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

# Your first Alchemy run

> Before you spend any credits, open a saved sample and follow it from the first click to the finished result, then try a live run with a question of your own.

Reading about Alchemy only goes so far; the quickest way to get a feel for it is to watch a run. Samples are real runs saved earlier, so they open in seconds, need no sign-in and spend no AI credits. Start by finding one.

## Where the samples are

The Alchemy page shows two lists under the event box.

* **Example chips.** A row of example events. The ones tagged **Try me** are samples. Hover one and it says "Sample run: opens instantly, no credits used".
* **Hot queries.** A short list under the **Hot queries** switch, labelled "What people are running". The first row, **Impact of AI Bubble crash**, is also a sample.

![Alchemy example chips, with JD Vance wins the 2028 election, Super El Niño this winter and Market impact due to increase in cyber hacks tagged Try me](https://mintcdn.com/openfinance/-pu5SodoQuhEwuR-/images/alchemy/alchemy-try-me-chips.png?fit=max\&auto=format\&n=-pu5SodoQuhEwuR-\&q=85\&s=f4f8cec4dcba3f947bbc90269d9742b3)

Example chips: the ones tagged Try me open a saved sample

> [!WARNING]
>
> Chips and Hot queries rows without a sample behind them start a real run. Signed out, that asks you to sign in. Signed in, it spends AI credits. Stick to the **Try me** chips and **Impact of AI Bubble crash** while you explore.

## Walk through a sample

With a sample picked out, here is the whole run, from the first click to the basket at the bottom.

**1. Open Alchemy**

Click **Alchemy** under **Trade Agent** in the sidebar, or go to openfinance.tech/alchemy.

**2. Pick a sample**

Click **Impact of AI Bubble crash** under Hot queries, or any chip tagged **Try me**. The event box folds into a slim bar at the top with your question in it.

**3. Watch the steps**

A checklist appears and ticks through the same eight steps a live run takes. A sample replays them in a few seconds. A live run shows a clock and the note "Can take 3–4 minutes".

**4. Read the summary**

The first card names the headline Polymarket question and its odds, then eight number tiles and the scenarios. See [Reading a result](https://docs.openfinance.tech/alchemy/reading-results.md) for each one.

**5. Read the Alchemy take**

The second card is Alchemy's call in its own words, with **Detailed analysis** on the right and three view tabs at the bottom.

**6. Switch views**

Click **Trade YES**, **Trade NO** or **Alchemy's pick**. Everything below the take follows the view you choose.

**7. Look at the basket**

**Basket trades** lists the trades for that view, sized to a \$10,000 budget by default. Switch between **Cards** and **List**.

![Alchemy replaying a sample run, with the summary card loading at the top and a checklist of steps below, the first four ticked and the fifth in progress](https://mintcdn.com/openfinance/-pu5SodoQuhEwuR-/images/alchemy/alchemy-run-steps.png?fit=max\&auto=format\&n=-pu5SodoQuhEwuR-\&q=85\&s=201a0c64852e3af85250834a463df5ea)

A run in progress: the summary fills in while the steps tick down

## The eight steps

The checklist you watched in step 3 names each stage of the work. This is what Alchemy does in each one.

| Step | What Alchemy is doing |
| - | - |
| Understanding the question | Reads your event and decides what it is really asking |
| Researching | Gathers context on the event |
| Comparing prediction markets | Searches Polymarket and picks the market that prices it |
| Finding the driver variables | Lists the forces that move the outcome, such as rates or AI valuations |
| Making connections across industries | Links each driver to the industries it reaches |
| Generating second-order effects | Traces the knock-on effects one step further |
| Estimating exposure across stocks, macro and crypto | Finds the instruments that carry each effect |
| Building the trades and the AI pick | Sizes the trades and writes Alchemy's own pick |

The summary card fills in early, as soon as Alchemy has found the market. You can read the market's odds while the rest is still working.

## What you just saw

![Alchemy result for the AI bubble sample showing the Polymarket question and its odds, eight stat tiles, the Scenarios and odds ladder, related markets and key dates](https://mintcdn.com/openfinance/-pu5SodoQuhEwuR-/images/alchemy/alchemy-result-summary.png?fit=max\&auto=format\&n=-pu5SodoQuhEwuR-\&q=85\&s=2289da1dec9d8114da24bed805ee054c)

The top of a result: the market's answer, Alchemy's numbers and the scenarios

In the AI bubble sample, the market prices the headline question low and Alchemy's own odds sit higher, so the **Edge** tile is positive and **Alchemy's pick** leans Yes. That gap between the two numbers is the whole thesis. The basket then shows how Alchemy would express it: a Yes share on Polymarket, short perps on the companies the event hits hardest, and a few hedges.

> [!NOTE]
>
> Numbers in a sample are frozen at the time it was captured, shown at the bottom of the summary card. Charts inside the trade cards still load live prices. A sample also shows a **Replay** button, which plays the steps again.

## After the sample

* Click **Start over** in the top bar to clear the result and return to the empty box.
* Signed in, the sample is saved under **My history**, so you can reopen it later.
* When you are ready, type your own event. [Writing good events](https://docs.openfinance.tech/alchemy/writing-good-events.md) shows how to phrase it.

## Your first live run

When you want an answer to a question of your own, a live run takes four steps.

**1. Sign in**

A live run spends your AI credits, so Alchemy asks you to sign in first.

**2. Type one event**

One clear outcome with a date works best, for example "Fed cuts rates in December".

**3. Pick a model, or keep the default**

The model picker sits next to the send button. See [History, models and credits](https://docs.openfinance.tech/alchemy/history-and-models.md).

**4. Send and wait**

Press Enter or the arrow. The square button stops the run. You can move to another page of the app and come back; the run keeps going in that browser tab and shows in the sidebar's **Recents**.

> [!TIP]
>
> Closing or refreshing the tab during a live run stops it, so the browser asks you first. Asking a second question while one is running opens the new one in a new tab.

## Next

> [card] [Reading a result](https://docs.openfinance.tech/alchemy/reading-results.md)
>
> What each number and section means.

> [card] [Writing good events](https://docs.openfinance.tech/alchemy/writing-good-events.md)
>
> Example events you can paste.


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