Pick the tutorial closest to where you are.
The best way to learn Dome is by doing something specific. Each tutorial has a clear goal, a realistic time estimate, and a defined next step when you're done.
Just installed. Start here.
These three tutorials cover the first things worth doing — not the most impressive things. Get your first chart up, set your risk limits before you need them, and run AI on a live prompt so you know what to expect.
Your First Session, Start to Finish
Open a chart, check the data, ask ORACLE one real question about what you're seeing, log a journal entry while the trade is open, and review the summary. That's it. One complete loop.
- Open Trade Desk and load an instrument you actually trade
- Check the data source — is it live or cached?
- Read what Quant Brain says about the current structure
- Write your first ORACLE prompt with chart context attached
- Log a journal entry with your reason and how you felt about the trade
- Close the session and read the summary
Set Your Risk Limits Before You Trade
The most important setup in the terminal. Do this before your first real session. A daily loss limit you haven't set isn't a limit — it's a number you'll negotiate with yourself later, when you're already down.
- Set your daily loss limit in Session Management — a real number, not aspirational
- Configure the soft warning threshold so you get a heads-up before you're at the wall
- Use the position size calculator with your actual risk percentage
- Create an exposure alert in Alert Manager so you're notified before you're overextended
- Trigger a simulated breach to confirm the alerts actually fire
- Get comfortable with where the risk panel lives before you need it under pressure
Run Your First Local AI — No Cloud Required
Download a small model, point Dome at it, and ask it something real about your chart. Completely offline. No API key. Your data stays on your machine. Takes about 10 minutes once the model downloads.
- Open Local LLM Lab and pick a model that fits your machine
- Download a 1.5B GGUF model — small enough to be fast, capable enough to be useful
- Set the model path and configure the context window size
- Ask it a real question about a chart you're looking at right now
- Compare the local model response to what ORACLE gives you on the same question
- Understand when to use local vs. cloud AI for different tasks
You've had a few sessions. Now make them repeatable.
These tutorials are for traders who've used the terminal at least a handful of times and want to stop improvising. Build a consistent pre-session routine, write a real indicator, and deploy your first paper bot.
Build a Pre-Session Routine You Actually Follow
Most sessions go wrong before the first trade — because you opened the chart with no context and jumped in on feel. This tutorial builds a repeatable routine: market state, news filter, journal review, then decide.
- Create a pre-session checklist you complete before touching the chart
- Configure Market Intel to filter only what's relevant to your watchlist
- Build a Quant Brain read sequence across your key timeframes
- Connect last session's journal review to your pre-session steps
- Set a pre-session reminder alert so the routine actually happens
- Save it as a template so next time takes 5 minutes, not 45
Write an Indicator That Does What You Actually Want
You've been waiting for a chart overlay that matches your setup. Stop waiting — this tutorial takes you from a blank script to a working indicator on your chart. Specific goal: EMA crossover signal, drawn on the chart, with an alert attached.
- Open the Script Editor — this is where you'll spend most of your time
- Write a simple EMA crossover signal in Dome Script syntax
- Add adjustable inputs so you can tune it without editing the code
- Draw it on the chart using plotmark and plotline
- Attach an alert so you're notified when the signal fires
- Add it to Trade Desk and verify it's showing what you expect
Deploy a Paper Bot and See How It Actually Behaves
A good backtest is a starting point, not a guarantee. This tutorial takes you through the full paper deployment: convert your script to a strategy, set the risk gates, activate paper mode, and read the log after 5 real signals. Does it behave like you expected?
- Convert your indicator script to a strategy script with entry/exit rules
- Run the validator and fix any warnings before deployment
- Configure Bot Manager in paper mode — no real orders yet
- Set your risk gates: daily loss ceiling, max positions, size cap
- Activate the bot and watch the first signals fire in real conditions
- Read the trade log after 5 signals and compare to your backtest assumptions
Already using the terminal regularly? Go deeper.
These tutorials are for traders who've been through the intermediate material and want to build serious research pipelines, run multiple bots safely, or get more out of ORACLE's Committee deliberation. You'll need to be comfortable with Python for Tutorial 07.
Find Out If Your Strategy Has Real Edge
This is the research pipeline that separates real edge from curve-fitting. You'll build a walk-forward validation from scratch: in-sample optimization, out-of-sample testing, Monte Carlo stress test, and PBO scoring. By the end you'll know whether your strategy has a hypothesis worth trading — or just a beautiful backtest chart.
- Load OHLCV data directly from the Dome Python data API — no CSV exports
- Define your in-sample and out-of-sample window sizes
- Optimize parameters on the in-sample window only
- Test on the subsequent out-of-sample window and roll forward
- Run a Monte Carlo permutation to stress-test the robustness
- Compute PBO score — does the probability of overfitting put you off live trading?
Run Multiple Bots Without Losing Control of Your Risk
Running one bot is manageable. Running three or four — on correlated instruments, in the same session — is where portfolio-level risk exposure can quietly spiral. This tutorial sets up the gates that keep all bots inside a shared risk budget.
- Configure a shared portfolio risk gate that covers all active bots
- Set per-bot and aggregate daily loss limits that don't double-count
- Build a Dome Script risk arbitration layer that pauses lower-priority bots first
- Simulate a correlated drawdown and verify the gate fires correctly
- Review the multi-bot exposure panel in Session Management during a live session
- Set alerts when two bots are trading the same direction on correlated instruments
Get More Out of ORACLE's Committee
Most traders use ORACLE like a chat window. This tutorial shows you a more useful approach: structured multi-round deliberation, adversarial mode where specialists argue against each other, and how to read specialist disagreement as a signal to reduce size — not pick a side.
- Learn what each of the eight specialists actually looks for — and where they're blind
- Structure a deliberation prompt sequence that gets you more than one generic response
- Activate adversarial mode: specialists argue against each other's conclusions
- When specialists strongly disagree, that's not noise — it's uncertainty. Learn to read it that way
- Read the Committee chain-of-thought log to see what they were actually reasoning from
- Use the audit trail after a trade to see whether the Committee was right — and why
The full tutorial library is in the docs.
Video walkthroughs, code samples, troubleshooting notes, and additional guides live in the Dome Terminal documentation. Download first, then start with Tutorial 01 — 30 minutes and you'll know the layout.