Before you pay for this

What you're actually buying

Read this before you connect a real account. It's not a pitch, it's the specific ways this can go wrong.

  • ArbTrader only opens a trade when you tell it to. On Base it's a reporting tool: it shows you what the scanner found, and you place every trade by hand. On Quant you can copy a single whale trade and submit Kalshi and Polymarket US order tickets, but only after you open that specific trade or opportunity, set the size, and confirm it. On Terminal, if you switch on Auto-place, it places every spread that meets the rules you set (minimum return, dollars per spread, daily cap) while your dashboard is open, without asking each time. In Instant mode one side fills alone more often and is sold back, usually for a small loss. The other order it can send without a fresh confirmation is an Auto-sell you armed yourself on a position you already hold: it sells that position once you'd be ahead by your target while your dashboard is open (when it's closed, we email you instead and you sell it yourself), and if the book moves mid-sale some legs can sell and others not. International Polymarket legs are always placed by you.
  • Sending every leg is not the same as locking in the spread. Each leg is its own order, a bundle can have many legs, and they don't fill as one transaction. If a price moves first, one leg can fill while another doesn't. A ticket then retries the missing leg for a few seconds, at up to the price where the bundle still breaks even after fees. If that fails too, it takes whichever way out loses less: finishing the bundle at a small locked loss, or selling back what filled near the market price, which usually costs a little. If neither works, you're left with an unhedged position until you close or complete it yourself.
  • Order tickets need a Kalshi or Polymarket US key that can trade. That key lives in your browser's key store on the computer you added it on. Anyone who can use that browser profile could place orders with it; lock your computer, disconnect the key when you're done on a shared machine, and use a read-only key if you only want P&L. Requests are signed in your browser and relayed through arbtrader.co, which never sees the key itself. Polymarket US hasn’t published rules specifically about third-party tools placing orders with a customer’s own key; check their terms before you rely on it.
  • Copying a whale is a bet, not an arbitrage. A copy buys one outcome. If it loses, everything you put in is gone; nothing on the other side pays you back. Whales lose too, and a big trade is often not a view at all: it can be a hedge, a market maker, or one half of someone else’s arbitrage, so copying half of it can leave you with the risky half. By the time a trade shows up in the Whale Tracker the price has usually moved, so you’ll often pay more than the whale did for less upside. “Odds now” is the market’s current price for that outcome, not our prediction, and a trader’s past profit says nothing about their next trade. Copying is optional, never automatic, and each copy is one fill-or-kill order you size and confirm; a Polymarket trade can only be copied where Polymarket US lists exactly the same market, so read its rules before you do. Automatic exits (a stop-loss, a take-profit, or selling when the whale sells) only run while your dashboard is open, and each sells at whatever the book pays at that moment: a fast drop can sell below your stop, and a position sold on the way down won’t pay you if the outcome swings back. Selling when the whale sells means selling after them, usually at a lower price, since their sale moves the market first; we check their wallet every 30 seconds, and only Polymarket wallets can be followed, since Kalshi trades are anonymous.
  • Genuine, capturable arbitrage on Kalshi, Polymarket, Polymarket US and other apps is rare and closes fast. These markets are watched by bots that are faster than ArbTrader; a spread you see may already be thinner or gone by the time you act. Alerts and order tickets shrink the time between a spread appearing and your orders going out; they don't guarantee a fill at the price you saw.
  • Cross-platform matching is only as good as what you personally confirm. A Kalshi-to-Polymarket pairing you haven't verified can turn a "hedge" into an unhedged position, because the two markets may not resolve on identical criteria. Pairs are matched automatically only when both platforms list the same question with the same wording, numbers and deadline, and you can hide any pair or add your own; an automatic match can still hide a different rule, so read both sets of rules before you trade.
  • A bundle is only a hedge if exactly one listed outcome wins. We only list events where the platform’s own data says so, but a platform can clarify or change an event’s rules; read them before you trade.
  • The order-ticket integrations are built from each platform's own published API documentation, and the founder uses them on his own live accounts, but we never see yours and a platform can change its API without notice. Spot-check them against the platform's own numbers the first time you use each.
  • Fees and size are estimates at the moment we checked. Bundle spreads use Kalshi’s published fee formula and each Polymarket market’s own fee settings, at the size resting at the best price; your fills can differ, and rules or fees can change. Cross-platform pairs use the same published fee rules. Capital sits tied up until the event resolves, so judge a spread by its return over that time, not the raw number.
  • Legality and tax treatment of prediction-market trading vary by state and country and change often. Confirm you're eligible to use each platform from your location, and that its terms allow API-based trading on your account, before connecting a real one.

Nothing on this page is financial, legal, or betting advice. Past spread detection does not guarantee future opportunities, and trading and betting both involve risk of loss, including total loss of amounts wagered or traded.