Live & alt-data bricks

The capability model

Bricks don’t hard-code a broker. Instead each live brick declares the capability it needs — a plain string like options or account — and TradeBricks matches that against whatever providers you have connected. Connect a provider that supplies the capability and every brick needing it turns on at once; connect nothing and the built-in bricks (Congress, SEC) still work because their data is public.

This keeps strategies portable: a graph that uses option_greeks declares it needs options, not “Schwab”. Anyone who forks it and has any options-capable provider can run it unchanged.

fundamentals is a forward-looking capability: Schwab supplies it and the gating model is ready, but no brick consumes it yet — today’s alt-data bricks are options-, Congress-, and SEC-driven. It’s listed for completeness so the capability model stays honest about what is and isn’t wired.

Choosing a provider

• Schwab is the most customizable — it supplies the richest data surface: option chains and greeks, dealer gamma, IV, options flow, fundamentals, plus a brokerage account. If you want the alt-data bricks at full strength, connect Schwab.
• Topstep is account-only: it gives the account capability (equity, positions, prop-firm rule tracking) but no options or fundamentals. Ideal if you trade a Topstep evaluation and just need live account numbers.
• Congress and SEC need no connection at all — the data is public, so those bricks are always available, even signed out of any broker.

See Connecting providers for the step-by-step connect flow, exactly what each provider unlocks, and the keys-stay-on-your-device guarantee — and The Trader for the account dashboard a connection powers.

The alt-data bricks

Below are the live / alternative-data bricks. The engine isn’t reachable right now, so this is the built-in reference list — it fills in live as soon as the engine is up. Each shows its outputs, the capability it requires, and its tunable params.

Example: auto-follow Congress

A classic alt-data strategy is to mirror Congressional buying. The wiring is three bricks: a congress_trades source, an entry that fires on its bought signal, and a barrier exit that defines the trade’s risk. Because Congress is built in, this runs with no broker connected — though you’ll need a connected account to actually place the orders the Watcher proposes.

congress_trades            # source: STOCK Act disclosures for the symbol
  params: chamber=both, member=any, min_amount=50000
  outputs: bought (signal), sold (signal), net (series)

      bought ──▶ enter_when      # entry: open a position on disclosed buys

                 entry ──▶ barrier_exit   # risk: stop / target / horizon
                            params: stop=2.0%, target=4.0%, horizon=20

In the Lab this backtests neutrally — historical disclosure timing is noisy and lagged, so treat the backtest as a sanity check on the wiring, not proof of an edge. The value of this strategy is realized live: the Watcher subscribes to new filings and acts the moment one lands.

The Watcher (always-on)

Backtests in the Lab are historical: they replay the past to estimate whether a graph has an edge. They do not place trades. To act on live and alt-data feeds in real time, TradeBricks runs the Watcher — an always-running state machine that subscribes to your connected data feeds and triggers your strategy’s actions live.

How it runs

1. You promote a strategy that survived the Lab to the Watcher and choose which symbols it covers.
2. The Watcher subscribes to exactly the feeds your graph needs — the capabilities its bricks require, from your connected providers.
3. As each bar, filing, or flow event arrives it advances the strategy’s state machine and evaluates your entry / exit logic against the live data.
4. When a condition fires it produces an action — and acts only through your own live-trading worker, which reads your encrypted broker credential from the server-side vault.

Live vs. the Lab — be honest

A clean backtest is necessary but not sufficient. Live execution adds slippage, latency, partial fills, data revisions (disclosures and filings get amended), and the simple fact that the future isn’t the past. Treat the Watcher as the place a researched idea meets reality — and size accordingly. A strategy that looked great on the holdout can still lose money live.