TraderAlice: Cursor for Vibe Trading

What is Vibe Trading?

When most code is written by Cursor, we want most trading to be done by AI. That's why TraderAlice exists.

---

When friends ask how to become a developer from scratch, we usually point them to AI IDEs like Cursor and Claude Code. These tools let people describe what they want in plain English and get working code in return. As of December 2025—when this post is being written—most new code is generated by AI.

The same need exists in trading. When people want to become traders, they face a learning curve that goes way beyond programming: data sources, order matching, risk management, strategy design... all essential knowledge, yet we couldn't find a tool as plug-and-play as Cursor to help people build and test trading strategies.

TraderAlice was born from a simple, almost naive idea:

What if people could describe trading strategies in natural language and let an AI Agent handle the execution?

IDE and Trading: A Surprising Parallel

The inspiration for TraderAlice wasn't accidental. We realized that trading and coding are structurally similar.

Cursor is best described as an "AI-powered IDE"—an agent that collaborates with human developers on the same codebase, syncing state changes to a git repository. This process mirrors trading almost exactly.

At their core, both processes follow the same loop: understand → modify → test → commit. One operates on code, the other on markets. The ReAct framework by Shunyu Yao captures this beautifully: AI agents complete tasks through cycles of thinking → acting → observing.

Taking it further, both trading and coding can be viewed as state modification. Whether it's the state of a codebase or the state of a portfolio, changes happen through a series of operations—performed by humans and AI agents working together.

So no, Vibe Trading isn't some distant, unattainable holy grail. We're actually one step away from the answer.

Meet Alice: Your Vibe Trading Agent

Drawing from Cursor, Claude Code, and Codex, we built TraderAlice—an AI Agent designed specifically for Vibe Trading. As you might have guessed, we just call her Alice.

Alice is built around a few core modules:

• AI-Native News Retrieval — Think grep for financial news. Alice can pull relevant information from massive news feeds with simple parameter configs, filtering and trimming as needed.

• DataFrame & Operators — Inspired by freqtrade, we let Alice load candlestick data into DataFrames and compute indicators using Excel formulas or Pandas operations. In plain terms: we gave the AI a spreadsheet, so she can crunch numbers like a human trader.

• Trade Primitive Generator — Alice constructs complex order groups using special trading primitives—like chase orders with built-in TP/SL. When a primitive is proposed, Alice automatically learns its "effect," ensuring orders are both executable and sensible.

• Trading as Git (coming soon) — We're building portfolio versioning tools. Trust me, this is going to be cool.

Alice is heavily inspired by Cursor, but with designs unique to the trading domain. Through these modules, she can Vibe Trade on virtually any asset.

Traditional Quant vs. Agentic Quant

If you're familiar with traditional quant trading, you'll notice TraderAlice takes a fundamentally different approach.

Traditional quant relies on explicit calculations: moving average crossovers, RSI overbought/oversold signals, fundamental metrics, ML models, even NLP-based sentiment analysis. These methods work, but they're all predefined strategies.

TraderAlice introduces what we call Agentic Quant—preserving quantitative capabilities while letting the AI decide when and how to use them. Triggers can be news events, market volatility, or just plain English instructions. She might skip quant entirely and just buy Tesla every time Trump gives an exciting speech.

Though as of December 2025, Trump's statements are more likely to tank Tesla than pump it.

The point is: Alice isn't a quant trading agent. She's a general-purpose Trading Agent with quant capabilities. She can use quantitative methods when needed, but isn't limited to them—just like Cursor can write quant code but isn't restricted to it.

Typecheck and Backtest

When people talk about Trading Agents, what they really want is for trading strategies to be testable like code—typechecked, built, compiled, linted, unit tested. People want to verify that things actually work.

We can't predict the future like fortune tellers, but we can backtest to validate AI strategies—at least proving they worked in the past. When Alice knows her bright ideas will be thrown into backtests for validation and iteration, she tends to propose more viable strategies.

TraderAlice agents run in what we call a Sandbox. In live trading mode, the Sandbox connects to real-world data. In backtest mode, it creates a "virtual world" containing all necessary information from the target time period.

Imagine sending Alice back to the 2020 market crash—what strategy should guide her? Once we see how a strategy performs in extreme conditions, we can apply those learnings to the 2022 crypto meltdown. That's the power of backtesting.

Unlike many AI startups that would use this section to paint visions and announce funding rounds, we're happy to say: this feature is already built. Head to app.traderalice.com and create your first backtest.

Final Words: You Are Absolutely Right!

TraderAlice is built by a team that genuinely loves trading and AI. Every time we ship a new feature, we carefully evaluate user needs—because we're users too.

But no matter how thorough our research and development, bugs and blind spots are inevitable.

So whatever feedback or ideas you have, tell us—You are absolutely right! We'll make improvements before you can't stand it anymore.

Again, come fall down the rabbit hole with TraderAlice.