AFL is not just a scripting language; it is a vector-oriented analysis tool that allows you to test decades of data in milliseconds. Whether you are coding a simple moving average crossover or a complex neural network, understanding is the skill that separates profitable quants from perpetual tinkerers.
RSI_14 = RSI(14); VolumeSurge = V > MA(V, 50) * 1.5; Filter = RSI_14 < 30 AND VolumeSurge; AddColumn(C, "Close", 1.2); AddColumn(RSI_14, "RSI", 1.2); AddColumn(V, "Volume", 1.0); Run this on 5,000 stocks. AmiBroker will return a list of only those meeting the criteria. Exploration allows you to output historical values into a grid. amibroker afl code
// --- Exploration for Equity Curve Analysis --- SetBarsRequired(500, 0); Equity = Foreign("~~~EQUITY", "C"); // Internal equity array MonthlyReturn = (Equity - Ref(Equity, -20)) / Ref(Equity, -20) * 100; Filter = 1; AddColumn(MonthlyReturn, "20-Period Return %", 2.2); AddColumn(Stdev(MonthlyReturn, 20), "Volatility", 2.2); Backtesting is where AFL truly shines. The default settings are good, but professional optimization requires custom metrics. 4.1 Custom Backtest Interface (CBI) You can override AmiBroker’s core logic using SetCustomBacktestProc . AFL is not just a scripting language; it
// --- Entry Conditions --- BuySignal = C < BBLower AND C > TrendMA; // Price below lower band but above 200 MA Buy = ExRem(BuySignal, SellSignal); // Remove consecutive buy signals AmiBroker will return a list of only those
// Add custom metrics stats = bo.GetPerformanceStats(0); // 0 = Long positions maxDD = stats.GetValue("Max system % drawdown");
// --- Real-time Position Tracker --- staticVar = Nz(StaticVarGet("MyPosition"), 0); currentPos = staticVar; if (Buy AND currentPos == 0) { currentPos = 1; StaticVarSet("MyPosition", 1); printf("Buy executed at " + WriteVal(C)); }
SetCustomBacktestProc(""); if (Status("action") == actionPortfolio) { bo = GetBacktestObject(); bo.Backtest(); // Run standard backtest first
