Strategy Development

A strategy is a named, versioned set of trading rules stored as a JSON definition. The platform evaluates your conditions against real-time indicator values and generates orders automatically — no code required to run a strategy.

---

What Strategies Are

Every strategy is a JSON document that describes:

• Which pairs to trade — one or more USDT pairs (e.g. BTCUSDT, ETHUSDT)
• What timeframe — the candle interval used to compute indicators
• Entry conditions — rules that must ALL be true before entering a position
• Exit conditions — rules where ANY one triggers a position close
• Position sizing — how much equity to allocate per position
• Position limit — how many positions to hold simultaneously

The platform's StrategyExecutor evaluates these rules against live indicator values computed by the IndicatorEngine (pure numpy — no external TA-Lib dependency). When entry conditions align, it places buy orders. When any exit condition fires, it places sell orders and updates trailing stop levels.

---

Strategy Lifecycle

draft → testing → validated → deployed → archived
                                  ↓
                              undeploy → validated

Status	Meaning
draft	Just created — not yet tested
testing	A test run is in progress
validated	At least one test run has completed
deployed	Actively evaluating conditions and placing orders
archived	Soft-deleted — no longer active

A strategy moves from draft to testing automatically when you start a test run. After the test completes it becomes validated. You can only deploy from validated. To archive a deployed strategy, undeploy it first.

---

Strategy Definition Schema

{
  "pairs": ["BTCUSDT", "ETHUSDT"],
  "timeframe": "1h",
  "entry_conditions": {
    "rsi_below": 30,
    "macd_cross_above": true,
    "adx_above": 25
  },
  "exit_conditions": {
    "stop_loss_pct": 3,
    "take_profit_pct": 8,
    "trailing_stop_pct": 2,
    "max_hold_candles": 48
  },
  "position_size_pct": 10,
  "max_positions": 3
}

Field	Type	Required	Description
pairs	string[]	Yes	One or more USDT trading pair symbols
timeframe	string	Yes	Candle interval: 1m, 5m, 15m, 1h, 4h, or 1d
entry_conditions	object	Yes	All non-null conditions must be true to enter
exit_conditions	object	Yes	Any condition triggers exit
position_size_pct	number	Yes	Percent of equity to allocate per position (1–100)
max_positions	number	Yes	Maximum simultaneous open positions (1–50)

---

A Complete Example

This strategy buys BTC and ETH when RSI is oversold, the MACD has just turned bullish, and the trend is strong. It exits via a stop-loss, take-profit, or if the position stagnates for more than 48 candles.

{
  "pairs": ["BTCUSDT", "ETHUSDT"],
  "timeframe": "1h",
  "entry_conditions": {
    "rsi_below": 30,
    "macd_cross_above": true,
    "adx_above": 25,
    "volume_above_ma": 1.3
  },
  "exit_conditions": {
    "stop_loss_pct": 3,
    "take_profit_pct": 8,
    "trailing_stop_pct": 2,
    "max_hold_candles": 48
  },
  "position_size_pct": 10,
  "max_positions": 3
}

Create this via REST:

curl -X POST http://localhost:8000/api/v1/strategies \
  -H "Authorization: Bearer $JWT" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "RSI Oversold + MACD",
    "description": "Buy oversold dips with trend confirmation",
    "definition": { ... }
  }'

Response:

{
  "strategy_id": "strat_abc123",
  "name": "RSI Oversold + MACD",
  "status": "draft",
  "current_version": 1,
  "created_at": "2026-03-01T10:00:00Z"
}

---

Entry Conditions — AND Logic

All 12 entry condition keys use AND logic. Every non-null condition must evaluate to true for the entry signal to fire. Set a key to null (or omit it) to ignore it.

See the Indicators page for a detailed breakdown of each condition, what it measures, and example values.

---

Exit Conditions — OR Logic

Any single exit condition firing closes the position immediately. The priority order when multiple conditions trigger on the same candle is:

stop_loss → take_profit → trailing_stop → max_hold_candles → indicator exits

This means a stop-loss always takes priority over a take-profit if both conditions are met on the same candle (this should not happen in practice but defines the tiebreak).

See the Indicators page for all 7 exit condition keys.

---

Versioning

Each PUT to /strategies/{id}/versions creates a new immutable version:

from agentexchange import AgentExchangeClient

client = AgentExchangeClient(api_key="ak_live_...")

v2 = client.create_version(
    strategy_id="strat_abc123",
    definition={
        "pairs": ["BTCUSDT", "ETHUSDT"],
        "timeframe": "1h",
        "entry_conditions": {"rsi_below": 25, "macd_cross_above": True, "adx_above": 28},
        "exit_conditions": {"stop_loss_pct": 2, "take_profit_pct": 6, "trailing_stop_pct": 1.5},
        "position_size_pct": 8,
        "max_positions": 3
    },
    change_notes="Tightened RSI threshold and stop-loss based on test results"
)
# v2["version"] == 2

Versions are never deleted. The current_version field on the strategy points to the latest version, but you can deploy any version by number.

---

REST API Reference

Method	Path	Description
POST	/api/v1/strategies	Create a strategy
GET	/api/v1/strategies	List strategies (paginated)
GET	/api/v1/strategies/{id}	Get detail, definition, and latest test results
PUT	/api/v1/strategies/{id}	Update name or description
DELETE	/api/v1/strategies/{id}	Archive (soft-delete)
POST	/api/v1/strategies/{id}/versions	Create a new version
GET	/api/v1/strategies/{id}/versions	List all versions
GET	/api/v1/strategies/{id}/versions/{v}	Get a specific version
POST	/api/v1/strategies/{id}/deploy	Deploy a version to live
POST	/api/v1/strategies/{id}/undeploy	Stop live execution

All endpoints require Authorization: Bearer {jwt} or X-API-Key.

---

Next Steps

• Built-in Indicators — all 7 indicators, conditions, and example values
• Strategy Testing — multi-episode testing and the recommendation engine
• Deploying Strategies — versioning, monitoring, and rematches