Skip to content

Market signals

Live pump, dump, breakout and volume signals on Hyperliquid and Aster, alerts you can subscribe to, and how well the signals have done.

On this page

Market signals is information only: it never trades for you. Bots are separate; they only do what you set up on Bots. On a computer, open Settings with the account chip at the bottom of the sidebar and pick Market signals in its menu. On a phone, open Profile > Market signals. (Older notifications and pages call it the scanner: it is the same thing.)

The scanner watches the most traded perps on Hyperliquid and Aster every minute and lists the ones moving unusually. You can get a push when a signal you care about fires, if push is on for this device. A line at the top of the page says whether it is live: how many markets it follows, the last candle it processed and when it last reported. Starting up means it is still loading price history; Signals delayed or Signals offline means signals may be missing.

The Arbitrage opportunities tab shows where the same coin pays differently on two venues, after both sides' fees, Prow's fee, slippage and the risk of the prices drifting apart. Each row says it in one line, for example "ETH funding differs between Aster and Hyperliquid: long on Aster, short on Hyperliquid", with what it would pay per $1,000 over the next 3 days if today's funding holds (the yearly pace is shown only as a note: it rarely lasts). By default it shows only what you could act on: rows that cost more than they earn, and Lighter rows (bots can't trade on Lighter), are hidden and counted; Show all brings them back. Details shows the costs, both venues' funding, the price gap, the depth, and what a pair needs (money on both venues). Funding differences close within days and price gaps within seconds; past ones don't predict future ones.

What it looks for

SignalWhen it fires
Volume spikeVolume in the last 5 minutes is far above its usual level for the coin (at least 4 robust deviations and 5× the 24-hour median), and the price moved at least 2%
BreakoutA 1-minute close above the coin's 24-hour high, on heavy volume
BreakdownA 1-minute close below the coin's 24-hour low, on heavy volume
PumpThe price rose at least 3 times its usual hourly range within 5 minutes, on heavy volume
DumpThe same move down
Squeeze setupThe hourly price band (Bollinger width) is narrower than in 90% of the past week's hours while open interest rose 5% or more in an hour. It says nothing about direction
Funding spikeFunding is far from its level of the last 7 days and at least 0.005% an hour. One side is crowded
Open interest surgeOpen interest (all futures positions open on the market) rose 10% in an hour

"Usual" is measured against the coin's own last 24 hours, so a quiet coin and a busy one are judged on their own terms. Each coin shows each signal at most once every 30 minutes (squeeze, funding and OI once every 4 hours).

Strength is how far past its thresholds a signal went: Weak just met the rule, Strong went about twice as far as the rule asks. Strong is rarer, not more reliable.

The table

Tap a signal's name (or the (i) on its filter) for its exact rule, what it doesn't tell you and how such signals did in our tests and live.

Each row shows the coin and venue, the signal, Move 5m before and Move 1h before (the price change up to the signal), Volume vs 24h median (×6.1 means six times), OI change 1h, Funding per hour, the level a breakout crossed and how long ago the signal fired. 1h after (before costs) fills in once the signal is 4 hours old; until then it says when (for a squeeze, the size of the move either way).

The line above the table says the period: Last 24 h, or, when the scanner started less than a day ago, Since the scanner started at …, so a short list isn't mistaken for a quiet day. Load older shows signals beyond the newest 200.

Direction (Both, Up, Down) is next to the list. The other filters open with Filters (kept in the page address, so a link opens the same view; Reset filters goes back):

  • Venue: all, Hyperliquid or Aster.
  • Markets: All ≥ $1M volume (the default) or Bot markets, the deep markets Prow bots can trade.
  • Strength (Medium+ by default: the weakest signals fire most and mean least).
  • Signal types: pumps, dumps, breakouts, breakdowns and volume spikes are on by default; squeeze, funding and OI are off.

A row shows the coin, the signal, the move over the last hour, volume and age. Details on a row opens the rest: the 5-minute move, open positions, funding, the breakout level and what the price did an hour later.

Chart opens the coin on Trade. On a bot market, Create bot opens the normal bot builder on that coin: that bot does not follow the signal, and nothing starts until you press Sign and start there. The scanner never places orders.

Scanner alerts

Press the bell on a row, or New alert, to get a matching signal in the bell. With push on for this device you get a push too.

  • Signals and Venues: which ones you want.
  • Markets: one coin, the bot markets, or all markets with at least $1M of daily volume.
  • Strength: any, medium and strong, or strong only. A new alert starts at Medium+.
  • At most once per coin every: 15 minutes, 1 hour, 4 hours or 1 day. A new alert starts at 4 hours.

You get at most 5 scanner alerts a day. They arrive in the bell's Notifications. Switch an alert off or delete it under My scanner alerts. Push needs to be on for this device, see Alerts.

How well do signals work?

Prow checks every signal against random moments on the same coin. Each signal is held for 15 minutes, 1 hour and 4 hours, with the fees you would pay: the taker fee both ways, the Prow fee and some slippage, about 0.24% per round trip. The random moments are taken at the same hour of day and in similarly volatile conditions, and pay the same fees.

Before launch we ran the same rules over past prices of the most traded markets: 60 on Hyperliquid (208 days of hourly prices) and 40 on Aster (30 days of 1-minute prices). With 1-hour holds:

SignalSignals traded in their direction, after feesRandom moments, after fees
Breakout−0.19% per trade on both venues−0.22%
Breakdown−0.40% to −0.44%−0.25% to −0.30%
Volume spike−0.25% to −0.29%−0.23% to −0.29%
Pump+0.06% (Hyperliquid) to +0.99% (Aster), few signals−0.21% to −0.24%
Dump−0.96% (Hyperliquid) to −1.66% (Aster)−0.24% to −0.27%

No signal did better than random by more than chance would explain. Pumps came closest, but there were few of them and the result was not reliable. Most signals lost the fees and a bit more. That is why the scanner only sends alerts and no bot trades on it.

The How signals did afterwards table on the page keeps score live from launch: for the last 30 days, per signal and venue, the share that made money after fees, the average after fees, the same for random moments, and the difference with its range. A range above zero means the signal beat random after allowing for the many signals tested. Past results do not predict future ones.

Last updated