Where to start

Two routes. Pick the one that matches why you're here; they meet in the middle.

"I want to test and trade my ideas"

You have a strategy, or an idea for one, and you want to know whether it survives honest testing before it sees money.

flowchart TD
    A["What you can do"] --> B["Install"]
    B --> C["Connect an account"]
    C --> D["Write a ruleset"]
    D --> E["Run a study"]
    E --> F["Read the verdict"]
    F --> G["Paper trade it"]
    G --> H["Go live"]
    H --> I["Explain a position"]

    style F fill:#1f6feb,stroke:#1f6feb,color:#fff

1	What you can do with Arvo — the seven things it's for
2	Install the window and the engine
3	Connect an account — Alpaca for bars and paper trading
4	Write a ruleset — your idea, and the parameters to try
5	Run a study and read the verdict — the step that matters
6	Start a paper session on a finding that passed
7	When a session freezes and halt trading
8	Explain a position — why do I own this?

The one page to read properly is step 5. Everything else is mechanics; that page is where you learn to read what Arvo is telling you, including when it tells you it doesn't know.

"I want to build on it"

You want to drive Arvo from code, an editor or an agent — or extend it.

flowchart TD
    A["Why Arvo is different"] --> B["Run it headless"]
    B --> C["Configure"]
    C --> D["Command line"]
    D --> E["Python, or an agent"]
    E --> F["The API"]
    F --> G["Extend it"]
    G --> H["Architecture"]

    style B fill:#1f6feb,stroke:#1f6feb,color:#fff

1	Why Arvo is different — the thesis the code is built around
2	The engine on its own — no window needed; it's a complete tool
3	Configure — the project folder, the handshake, the two tokens
4	Command line and engine examples
5	Research from Python, or with Claude Code
6	The API and the packages
7	Extensions and writing a provider
8	Architecture and the decisions behind it

The one thing to understand early is the two-token split in step 3: an agent or a script holds a token that reaches research and nothing else, so it cannot place an order or touch a credential. Most of the design follows from that.

Where they meet

Both paths need the same three ideas, and they're in Core concepts: what a rule and a ruleset are, what makes a result out-of-sample, and why the number of things you tried changes what the winner has to beat.

If you read nothing else, read Why Arvo is different — it's short, and it's the reason the rest is shaped this way.