Español ES
Ver en GitHub

SVM Trader Strategy

Overview

SVM Trader Strategy demonstrates how a combination of classic technical indicators can approximate the behavior of a support vector machine (SVM) model for generating trading signals. The original MQL example trained two separate SVMs for buy and sell decisions. In this StockSharp conversion, we emulate the decision process with a simple scoring system derived from seven indicators:

  • Bears Power and Bulls Power – measure the balance between sellers and buyers.
  • Average True Range (ATR) – captures current volatility.
  • Momentum – checks price acceleration.
  • Moving Average Convergence Divergence (MACD) – identifies trend direction.
  • Stochastic Oscillator – detects overbought and oversold levels.
  • Force Index – combines price movement and volume.

Each indicator contributes to a cumulative score. When the score exceeds a threshold, the strategy opens a long position; when the score falls below the opposite threshold, a short position is opened. This setup mirrors the classification step of the original SVM approach while keeping the implementation lightweight and transparent.

Parameters

Name Description
CandleType Candle timeframe used for calculations.
Volume Order volume for new trades.
TakeProfit Distance for take-profit in absolute price units.
StopLoss Distance for stop-loss in absolute price units.
RiskExposure Maximum cumulative position volume allowed.

Trading Logic

  1. Subscribe to candles of the specified type and bind all indicators using high-level API.
  2. For each finished candle, retrieve indicator values from the binding callback.
  3. Calculate a score:
    • Bulls Power greater than Bears Power
    • Momentum above zero
    • MACD line above its signal line
    • Stochastic %K above %D
    • Force Index above zero
  4. If at least three conditions are true and the current position is non-positive, a market buy order is placed.
  5. If two or fewer conditions are true and the current position is non-negative, a market sell order is placed.
  6. StartProtection applies both stop-loss and take-profit for every opened position.

Notes

  • Indicator periods are fixed to values from the original MQL example (mostly 13 for symmetry and smoothness).
  • The scoring system is a simplified proxy for SVM classification and can be replaced with a more advanced model if required.
  • RiskExposure prevents over-allocation by limiting the total position size.
  • The strategy uses tabs for indentation and English comments as required by project conventions.

Disclaimer

This strategy is provided for educational purposes. It demonstrates indicator binding and basic risk management in StockSharp. Use and modify at your own risk.

using System;
using System.Collections.Generic;

using Ecng.Common;
using Ecng.Serialization;

using StockSharp.Algo.Indicators;
using StockSharp.Algo.Strategies;
using StockSharp.BusinessEntities;
using StockSharp.Messages;

namespace StockSharp.Samples.Strategies;

/// <summary>
/// SVM Trader Strategy using multiple indicators to approximate SVM classification.
/// </summary>
public class SvmTraderStrategy : Strategy
{
	private readonly StrategyParam<DataType> _candleType;
	private readonly StrategyParam<decimal> _takeProfit;
	private readonly StrategyParam<decimal> _stopLoss;
	private readonly StrategyParam<decimal> _riskExposure;
	private readonly StrategyParam<int> _buyThreshold;
	private readonly StrategyParam<int> _sellThreshold;
	private readonly StrategyParam<int> _cooldownBars;

	private BearPower _bears;
	private BullPower _bulls;
	private Momentum _momentum;
	private MovingAverageConvergenceDivergenceSignal _macd;
	private StochasticOscillator _stochastic;
	private ForceIndex _force;
	private int _previousScore;
	private bool _hasPreviousScore;
	private int _barsSinceTrade;

	/// <summary>
	/// Candle type for processing.
	/// </summary>
	public DataType CandleType
	{
		get => _candleType.Value;
		set => _candleType.Value = value;
	}

	/// <summary>
	/// Take profit in price units.
	/// </summary>
	public decimal TakeProfit
	{
		get => _takeProfit.Value;
		set => _takeProfit.Value = value;
	}

	/// <summary>
	/// Stop loss in price units.
	/// </summary>
	public decimal StopLoss
	{
		get => _stopLoss.Value;
		set => _stopLoss.Value = value;
	}

	/// <summary>
	/// Maximum allowed cumulative position volume.
	/// </summary>
	public decimal RiskExposure
	{
		get => _riskExposure.Value;
		set => _riskExposure.Value = value;
	}

	/// <summary>
	/// Minimum score required for a long signal.
	/// </summary>
	public int BuyThreshold
	{
		get => _buyThreshold.Value;
		set => _buyThreshold.Value = value;
	}

	/// <summary>
	/// Maximum score allowed for a short signal.
	/// </summary>
	public int SellThreshold
	{
		get => _sellThreshold.Value;
		set => _sellThreshold.Value = value;
	}

	/// <summary>
	/// Bars to wait after a position is closed.
	/// </summary>
	public int CooldownBars
	{
		get => _cooldownBars.Value;
		set => _cooldownBars.Value = value;
	}

	/// <summary>
	/// Initializes a new instance of the <see cref="SvmTraderStrategy"/> class.
	/// </summary>
	public SvmTraderStrategy()
	{
		_candleType = Param(nameof(CandleType), TimeSpan.FromHours(4).TimeFrame())
			.SetDisplay("Candle Type", "Type of candles", "General");

		_takeProfit = Param(nameof(TakeProfit), 1400m)
			.SetDisplay("Take Profit", "Take profit in price units", "Risk");

		_stopLoss = Param(nameof(StopLoss), 900m)
			.SetDisplay("Stop Loss", "Stop loss in price units", "Risk");

		_riskExposure = Param(nameof(RiskExposure), 1m)
			.SetDisplay("Risk Exposure", "Max cumulative position", "Risk");

		_buyThreshold = Param(nameof(BuyThreshold), 4)
			.SetDisplay("Buy Threshold", "Score required for a long signal", "Signal");

		_sellThreshold = Param(nameof(SellThreshold), 1)
			.SetDisplay("Sell Threshold", "Score required for a short signal", "Signal");

		_cooldownBars = Param(nameof(CooldownBars), 2)
			.SetDisplay("Cooldown Bars", "Bars to wait after a completed trade", "Risk");
	}

	/// <inheritdoc />
	public override IEnumerable<(Security sec, DataType dt)> GetWorkingSecurities()
	{
		return [(Security, CandleType)];
	}

	/// <inheritdoc />
	protected override void OnReseted()
	{
		base.OnReseted();

		_previousScore = 0;
		_hasPreviousScore = false;
		_barsSinceTrade = CooldownBars;
	}

	/// <inheritdoc />
	protected override void OnStarted2(DateTime time)
	{
		base.OnStarted2(time);

		_bears = new BearPower { Length = 13 };
		_bulls = new BullPower { Length = 13 };
		_momentum = new Momentum { Length = 13 };
		_macd = new MovingAverageConvergenceDivergenceSignal();
		_stochastic = new StochasticOscillator();
		_force = new ForceIndex { Length = 13 };

		var subscription = SubscribeCandles(CandleType);
		subscription
			.BindEx(_bears, _bulls, _momentum, _macd, _stochastic, _force, ProcessIndicators)
			.Start();

		StartProtection(
			new Unit(TakeProfit, UnitTypes.Absolute),
			new Unit(StopLoss, UnitTypes.Absolute));

		var area = CreateChartArea();
		if (area != null)
		{
			DrawCandles(area, subscription);
			DrawOwnTrades(area);
		}
	}

	private void ProcessIndicators(
		ICandleMessage candle,
		IIndicatorValue bearsValue,
		IIndicatorValue bullsValue,
		IIndicatorValue momentumValue,
		IIndicatorValue macdValue,
		IIndicatorValue stochasticValue,
		IIndicatorValue forceValue)
	{
		if (candle.State != CandleStates.Finished ||
			!bearsValue.IsFinal || !bullsValue.IsFinal || !momentumValue.IsFinal ||
			!macdValue.IsFinal || !stochasticValue.IsFinal || !forceValue.IsFinal)
			return;

		if (!IsFormedAndOnlineAndAllowTrading())
			return;

		if (_barsSinceTrade < CooldownBars)
			_barsSinceTrade++;

		var score = 0;
		var bears = bearsValue.ToDecimal();
		var bulls = bullsValue.ToDecimal();
		var momentum = momentumValue.ToDecimal();
		var macdTyped = (MovingAverageConvergenceDivergenceSignalValue)macdValue;
		var stochasticTyped = (StochasticOscillatorValue)stochasticValue;
		var force = forceValue.ToDecimal();

		if (bulls > bears)
			score++;
		if (momentum > 100m)
			score++;
		if (macdTyped.Macd > macdTyped.Signal)
			score++;
		if (stochasticTyped.K > stochasticTyped.D && stochasticTyped.K > 55m)
			score++;
		if (force > 0m)
			score++;

		if (!_hasPreviousScore)
		{
			_previousScore = score;
			_hasPreviousScore = true;
			return;
		}

		var longSignal = _previousScore < BuyThreshold && score >= BuyThreshold;
		var shortSignal = _previousScore > SellThreshold && score <= SellThreshold;
		var openVolume = Math.Abs(Position);

		if (_barsSinceTrade >= CooldownBars && openVolume + Volume <= RiskExposure)
		{
			if (longSignal && Position <= 0)
			{
				BuyMarket(Volume + (Position < 0 ? -Position : 0m));
				_barsSinceTrade = 0;
			}
			else if (shortSignal && Position >= 0)
			{
				SellMarket(Volume + (Position > 0 ? Position : 0m));
				_barsSinceTrade = 0;
			}
		}

		_previousScore = score;
	}
}