AI Essentials is free. Paid enrollment opens soon.
Written practical guide
Downloads and references for this free lesson. Library resources may require course access.
Choose Track A: an alert assistant if you want a watchlist event, a source-backed explanation and a notification you can review. Choose Track B: paper automation if you also want to turn an approved event into a simulated order and reconcile its lifecycle. Both tracks share the data, testing and reliability lessons. Track A can finish without a broker account.
The system is a pipeline: completed market data → deterministic signal → optional AI evidence annotation → risk policy → review → notification or paper order → audit log. The model does not invent prices, decide position sizes, receive account secrets or bypass a stop condition. An explanation can fail while the underlying event remains recorded.
Write down one instrument universe, one timeframe, one signal rule, one event expiry, one notification destination and one operator. Choose a deliberately small problem: “A completed bar crosses a moving-average threshold; create one reviewable alert with the bar timestamp and a link to evidence.” A rule being easy to implement says nothing about whether it makes money.
Our included DEMO instrument is fictional. The prices, risk amounts and output figures are classroom fixtures. The Python starter runs offline and creates order previews only; the course explains the additional adapter, state and operational work for a broker paper environment. No lesson asks you to fund an account or enables live-money execution.
A green test proves the behavior that test checks. A positive backtest is a historical result under assumptions. A paper fill is a simulator result. None establishes future returns. AI can also fabricate confident financial explanations; independently review its sources. The linked joint investor alert explains why guaranteed AI trading returns are a red flag.
Create README.md with your chosen track, user, inputs, outputs, failure behavior and evidence required to pass. Include: “When the same event is delivered five times, exactly one action intent exists. When data is stale or the kill switch is active, no new paper intent is created.” This observable statement is the course's starting point.
An API key identifies a client and grants particular access; it is not an AI model, a strategy, or a subscription to every market. Sketch this division before creating keys:
| Connection | What it contributes | What it must not do in this design |
|---|---|---|
| Twelve Data or a broker data feed | Quotes, completed candles and instrument metadata | Decide your risk policy or place orders merely because it supplied a price |
Explore the full curriculum, projects, and access details.
| eToro Demo or Alpaca Paper | Account state, order status and optional simulated execution | Receive unvalidated model commands |
| Optional AI provider | Explain an already recorded event from selected evidence | Receive broker secrets, invent current prices or size an order |
| Notification provider | Deliver an alert to your private test destination | Treat delivery acknowledgement as reviewer approval |
| Your application | Normalize inputs, apply rules, authorize actions and keep the ledger | Confuse an API response with a verified outcome |
You can build Track A with a data source and notifications, adding AI only after the event pipeline works. Track B adds a broker adapter. An eToro API exists and is documented publicly; access to a particular account, product or permission still depends on provider eligibility. The lessons demonstrate Twelve Data, eToro and Alpaca as examples, without requiring you to buy a plan or open all three accounts.
A useful design is Twelve Data bars → your signal → evidence annotation → private alert. A second design adds eToro Demo or Alpaca Paper after approval. If the data provider and broker differ, map instrument, exchange, asset type and currency explicitly. The same ticker text is not proof of the same tradable contract. Keep the first build on fixtures while you complete the provider worksheet.