How do you run a trading bot in the cloud?
Updated November 13, 2026
A trading bot needs three things: your code, a machine that is on when the market is, and your broker's API keys. There are three ways to get the machine: rent a server and run everything yourself, deploy to a platform that hosts your algorithm, or hand the code to a service that runs it on a schedule.
On QuantGPT Pro, QuantGPT Cloud runs your own Python bot on the schedule you set and sends its orders to your own Alpaca account, paper or live, under your own keys. There is no server to set up. QuantGPT never decides a trade: the orders come only from the code you saved.
- Three ways to run a bot in the cloud
- How QuantGPT Cloud runs your bot
- What QuantGPT Cloud does not do
- Questions
Three ways to run a bot in the cloud
| Way | Who runs the machine | Brokers | What you write | Price, per its own pages |
|---|---|---|---|---|
| QuantGPT Cloud, on QuantGPT Pro | QuantGPT: each run in a fresh, isolated container, on your schedule in New York time | Alpaca today (paper and live accounts); U.S.-listed stocks and ETFs | A Python file, or one compiled from your own backtest that you can read and edit | Part of QuantGPT Pro, $149 a month with a 3-day trial: up to 5 bots, nothing per trade |
| QuantConnect live trading | QuantConnect: its docs say live algorithms run on co-located servers racked in Equinix, one live trading node for each algorithm | Its brokerages list: Interactive Brokers, TradeStation, tastytrade, Public, Alpaca, Charles Schwab, Webull and more, plus crypto exchanges | An algorithm in Python or C# | Paper and live trading are not on its $0 plan; Researcher shows $84 a month in its recommended setup |
| A server you rent | You: install, schedule, secure and watch it yourself | Any broker with an API | Your code and everything around it | DigitalOcean's Basic Droplets start at $4.00 a month (512 MiB, 1 vCPU) |
Alpaca, the broker QuantGPT Cloud connects to, is the other half of any of these: its Trading API takes the orders your code sends, commission-free for stocks, ETFs and options per its page, with paper trading on real-time data and a live account behind the same calls. Its page also offers an MCP server for trading from an AI assistant. It does not describe running your code for you.
How QuantGPT Cloud runs your bot
- Start from a backtest or a blank file. Press Run on QuantGPT Cloud on a backtest in your Journal and QuantGPT compiles its filters, ranking, number of names and cadence into a Python file, shown in full before anything runs. Or press + New bot and write your own with the bot library.
- Connect your Alpaca keys. Your browser seals them to QuantGPT Cloud's own key before they leave the page; they are stored, encrypted, only on QuantGPT Cloud, and the page never shows the secret back.
- Set when it wakes. A common schedule, days and a time, or five cron fields, in New York time, with an option to skip weekends and NYSE holidays. There is no default schedule.
- Paper or live. A new bot trades at your paper account. Before a live account: a pre-flight, an attestation that you wrote or reviewed every line and choose what it trades, and your recorded choice about a 30-day paper window, which is offered and never required.
- Read what happened. Every run and its output, every start, pause and change, and the orders as Alpaca reported them. Run once now runs it off schedule, up to 60 times a day per bot.
Every screen, step by step, with getting Alpaca keys: QuantGPT Cloud, step by step.
What QuantGPT Cloud does not do
- It does not choose what your bot trades, size a position, set a limit, time an order or stop a bot because of its results.
- It does not hold or move your money: your money stays at Alpaca.
- It does not trade crypto, options or futures: U.S.-listed stocks and ETFs only.
- It offers only the order types Alpaca has. A bot reaches only Alpaca's APIs and QuantGPT's own data host.
Questions
How do I run a trading bot in the cloud?
Put your code on a machine that is on during market hours and give it your broker's API keys. You can rent a server and run it yourself, deploy to a platform that hosts your algorithm, such as QuantConnect's live trading nodes, or, on QuantGPT Pro, save the bot on QuantGPT Cloud, which runs your own Python code on your schedule and sends its orders to your own Alpaca account, paper or live. Step by step: quantgpt.co/learn/quantgpt-cloud.
Can I run a trading bot without managing a server?
Yes. A hosted service runs the machine for you. On QuantGPT Pro, QuantGPT Cloud runs each run of your bot in a fresh, isolated container on the schedule you set, keeps every run's output and the orders as Alpaca reported them, and has nothing for you to install. QuantConnect hosts live algorithms written in Python or C# on its own servers.
Does Alpaca host trading bots?
Alpaca's Trading API takes the orders your code sends, with paper trading on real-time data and a live account behind the same calls; its algotrading page does not describe running your code for you (read 2026-11-13). QuantGPT Cloud, on QuantGPT Pro, is one place that runs it: your own code, at your own Alpaca account, under your own keys.
Does QuantGPT trade for me?
No. QuantGPT never decides an order. Orders come only from the code you saved, when your schedule wakes it or when you press Run once now, under your own keys. QuantGPT does not pick, size, time or manage any trade, and no QuantGPT AI tool can place, change or cancel an order.
What does QuantGPT Cloud cost?
It is part of QuantGPT Pro, $149 a month, with a 3-day trial: up to 5 bots per account. QuantGPT charges nothing per trade, nothing on account size and nothing on results; trading costs are Alpaca's. The QuantGPT plan, $49 a month, does not include it.
Related
- LearnQuantGPT Cloud, step by step
- ExecutionFrom a backtest to your broker
- LearnBacktest a trading strategy without coding
- LearnThe best AI for stock research, by the job
- PlansQuantGPT pricing
QuantGPT is research software and hosted compute, not a broker and not investment advice. Trading involves risk of loss. The disclosures and terms govern.