The usual answer to "how do I build a Kalshi trading bot" is a Python tutorial: get an API key, poll the markets, write the order logic, host the loop somewhere, add the risk controls you forgot, rotate the key when it leaks. That works, and it is a weekend for someone who already writes code. This guide is the other route: a Kalshi bot built in Rivo's bot builder in three steps, tested before it runs, and live through your own key.
Why Kalshi is a good venue for a copy bot
Kalshi is unusual among prediction markets in that many traders are visible by name. Rivo tracks those traders, records every trade we observe them make, and scores each position when the market settles. A trader's profile carries a real profit curve, a win rate against the odds they paid, results by bet size and category, and the markets they trade most. That record is what a copy bot is built from. The raw Kalshi fill feed, by contrast, carries no identity at all, which is why a bot cannot be built from fills alone.
Step 1: pick who or what it trades
Open Bots and choose New bot. Start from Sports Whales or another template. Choose Build from scratch on the empty dashboard to open a blank signal. The builder lets you set the venue and category then refine the entry price and source trade size.
Copy a trader. Open a trader on the Kalshi whale tracker or the ranked traders. Read the record. If it holds up, the profile's "Build a bot that copies" action drops the trader into the builder. The bot reacts every time they open a position, in the market they chose, on the side they chose, at the size you set.
Build a signal. Kalshi prints a steady stream of large trades across sports, politics, economics and weather. A signal picks out the ones you want: a category or league, a minimum size, a keyword in the market title, a specific market, an entry price band, and whether the trade opened a position, added to one or flipped sides. Four templates give you a working signal in one click.
Step 2: read the backtest at your size
Choose a fixed amount per trade or a percentage of the source trade with a cap. The builder replays the trigger at that size on Kalshi markets that already settled. For a trader you get the record, the return and the drawdown over 7, 30 or 90 days. For a signal you get every match from the last 90 days, split into a training window and a validation window, with an overfit flag when the two disagree.
One Kalshi-specific detail: the price band in a signal is the YES price. A trader buying NO at a "YES under 20 cents" market paid the complement, over 80 cents. The signal summary says so when it applies, and the backtest scores the trade at the price the trader actually paid.
Step 3: choose how it runs
Choose paper to track simulated results or select supported live execution. Set the schedule and choose daily summaries or every-trade updates for a signal bot. The dashboard keeps your bots together. Click a card to inspect its trades and change its settings. Use the separate Alerts column to follow traders for notifications.
Alerts only tells you when the trigger fires and records nothing. Paper copies every entry with virtual money at your size and scores it on settlement. Live places the same entries as real orders through your own Kalshi API key inside a per-trade and a per-day ceiling.
Start on paper and promote later from the bot's own page; the trigger and size carry over. When you promote, the builder asks you to connect a Kalshi API key in settings. The key is encrypted at rest, wiped when you disconnect, and cannot withdraw. Funds stay in your Kalshi account.
Control how your Kalshi bot trades
- It places entries only. Closing stays your decision, or a stop-loss and take-profit rule you arm per position.
- It sizes down to the depth available inside your slippage boundary rather than pushing an order through a thin book.
- It stops at both ceilings, and you can pause it from its page.
- It skips a trade that has aged past the copy window and logs the reason.
- It pauses itself after three consecutive execution failures.
Build on a record you can inspect
A Rivo bot starts with activity you can inspect. A trader bot follows the accounts you choose. A signal bot selects incoming trades that match your criteria. Both connect the historical record to the trades you track next. Explore the builder to choose your starting point.
Driving the bot from code anyway
If you do write code, the same bots are available through the Rivo REST API and the MCP server. A trader bot is a follow, a signal bot is a strategy, a live bot needs a trade-scoped key with an explicit confirmation on the call, and a per-key daily spend cap fails closed. You keep the guardrails and lose the hosting.