Strategies
A strategy is your trading playbook — the complete set of rules that tells your agent what to look for in the market, when to trade, and how to manage risk. Think of it like written instructions you'd give to a human trader: "Watch these indicators, enter when these conditions are met, set stops here, exit like this."
You build strategies in the Strategy Builder, then assign them to an agent. The agent follows your playbook automatically — you decide when and where to deploy it.
Build a strategy in 5 steps
┌─────────────────────────────────────────────────────────────────────────┐
│ │
│ 1. INDICATORS → 2. RULES → 3. ENTRY → 4. LEVELS → 5. AGENT │
│ What to When to How to Risk & Extra │
│ watch trade get in reward guidance │
│ │
└─────────────────────────────────────────────────────────────────────────┘
Step 1: Pick your indicators
Indicators are the numbers your agent watches — RSI, moving averages, volume, volatility, and more. You organize them into panels in the Agent briefing, which is the data table your agent reads for each coin.
What you decide: Which indicators matter for your trading style? A momentum trader might watch RSI and MACD. A mean-reversion trader might focus on Bollinger Bands and oversold readings.
→ Indicators guide · Indicator examples by goal
Step 2: Set your rules
Rules are true/false tests that decide when your agent trades. They read your indicators and answer: Should we trade this coin? Which direction? When should we exit?
What you decide: What conditions must be true before entering? What's a filter (gates all trades) vs. a signal (picks direction)?
Step 3: Choose your entry
Entry controls when and how your agent gets into a trade after your rules approve it. Enter immediately at bar close, wait for a breakout, or place a limit order for a pullback.
What you decide: Do you want to chase momentum or wait for better prices?
Step 4: Set trade levels and exit
Trade levels control your risk and reward — minimum volatility, stop-loss band, and the reward-to-risk ratio you require.
Exit controls how your stop moves after you're in — break-even protection, trailing stops, and time decay for stalled trades.
What you decide: How tight or wide should stops be? How much reward do you need for the risk? How should profits be protected?
→ Trade levels guide · Exit guide
Step 5: Add agent instructions
Agent instructions are plain-language guidance your agent reads with every decision. They shape judgment — "prefer structure over short-term noise" — but don't enforce hard rules.
What you decide: What context or priorities should your agent consider that aren't captured in rules?
Start here if you're new
New to strategies? Follow this path:
| Step | What to do | Link |
|---|---|---|
| 1. Learn the vocabulary | Read the glossary so you know what each term means | Glossary |
| 2. Understand indicators | Learn how to add data to your Agent briefing | Indicators |
| 3. See real examples | Browse indicator setups organized by what you're trying to do | Indicator examples |
| 4. Build your first rule | Learn how rules control when your agent trades | Rules |
| 5. Troubleshoot | If your strategy isn't trading, check this list | "Why isn't my strategy trading?" |
What a strategy contains
A strategy has two layers:
Agent briefing — what your agent sees
| Component | What it does |
|---|---|
| Indicators | The numbers your agent reads — organized into panels |
| Rules | True/false tests over those indicators |
| Agent instructions | Plain-language guidance for your agent |
Policy — what the platform enforces
| Component | What it does |
|---|---|
| Entry | When and how to enter trades |
| Trade levels | Stop-loss band and minimum R:R |
| Exit | How stops move after entry |
Using your strategy
Once you've built a strategy:
- Assign it to an agent — Connect the strategy to an agent so it can trade
- Deploy the agent — Choose where it runs:
- Radar — Continuous coin monitoring
- Arena — Prediction game sessions
- Trade terminal — You approve each trade
Creating a strategy doesn't risk capital. Deploying the agent does.
Good to know
-
Strategies and agents are separate. The strategy is the playbook. The agent follows it. You can assign one strategy to multiple agents, or change which strategy an agent uses.
-
Changes apply on save. When you save a strategy, all agents using it pick up the changes on their next evaluation cycle.
-
Commander can help. You can describe changes in plain language to Commander, which stages edits into your strategy's draft for you to review and save.
-
Open trades keep their snapshot. If you change a strategy mid-trade, existing positions keep the rules they entered with until they close.
Advanced: Strategy identity and settings
Identity fields
| Field | What it does |
|---|---|
| Name, tagline, description | How you recognize the strategy in your library |
| Primary timeframe | The candle length your strategy trades on (e.g. 1h, 4h) |
| Regime timeframe | The shortest enabled timeframe at least three times your primary timeframe — set automatically |
| Cadence | Display badge: Scalper (≤15m), Intraday (30m–2h), Swing (4h–12h), Position (≥1d) |
| Active | Whether the strategy can be assigned to agents |
Timeframe constraints
- Primary timeframe must have both a lower and higher rung available (floor and ceiling of enabled timeframes are not available)
- Regime timeframe is always the shortest enabled timeframe at least three times the primary timeframe (a 30m strategy reads 2h, a 1h strategy reads 4h)
- Changing primary timeframe is a structural edit — the builder asks you to confirm
Preview cohort
The universe select in the builder controls which coins appear in your preview while building. This is separate from what your agent trades live:
| Mode | What it does |
|---|---|
| Ranked | Top coins by 24h volume (up to 36) |
| Explicit | Specific tickers you name (up to 36) |
Preview cohort ≠ live trading universe. The Agent briefing preview helps you test, but live evaluation uses the platform's active universe.
Version control
When you save, the platform checks you're editing the latest version. If someone else (or Commander) saved in between, your save is refused — reload and try again.
Advanced: Building via Commander
Commander lets you describe strategy changes in plain language:
- Stage — Commander writes proposed changes into your strategy's draft
- Review — See the changes in the builder before anything goes live
- Commit — Save when you're ready (or ask Commander to commit after you confirm)
If the strategy changed since Commander read it, the save is refused and Commander reads the latest version.
← Introduction · Next: Glossary · Indicators