A hedge-fund Squad is a team of AI analysts you create, which debates a potential trade the way a real investment committee would — bulls and bears arguing each side, and managers weighing the argument — and publishes a decision you can act on.
It exists to answer one question your indicators cannot: given everything — the technicals, the fundamentals, the news, the mood — is this actually a good trade right now? And because one Squad can review many strategies, you build a view you trust once and apply it everywhere.
Squads are called committees in the API, the CLI, the MCP tools and in URLs. That naming is stable for automation — scripts, routes and tool names keep working unchanged.
Four angles, a bull and a bear on each#
A Squad argues a trade from four analytical domains, and every domain has both a bull and a bear, so no view goes unchallenged.
The eight specialists are Technical, Fundamentals, News and Sentiment, each as a bull and a bear. They post their theses, then cross-examine each other — bears attacking the bulls' case, bulls rebutting. Two managers then take over: the Research Manager synthesises the debate, and the Portfolio Manager makes the call. The step-by-step is in how a Squad deliberates.
One committee, many strategies#
This is the part that makes Squads worth building. A Squad is not tied to a single strategy. You create one — say a macro risk Squad, or a news-driven one — point it at the markets it should cover, and any strategy trading those markets can read its decisions. Every one of them gets the same disciplined second opinion, and the Squad is scored once, on its own calls, rather than separately inside each strategy.
You do the mapping from a strategy's Squad tab. See using a Squad in a strategy for the how and the signal-triggered option.
A committee never trades#
This is the important thing to understand about them: a Squad has no ability to open, close or size a position, and it does not ask you to approve one. It wakes on its schedule, deliberates, and records a decision — long, short, hold, or one of two risk-off states — stamped with the market price at that moment.
What happens next is entirely your strategy's business. A strategy that has added the Squad's indicator sees that decision as a line and applies its own rules to it, including whether to trade at all.
Because each decision is graded against what the market did next, a Squad accumulates a score you can read before you decide to follow it. See cadence, decisions and the track record.
What you need to run one — a model is included#
A Squad thinks by calling a language model, and one comes with the platform. There is a built-in default model, so a Squad works out of the box with nothing to connect and nothing extra to pay for.
When you want a stronger or specific model, you connect your own — either your own provider key or a model you host yourself. That is the trade: the included model is free and instant; the best models are yours to bring.
Start on the included model. It is enough to see how a Squad argues and to build a first paper track record. Upgrade to your own model once you know a Squad is worth the spend.
Connecting your own model, routing different jobs to different models, and capping daily spend are all in connecting AI models.
Getting started#
- Open Squads in the sidebar and create one with the guided wizard — start from a template (each presets real config: indicator readings, bias lines, cadence, context window and suggested markets) or build one from scratch. Template squads start deliberating on their cadence right away unless you turn the toggle off in the wizard.
- Make sure a model provider is connected.
- In its Config tab, fine-tune per market — everything a template presets stays editable there.
- Run a deliberation from a market's badge and read the argument in Chat.
- Let it build a record. Watch the score on its Performance tab.
- Once you trust it, add the Committee Decision indicator to a strategy — see using a Squad in a strategy.