Strategy input formats

The terminal accepts three formats. You can mix them freely — the AI reads all three.

Natural language

The simplest way. Describe what you want in plain English:

Buy ETH when the 15-minute RSI drops below 28. Sell when price is up 4%
or RSI goes above 72. Use 8% of my portfolio.

Works well for strategies with 1–3 conditions. The more specific you are, the less editing you'll need to do afterward.

Tips:

• Name the market: BTC/USDT, ETH/USD, or just BTC. If the market is ambiguous, we'll ask.
• Mention the timeframe: on the 4h, daily candles. Defaults to 1h if absent.
• Use numbers: RSI below 30, price up 5%, EMA(20). Vague terms like "low RSI" are less reliable.

Python pseudocode

# SMA crossover with trailing stop
entry_condition = sma_50 < sma_200  # "death cross" - short signal
exit_condition = trailing_stop(2.5) or profit >= 3
 
market = "BTC/USDT"
interval = "1h"
capital = 5  # % of portfolio
side = "short"

Python pseudocode reads identically to natural language for simple expressions. It's more reliable for:

• Multi-condition entries with and / or
• Indicator parameters: ema(close, 20) vs ema(20) vs EMA(20) all parse the same
• Comments as annotations — the AI reads them

Pine Script (TradingView)

Paste a Pine Script strategy or study. The AI extracts indicator parameters, entry/exit logic, and alert conditions. It ignores anything that doesn't apply to Coinrule (e.g. strategy.entry labels, position sizing in lots, etc.).

//@version=5
indicator("EMA Cross", overlay=true)
fastLen = input.int(9, "Fast EMA")
slowLen = input.int(21, "Slow EMA")
fast = ta.ema(close, fastLen)
slow = ta.ema(close, slowLen)
longEntry = ta.crossover(fast, slow)
longExit  = ta.crossunder(fast, slow)

What the AI cannot parse

• Cross-exchange arbitrage (e.g. "buy on Binance when price is 0.3% below Coinbase")
• Market-making / order book depth strategies
• Strategies that require streaming tick data below the 1-minute interval
• Strategies referencing custom data sources you host yourself

If your prompt includes unsupported logic, the AI will tell you what it dropped. See Supported vs unsupported strategies.