop and target stay armed and the position is re-adopted on the way back up.
Strategies

🌲 Strategies — overview

Each pair runs one strategy. Starter runs the first three; BBReversion needs Pro or Elite. Pick yours in My account, or per pair in Strategy settings:

EMABreakout — trend RSIMeanReversion — range ChandelierZLSMA — trend + trail BBReversion — Bollinger mean-revert

Every entry must clear a confidence threshold. Each strategy grades the current setup on a 0–1 scale and opens a trade only when that grade reaches its own min_confidence_* value (unset → the global min_confidence). A higher threshold means fewer but stronger trades; a lower one means more but looser trades. You set it in Strategy settings.

Each strategy has its own page in the sidebar. The worked examples there all assume the threshold is set to 0.70.

Strategies

📈 EMABreakout

Trend breakout · threshold key min_confidence_ema · available on every plan.

Buys when a fast EMA crosses above a slow EMA inside an up-trend (mirror for shorts). It adds points for each thing that confirms the move, then divides by 6 to get the grade.

Open a LONG with threshold 0.70 — you need enough of these:
  • EMA fast crosses above the slow EMA → +1.5 (required: no cross, no trade)
  • Price is above the trend EMA → +2.0
  • RSI in the bullish zone (52–70) → +1.5
  • A strong green candle (high volume) → +0.5

Cross + price above trend + RSI = 5.0 ÷ 6 = 0.83 ≥ 0.70 → ENTER.
Cross + price above trend only = 3.5 ÷ 6 = 0.58 → wait (even with volume, 4.0 ÷ 6 = 0.67, still under 0.70). So at 0.70 the RSI confirmation is effectively required. On top of the grade, ADX must show a real trend and Supertrend/KST (if enabled) must agree with the direction.

DirectionEntry conditions (all required)Confidence score (÷6 ≥ threshold)
LONG EMA fast crosses above slow (hard gate + grace window) · ADX regime OK (slope-zoned around adx_min_trend) · Supertrend and KST bullish (when enabled) cross 1.5 + close > EMA-trend 2.0 + RSI 52…rsi_overbought 1.5 + volume > SMA×volume_multiplier (green) 0.5
SHORT Mirror: EMA fast crosses below slow · ADX OK · Supertrend and KST bearish cross 1.5 + close < EMA-trend 2.0 + rsi_oversold…48 RSI 1.5 + volume (red) 0.5
Exit / stopTrigger
EXIT_LONGOpposite EMA cross down, or RSI(closed) ≥ rsi_exit_overbought (80). Can immediately reverse to SHORT if that side is itself valid.
EXIT_SHORTOpposite EMA cross up, or RSI(closed) ≤ rsi_exit_oversold (20). Can reverse to LONG.
SL / TPSL = ATR × atr_multiplier_sl · TP = ATR × atr_multiplier_tp, plus the adaptive trailing stop (moves only in profit, never locks a loss).
Strategies

🔄 RSIMeanReversion

Range fade · threshold key min_confidence_rsi · available on every plan.

Buys a dip when RSI is oversold but price is still above its long-term average (SMA200) — and only while the market is ranging, not trending. The grade blends how oversold the RSI is with how far price sits above the SMA200.

Open a LONG with threshold 0.70 — the grade adds up two parts:
  • How oversold RSI is (below 30) → up to +0.5 (the lower the RSI, the more)
  • How far price sits above the SMA200 → up to +0.5 (the further, the more)

A deep dip (RSI ≈ 12) with price ≈ 0.8% above the SMA200 ≈ 0.70 → ENTER. A shallow dip (RSI just under 30) tops out around 0.58 → wait, even far above the SMA200. So at 0.70 the RSI must be genuinely oversold. ADX must also stay below adx_max_mean_reversion (25) — no fading a real trend.

DirectionEntry conditions (all required)Confidence score
LONG RSI < rsi_oversold (30) · price above SMA200 · ADX ≤ adx_max_mean_reversion (25) 0.5 × (30 − RSI)/30 + min(0.5, distance to SMA200 %) ≥ threshold
SHORT RSI > rsi_overbought (70) · price below SMA200 · ADX ≤ adx_max_mean_reversion 0.5 × (RSI − 70)/30 + min(0.5, distance to SMA200 %) ≥ threshold
Exit / stopTrigger
EXIT_LONGRSI ≥ 50 (mean recovered). Never blocked by ADX — always allowed to exit.
EXIT_SHORTRSI ≤ 50.
SL / TPSL = ATR × atr_multiplier_sl · TP = ATR × atr_multiplier_tp + adaptive trailing.
Strategies

🕯️ ChandelierZLSMA

Trend following with a built-in trail · threshold key min_confidence_ce · available on every plan.

Follows the Chandelier Exit trend-flip (its stop line also becomes your trailing stop), but only takes the flip when price has clearly separated from the ZLSMA trend line — a flip glued to the line means a weak, choppy trend.

Open a trade with threshold 0.70:
  • Grade starts at 0.60 on any valid flip
  • + 0.20 × (distance from price to ZLSMA, measured in ATR), capped at 1.00

0.60 + 0.20 × 0.5 = 0.70 → ENTER when price is at least half an ATR away from the ZLSMA. A flip right on the line stays at 0.60 → wait. LONG also needs price above the ZLSMA, SHORT below it.

DirectionEntry conditions (all required)Confidence score
LONG Chandelier direction flips −1 → +1 (buy flip) · price above ZLSMA (when ce_require_zlsma) 0.6 + 0.2 × |price − ZLSMA| / ATR, capped 0.6–1.0 ≥ threshold
SHORT Chandelier flips +1 → −1 (sell flip) · price below ZLSMA · ce_allow_short on Same distance-to-ZLSMA score ≥ threshold
Exit / stopTrigger
EXIT_LONGOpposite Chandelier flip (sell flip). Can reverse to SHORT if ZLSMA confirms.
EXIT_SHORTOpposite flip (buy flip). Can reverse to LONG.
SL / TPSL = Chandelier stop (else price ∓ ATR × ce_sl_atr_mult) · optional TP = ATR × ce_tp_atr_mult (0 = off, let it run until the flip).
Strategies

🎈 BBReversion

Bollinger mean reversion · threshold key min_confidence_bb · Pro and Elite only. This is the strategy the reference account runs.

Fades a close that pokes outside a Bollinger band back toward the middle band, when a fast RSI confirms panic, the ADX shows range-strength, the bands are wide enough, and the move aligns with the higher trend EMA. The grade grows with how deep the close pierced the band.

Open a LONG with threshold 0.70:
  • Grade starts at 0.65 once price closes below the lower band
  • + 0.15 × (how far below the band, measured in ATR), capped at 0.95

0.65 + 0.15 × 0.33 ≈ 0.70 → ENTER when the close pierces the band by about a third of an ATR. A close just grazing the band stays at 0.65 → wait. It also needs RSI ≤ mr_rsi_buy (15), ADX inside its range, wide-enough bands, and price above the trend EMA.

DirectionEntry conditions (all required)Confidence score
LONG Close below lower band (mr_bb_period/mr_bb_std) · RSI ≤ mr_rsi_buy (15) · ADX in [mr_adx_min, mr_adx_max) · band width ≥ mr_min_bb_width_pct · close > trend EMA (when mr_trend_filter, unless strong aligned news overrides) 0.65 + 0.15 × (band − close)/ATR, capped 0.95 ≥ threshold
SHORT Close above upper band · RSI ≥ mr_rsi_sell (85) · same ADX + width gates · close < trend EMA (when filtered) 0.65 + 0.15 × (close − band)/ATR, capped 0.95 ≥ threshold
Exit / stopTrigger
TP (mean)Close reaches the mid-band captured at entry — the mean it reverts to (mr_tp_mid_band).
Time stopPosition held ≥ mr_max_hold_candles — a reversion that never happened is a failed thesis.
Opposite bandOptional (mr_exit_opposite_band_enabled): close crosses the far band for the full reversion.
SLSL = ATR × mr_sl_atr_mult; can ratchet toward the mid-band once back at breakeven (mr_sl_to_mid_enabled).

🎯 Risk & exits

Every position is bracketed on the exchange from the moment it fills, and the bot never widens a stop — it only ever moves it in your favour.

The protective orders

  • Stop-loss and take-profit are real exchange orders, placed right after the entry fill. They stay armed if the bot restarts or goes offline.
  • Both are re-anchored to the actual fill price. If the market moves between the signal and the fill, the distance you intended is preserved instead of landing on the wrong side of the entry.

Adaptive trailing stop

  • The trailing distance is sized per pair from its own volatility (ATR-based), so a calm pair trails tighter and a volatile one gets room to breathe.
  • It never locks in a loss: it only starts moving once the position is in profit, and it only ratchets in the protective direction.
  • Once it has moved your stop, an exit at that level is reported as TRAILING rather than a plain SL, so the Strategy tab can show you what the trail actually captures — or gives back.

Guardrails

  • Daily loss cap — exchange-side stop fills count toward it too, so a run of stop-outs really does trip it.
  • Position cap — 2 on Starter, 10 on Pro, unlimited on Elite.
  • Dead-man switch — if the trading loop ever stalls, the bot cancels its working orders. It leaves the stop-loss and take-profit of a live position alone.
  • Manual positions — anything you open by hand is tagged and kept out of the bot's statistics, so your performance figures only reflect the bot's own trades.
Settings

⚙️ Strategy settings

The full parameter panel, on Pro and Elite. Starter gets a curated settings card in My account instead.

Everything you change here applies to your account only, and takes effect on the next order — roughly a second later. Nothing you touch can affect anyone else's account, and you always start from the strategy defaults.

What you can set

  • Global — position size, leverage, confidence threshold, maximum spread, daily loss cap, trailing and break-even.
  • Per strategy — the indicator parameters of whichever strategy you run.
  • Per pair — enable or disable it, and override its strategy, leverage and position size. Disabling a pair blocks new entries; it never closes what is already open.
The timeframe is global to your account — there is no per-pair timeframe. Changing it re-seeds the candle history for every pair you trade.
Settings

Backtest

Replay your parameters over past data before you commit real money. The Backtest panel lives in Strategy settings — Pro and Elite.

  • Pick a pair, a timeframe and a strategy, then run it over the last 180 days on Pro or 365 days on Elite. Fees, funding and leverage are all included.
  • 10 runs a day on Pro, 30 on Elite. One simulation runs at a time, so the allowance is what keeps the queue fair.
  • Live progress while it runs, and a Stop button.
It never disturbs the bot. The simulation runs in a separate process on history downloaded ahead of time — your positions and the trading loop are untouched.
Read the result as indicative. The candles come from Binance perpetuals, not Hyperliquid: the levels track closely, but wicks, spread and funding differ. A backtest is evidence, not a promise.
Settings

Live control

Your on/off switch lives in My account.

  • ⚡ Authorize bot (sign) — your wallet signs a consent message. No transaction, no gas. The bot starts trading your account on the next loop.
  • 🛑 Stop bot on my account — one click. It stops opening new positions immediately and leaves your existing exchange stop-loss / take-profit orders in place, so anything still open stays protected.

Both act on your account only. If the bot is not trading you, the same page lists every reason why, so "authorized" can never look like "trading" when the engine is actually skipping you.

🛎️ Maintenance & status

Announced downtime — ours and Hyperliquid's own — appears in the bottom bar, just left of the clock. Click the warning triangle to fold the notice away; the choice sticks, and the triangle brings it back.

What a scheduled restart means for you

  • Your open positions stay open. The exchange holds them, along with their stop-loss and take-profit, for the whole restart.
  • The bot re-adopts them on the way back up and carries on managing them.
  • New entries pause for the couple of minutes the restart takes.

Hyperliquid's own scheduled maintenances are read from their status page, so you see an exchange-side window in the same place as ours. A Details link opens their notice.

FAQ

Can the bot withdraw my funds?

No. It holds a Hyperliquid API agent, which can place and cancel orders and nothing else. Withdrawals and transfers are impossible with an agent key.

What happens if the bot goes down while I have a position open?

Nothing changes on the exchange. Your stop-loss and take-profit are real orders sitting on Hyperliquid, and they stay armed. The bot picks the position back up when it returns.

Why did the bot not take a trade I expected?

Open Performance → Fundamental impact → Blocked. Every refusal is recorded with its reason: news opposition, disabled pair, risk limit, spread too wide or a failed order.

Why is a position half the size I configured?

The fundamental filter halved it — either a macro risk-off window or news opposing your direction. The position carries a −50% tag saying which, and on what evidence.

Can I trade manually on the same account?

Yes. Positions you open yourself are tagged MANUAL and excluded from the bot's statistics, so they never distort your performance figures.

Does changing a setting affect anyone else?

No. Pairs, parameters, caps and timeframe are scoped to your account.

Do I lose my settings if I upgrade my plan?

No. Upgrading raises your caps and unlocks modules; your configuration stays as it is.

📬 Contact & links

Hyperliquid   Bot Telegram

For any question or information request, contact us by email only at [email protected].

What's new

Every user-visible change, newest first. The most recent release is also announced once in a popup — dismiss it and it won't come back until the next one.

Bot control
Control channel:

Global defaults

applied to markets without an override
Markets
Priority ranking — when several pairs signal at the same time, the top-ranked (highlighted) pairs claim the open-position slots first.
Filter active — ranking is frozen. Clear the filter to reorder.
No pair matches this filter.