Program Trading

Program Trading lets you define trading logic with code. When signals trigger, your strategy executes automatically without AI involvement.

Program Trading vs AI Trader

Feature	AI Trader	Program Trading
Decision	AI model analyzes and decides	Code logic executes directly
Flexibility	High, handles complex scenarios	Medium, follows preset rules
Predictability	Low, AI may judge differently	High, same conditions = same results
Speed	Slower (waits for AI response)	Fast (direct code execution)
Best for	Complex market analysis	Clear rule-based strategies

When to use which?

• AI Trader: When you need AI to analyze market conditions and handle ambiguous situations
• Program Trading: When you have clear trading rules and want fast, predictable execution

Workflow Overview

Market Data Collection → Signal Detection → Program Execution → Order Execution → Attribution

Core Concepts

1. Signals and Signal Pools

Signals define "what conditions are worth watching":

• Example: CVD exceeds 1M, OI growth > 5%, Funding Rate > 0.01%

Signal Pools combine multiple signals for more reliable triggers:

• AND mode: All signals must be satisfied (stricter)
• OR mode: Any signal triggers (more sensitive)

The system uses edge triggering: triggers only once when conditions are first met.

2. Market Regime Detection

The system automatically identifies current market state:

Regime	Meaning	Trading Suggestion
Breakout	Trend starting	Enter with trend
Continuation	Trend continuing	Hold or add
Exhaustion	Trend weakening	Consider taking profit
Absorption	Large orders absorbed	Wait and watch
Stop Hunt	Stop loss hunting	Wait for reversal
Trap	Bull/bear trap	Trade cautiously
Noise	No clear direction	Stay out

Available Data in Strategies

Account Info

Field	Description
available_balance	Available balance
total_equity	Total equity
positions	Current positions (dict by symbol)
open_orders	Pending orders
recent_trades	Recent closed trades

Trigger Info

Field	Description
trigger_symbol	Triggered symbol (e.g., BTC)
trigger_type	Trigger type (signal / scheduled)
signal_pool_name	Name of triggered signal pool
triggered_signals	List of triggered signal details
trigger_market_regime	Market regime at trigger time

Environment Config

Field	Description
environment	Environment (testnet / mainnet)
max_leverage	Maximum leverage limit
default_leverage	Default leverage

Available Functions

Get K-line Data

get_klines(symbol, period, count)

• Purpose: Get historical candlesticks
• Parameters: Symbol, period (5m/15m/1h/4h/1d), count
• Returns: List of candles (time, OHLCV)

Get Technical Indicators

get_indicator(symbol, indicator, period)

• Purpose: Get calculated technical indicators
• Parameters: Symbol, indicator name, period

Get Market Flow Data

get_flow(symbol, metric, period)

• Purpose: Get market flow metrics
• Parameters: Symbol, metric type, time window

Get Market Regime

get_regime(symbol, period)

• Purpose: Get current market state classification
• Returns: Regime type, confidence, direction, reason

Get Price

get_price(symbol)              # Get latest price
get_price_change(symbol, period)  # Get price change percentage

Built-in Technical Indicators

Moving Averages

Indicator	Description
MA5 / MA10 / MA20	Simple Moving Average
EMA20 / EMA50 / EMA100	Exponential Moving Average

Momentum Indicators

Indicator	Description
RSI7 / RSI14	Relative Strength Index
MACD	Moving Average Convergence Divergence (returns macd, signal, histogram)
STOCH	Stochastic Oscillator (returns K, D values)

Volatility Indicators

Indicator	Description
BOLL	Bollinger Bands (returns upper, middle, lower)
ATR14	Average True Range

Volume Indicators

Indicator	Description
OBV	On-Balance Volume
VWAP	Volume Weighted Average Price

Market Flow Metrics

Metric	Description	Use Case
CVD	Cumulative Volume Delta	Buy/sell pressure comparison
OI_DELTA	Open Interest change rate	Capital flow direction
TAKER	Taker buy/sell ratio	Market sentiment
DEPTH	Order book depth ratio	Support/resistance strength
IMBALANCE	Order book imbalance	Short-term direction
FUNDING	Funding rate	Long/short sentiment

Decision Output

Your strategy must return a Decision object:

Field	Required	Description
operation	Yes	Operation type: buy / sell / close / hold
symbol	Yes	Trading pair
reason	Yes	Decision reasoning
target_portion_of_balance	For buy/sell	Balance portion to use (0.1-1.0)
leverage	For buy/sell	Leverage multiplier
take_profit_price	No	Take profit price
stop_loss_price	No	Stop loss price

Step-by-Step Guide

Step 1: Configure Signals (Optional)

If you want signal-triggered programs, first create signals and signal pools in Signal Pools.

Common monitoring metrics:

• CVD: Cumulative Volume Delta - buy/sell pressure
• OI Change: Open Interest change - capital flow
• Funding Rate: Long/short sentiment indicator
• Order Book Depth: Buy/sell pressure comparison

Step 2: Create AI Trader Account

1. Go to Account Management page
2. Create new account, select "AI" type
3. Configure AI model and API Key
4. Enter Hyperliquid private key (use testnet first)
5. Enable "Auto Trading" switch

Step 3: Create Trading Program

1. Go to Program Management page
2. Click AI Code Assistant, describe your strategy in natural language
3. AI generates and tests code, outputs to editor when passed
4. Click Save

Step 4: Bind Program to Account

1. Select the AI Trader account to bind
2. Check the signal pools to trigger on
3. Set scheduled trigger interval (optional)
4. Click Save Binding

Backtesting

After binding, you can backtest your program to validate it.

Backtest Types

Type	Purpose	Description
Program Backtest	Validate strategy code	Simulate strategy with historical K-lines
Signal Backtest	Validate signal effectiveness	Check historical trigger patterns

How Backtesting Works

Load historical K-line data
    ↓
Iterate from candle #50 onwards
    ↓
Build MarketData for each candle (only sees current and past data)
    ↓
Execute strategy code, get Decision
    ↓
Simulate trade execution (calculate fees, update virtual positions)
    ↓
Output backtest results

Data Requirements

Backtesting requires historical data. If the system wasn't running during a period, there's no market flow data for that time, making backtest results inaccurate. Keep the system running 24/7 or deploy to a server.

Backtest Result Metrics

Metric	Description
Total Trades	Number of open/close trades
Win Rate	Percentage of profitable trades
Total PnL	Cumulative profit/loss
Max Drawdown	Maximum equity decline from peak
Sharpe Ratio	Risk-adjusted return
Equity Curve	Account value over time

Monitoring

View Execution Logs

Check detailed trigger records in the Signal System page:

View Positions and Assets

Monitor real-time positions, PnL, and asset curves in the Dashboard:

Best Practices

1. Test before live trading: Validate all strategies on testnet first
2. Control position size: Don't use too large a portion per trade
3. Use reasonable leverage: Choose leverage based on strategy characteristics
4. Regular review: Analyze strategy performance through execution logs
5. Diversify risk: Run multiple accounts with different strategies

Next Steps

• Dashboard & Monitoring - Monitor trading performance
• FAQ - Common questions and solutions