From idea to execution.

Signal contract

A strategy receives a pandas DataFrame indexed by time and returns buy/sell boolean series or signal events. The same strategy object is used for backtesting and execution.

def on_bar(df, params):
    close = df["close"]
    fast = close.ewm(span=params["fast"]).mean()
    slow = close.ewm(span=params["slow"]).mean()
    buy = (fast > slow) & (fast.shift(1) <= slow.shift(1))
    sell = (fast < slow) & (fast.shift(1) >= slow.shift(1))
    return {"buy": buy, "sell": sell}

Execution modes

• Market — immediate execution at the next available price.
• Maker-rest — place a limit order at a configured offset and wait for a pullback.
• Maker-then-market — attempt maker execution, then fall back to market within the configured window.

Risk configuration

• Stop-loss and take-profit as price percentages
• Trailing activation and trailing distance
• Breakeven trigger
• Circuit breaker after abnormal volatility
• Per-strategy exposure and leverage limits

Backtest assumptions

Backtests include commissions, slippage, and optional funding. Results are simulations; live behavior differs with liquidity, latency, partial fills, and exchange outages. Always compare paper results with backtest expectations before increasing size.

Paper-to-live checklist

• Strategy has a documented edge and defined invalidation
• Backtest includes realistic costs and drawdown
• Paper trading matches the expected trade frequency
• Stop-loss, trailing, and max-loss limits are tested
• Exchange keys are trade-only and IP-restricted
• Monitoring and an independent exit plan are in place