Crypto Trading Bots

Crypto trading bots are computer programs that buy and sell cryptocurrencies automatically according to rules set by their user. A bot watches prices and order books around the clock and places orders when its conditions are met, for example "buy when the price falls 2%" or "sell when the gap between two exchanges is wide enough." They are a retail form of algorithmic trading, used by individuals, trading firms and market makers.

Bots usually connect to an exchange through an application programming interface (API) key, or run directly on a blockchain as programs that interact with decentralised exchanges. They do not make trading profitable on their own: a bot simply carries out a strategy faster and more consistently than a person. Regulators warn that "guaranteed profit" bots are a common lure in investment fraud.

Think of a bot as a very obedient shop assistant who never sleeps. You leave instructions: "If apples drop below a certain price, buy a box; if they rise above another price, sell one." The assistant follows them exactly, all night, without getting nervous or bored.

The catch is that the assistant does only what you told it. If your instructions are bad, or the market changes in a way you did not foresee, it will keep following them anyway. And if someone steals the keys to the shop, the assistant cannot tell the difference.

• Grid bots. They place a ladder of limit orders above and below the current price and profit from prices moving back and forth within a range.
• Dollar-cost averaging bots. They buy a fixed amount at regular intervals, spreading purchases over time.
• Arbitrage bots. They exploit price gaps between venues, such as the kimchi premium in South Korea, in line with the general idea of arbitrage.
• Market-making bots. They keep buy and sell quotes on both sides of the order book and earn the spread, the job of a market maker.
• Signal and trend bots. They trade on indicators such as the relative strength index or moving averages, or copy the trades of others, like a copy trader.
• On-chain bots. A sniper bot buys new tokens the moment trading opens, and some bots engage in front-running by spotting pending transactions and jumping ahead of them.

Professional firms run similar strategies at much higher speed, the field known as high-frequency trading.

A grid bot divides a price range into equal steps. The spacing between orders is:

• $\Delta$ is the gap between neighbouring orders.
• $P_{\text{upper}}$ and $P_{\text{lower}}$ are the top and bottom of the chosen range.
• $n$ is the number of grid steps.

Example: a user sets a range of $90 to $110 with 10 steps, so $\Delta = (110 - 90) \div 10 = 2$. The bot places buy orders at $98, $96, $94 and so on below the price, and sell orders $2 above each filled buy. Each completed buy-and-sell pair earns about $2 per coin before fees. If the price breaks out of the range, the bot stops trading or is left holding coins bought at higher prices, so trading fees, slippage and trends can wipe out the small gains.

Bots that run on a centralised exchange need an API key, a credential that lets software act on the account. Exchanges let users limit what a key can do, and good practice is to:

• Disable withdrawals. A trading-only key cannot move funds off the exchange.
• Restrict by IP address. The key works only from the bot's own server.
• Protect the account. Use multi-factor authentication and revoke keys that are no longer used.

Even trade-only keys carry risk. In December 2022 the bot platform 3Commas confirmed that a large set of its users' API keys had leaked; attackers had used such keys to place unauthorised trades that drained value from accounts, for example by buying illiquid coins at inflated prices. Leaked keys can also be used for wash trading.

Many investment frauds use a "trading bot" as their cover story. BitConnect told investors that a "volatility software trading bot" generated its daily returns; the SEC said the money was diverted instead, the story behind the Hey Hey Hey meme. In January 2024 the US Commodity Futures Trading Commission issued a customer advisory titled "AI Won't Turn Trading Bots into Money Machines", warning that claims of high or guaranteed returns from bots and AI are red flags. It cited the Mirror Trading International case, in which the CFTC said more than $1.7 billion in bitcoin was taken from at least 23,000 people in a Ponzi scheme marketed as automated trading.

Warning signs include fixed daily returns, refusal to explain the strategy, pressure to recruit others and requests to send funds to the bot operator rather than keep them in your own account. A bot can be a useful tool for executing a plan; it cannot remove volatility or risk from a portfolio.

• CFTC: Customer Advisory Cautions the Public to Beware of Artificial Intelligence Scams