Can your notebook run a complete trading team locally?

The trading bots available on the market are basically one type: rule-driven, fixed parameters, and once online, they remain unchanged. When the market style changes, they become ineffective.

But the product we are introducing today is not that way.

The first sentence of its README is: "Your research desk, quantitative team, trading floor, and risk control department—all running on your notebook, 24/7."

This project is called OpenAlice. The author is @0xcherry.

It is an open-source AI trading agent engine, fully running locally, 24/7. The official statement is "Your research desk, quantitative team, trading floor, and risk control department." The official positioning is: "Your one-person Wall Street."

Over 3100 GitHub stars, fully open source, clone it and run it on your own machine, data does not leave local.

What can it help you do?

1. Market Research Automation

On the data side, OpenAlice has a built-in engine called TypeBB, modified from OpenBB, supporting stock quotes, cryptocurrencies, forex, macroeconomic indicators, and BLS Labor data. Technical analysis uses Excel-like formula syntax; if you write SMA(close, 20), it will give you the 20-day moving average.

Alice can also continuously monitor your specified RSS news sources to automatically fetch, archive, and analyze relevant information. For example, if you tell it to "follow all news about the Federal Reserve interest rates," it will read through, summarize impacts, and determine whether the effect on your position is bullish or bearish.

2. Multi-Asset Market Analysis

Supports stocks, cryptocurrencies, forex, options, futures, ETFs. Built-in Excel-like technical analysis formula engine (based on OpenBB) can calculate moving averages, RSI, Bollinger Bands, etc., and also supports macro data (like BLS data from the US Department of Labor).

3. Autonomous Order Placement

OpenAlice supports over 100 cryptocurrency exchanges, including Binance, OKX, and Bybit. In addition to cryptocurrency asset trading, it also covers US stocks, stock options, futures, and bonds.

After AI analysis, the order is "staged," and you see it with a note "confirm buy," which is equivalent to commit; you then push, and only then is it sent to the broker for execution. It will not automatically buy without your intervention.

4. Risk Control Shield

Each account has a "Guard Pipeline" to set the maximum position per trade, trading cooldown times, and allow trades for only certain assets. No matter how smart the AI is, it cannot bypass this requirement.

5. Public Company Research

For public company research, OpenAlice can pull financial statements, analyst expectations, insider trading records, and earnings calendar to help you conduct fundamental analysis.

6. Continuous Operation + Memory

Alice has a persistent "cognitive state"; it remembers the last analysis conclusions, tracks emotional states (yes, it has "emotions," used to influence decision styles), and supports cross-session working memory. You don't need to explain everything from scratch every time.

Three Basic Design Principles

OpenAlice has three design principles worth mentioning separately.

1. File-Driven

All states of OpenAlice exist in a local data/ folder. Markdown files, JSON configurations, conversation logs, human-readable.

You can open a text editor to see what OpenAlice remembers and what it is thinking. No database, no cloud, data sovereignty is completely in your hands.

No database, no cloud synchronization, nowhere you don't know about. This means that data sovereignty is entirely in your hands.

2. Reasoning-Driven

OpenAlice does not place orders based on rules but on continuous reasoning: reading quotes, reading news, calculating indicators, then providing judgments on what to do now, why, and where the risks are. Each decision has a complete reasoning chain behind it that is visible in real time, not a black box.

3. Operating System Native

Alice can directly utilize your operating system's capabilities: searching the web through a browser, sending you messages via iMessage, even connecting to local network speakers to wake you when necessary. It is not API in a sandbox, but real system permissions.

Users can also incorporate large models hosted locally into Alice as its "brain," with data at the model level also remaining local.

What are the specific application scenarios for Alice?

Scenario One: Monitoring Cryptocurrency Swing Trading

For example, if you hold BTC and ETH, you can have Alice monitor the 4-hour K-line and Coindesk news. When news or trends such as "Federal Reserve officials express dovish comments" or "BTC breaks past highs" occur, Alice automatically generates an analysis report + suggests a position increase plan, pushing it to your Telegram for your confirmation before executing with one click.

Scenario Two: US Stock Earnings Season

Before earnings season, you can have Alice organize analyst expectations and historical earnings performance of the companies you hold, generating a comparison report to tell you which ones have a "high probability of beating expectations." Alice will also remind you on the earnings release day and analyze the results immediately.

Scenario Three: Natural Language Order Placement

You can directly say in the Web UI chat box: "I think tech stocks will be under pressure, help me reduce my NVDA position by 30%, executed over three days with an average daily sell price."

Then Alice will break down the task, calculate price ranges, and stage three orders, waiting for you to confirm each one before pushing.

Scenario Four: 24/7 Unsupervised Monitoring

You can set Alice to check your position risk once an hour, and if any position loses more than 5%, it will wake you up via iMessage or Telegram. Truly achieving, while you sleep, AI watches the market.

How does it control itself?

The biggest risk of automated trading is: AI does things you don’t want it to do when you are unaware.

Through three layers, OpenAlice structurally blocks this possibility.

One is as mentioned earlier; all states of OpenAlice exist in the local data/ folder. We can always see what it is thinking and what it remembers.

The OpenAlice team believes that for a system that will handle real funds, hosting account credentials in the cloud is inherently unsafe, and regardless of technology used, local operation is the correct form for a trading agent.

Additionally, for order placement, the process follows: stage → commit → push, exactly like Git. This means OpenAlice cannot execute the order by itself after generating it; it requires your confirmation, signature, and then push before sending it to the broker. Without your two operations, Alice cannot access real funds.

Furthermore, on OpenAlice, each account has an independent Guard Pipeline. Maximum position limits, trading cooldowns, asset whitelists—you set them, and Alice cannot bypass them. No matter how aggressive the strategy is, if it cannot pass this barrier, it will not go out.

Some Unanswered Questions

The requirements for using OpenAlice are not high, but they are not zero either.

First, you need Node.js 22 or higher, pnpm 10 or higher, and the Claude Code CLI logged in correctly. Overall, someone with a bit of command line knowledge can get it running within two hours. Those who do not understand may get stuck on environment configuration for a day.

Currently, OpenAlice has over 3100 stars, 425 forks, and 463 commits on GitHub.

Open source happened in early 2026. The author's statement is that maintaining a commercial product requires handling pricing schemes, production environments, and permission systems; these tasks left him no time to improve OpenAlice itself. With open sourcing, he can focus on the core instead.

The team has just one request: no matter how you fork it, keep the name Alice.

For later commercialization directions, the author envisions three: data subscription, hosting services, transaction fee sharing are all options. But he says, before that, ensuring Alice is well-built is the most important.

But can reasoning-driven decisions continue to be profitable in real markets?

Not necessarily.

After all, the quality of Alice's decisions directly depends on the underlying large model. Large models can hallucinate, news can mislead, and technical indicators can become dull. How much the Guard Pipeline and manual approvals can compensate for this depends on the market environment, with no backtested data available for reference.

Even with the most powerful tools, we cannot guarantee our trades will be successful.

Moreover, the official also indicates this is experimental software and does not recommend using real funds unless you completely understand and accept the risks.

Project address: github.com/TraderAlice/OpenAlice

The Rhythm BlockBeats is looking for more AI Agent Trading products. We welcome recommendations, as we can bring you maximum exposure and early users. Feel free to DM us on Twitter @BlockBeatsAisa

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。