> ## Documentation Index
> Fetch the complete documentation index at: https://docs.everstrike.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Strategy Lab: Turn an Idea Into Tested Prompt Logic

> Describe a strategy idea, and the Strategy Lab turns it into testable prompt logic, backtests every revision on historical candles and saves your ideas.

The [Strategy Lab](https://app.everstrike.io/app/strategy-lab) turns an idea into testable prompt logic, lets you refine it in conversation, and measures every revision against historical candles.

## How it works

**1. Describe the idea**

Write an entry, an exit, or simply the behaviour you want. For example: "Buy BTC when momentum turns positive, trail the position with ATR, and exit when the trend breaks."

**2. The lab writes and tests it**

The lab writes the strategy as prompt logic, then backtests it. It keeps running if you leave the lab; the result is waiting when you come back.

**3. Refine it in conversation**

Ask for changes such as "make entries stricter", "add an RSI filter" or "reduce drawdown". Each accepted revision is tested automatically.

**4. Create the live strategy**

When you're happy with it, choose **Create live strategy**.

## Editing directly

The **Prompt logic** is editable. After you stop typing, it's re-tested automatically.

You can also change the test settings: **Market**, **Candles**, **Default qty**, **Starting capital**, **Max leverage**, **Fee** and **Slippage** (in basis points), **Confidence** and the start date (**From**).

## Results

Each test shows **Net return**, **Ending equity**, **Max drawdown**, **Trades**, **Win rate**, **Sharpe**, **Sortino** and **Profit factor**, with the **equity curve** and the **drawdown path**.

**Prompt coverage** shows how much of the prompt the backtest could test: parts **included**, **partial** and **unsupported**. See [Reading a Fast Estimate](https://docs.everstrike.io/ai/reading-fast-estimates.md).

> [!NOTE]
>
> Lab backtests are Fast Estimates: candle-based approximations, not replays of live AI decisions. They run in your browser and don't use AI credits.

## Past ideas and saved runs

Ideas are saved once their prompt logic has been generated, under **Past ideas**, with their return and trade count. Every run is kept under **Saved runs**.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.