日本語 JA
GitHub で見る

Williams R Mean Reversion Strategy

Williams %R oscillates between 0 and -100 to show when price closes near the extremes of its recent range. This strategy fades those extremes once the indicator stretches far from its own average.

Testing indicates an average annual return of about 154%. It performs best in the stocks market.

A long trade triggers when Williams %R falls below the average minus DeviationMultiplier times the standard deviation. A short trade is taken when it rises above the average plus that multiplier. Exits occur when Williams %R moves back toward its average level.

The approach suits traders who rely on momentum exhaustion to time entries. A protective stop-loss limits risk if price keeps moving to new extremes.

Details

  • Entry Criteria:
    • Long: %R < Avg - DeviationMultiplier * StdDev
    • Short: %R > Avg + DeviationMultiplier * StdDev
  • Long/Short: Both sides.
  • Exit Criteria:
    • Long: Exit when %R > Avg
    • Short: Exit when %R < Avg
  • Stops: Yes, percent stop-loss.
  • Default Values:
    • WilliamsRPeriod = 14
    • AveragePeriod = 20
    • DeviationMultiplier = 2m
    • CandleType = TimeSpan.FromMinutes(5)
  • Filters:
    • Category: Mean Reversion
    • Direction: Both
    • Indicators: Williams %R
    • Stops: Yes
    • Complexity: Intermediate
    • Timeframe: Intraday
    • Seasonality: No
    • Neural networks: No
    • Divergence: No
    • Risk Level: Medium
namespace StockSharp.Samples.Strategies;

using System;
using System.Collections.Generic;

using Ecng.Common;

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

/// <summary>
/// Williams %R Mean Reversion strategy.
/// This strategy enters positions when Williams %R is significantly below or above its average value.
/// </summary>
public class WilliamsRMeanReversionStrategy : Strategy
{
	private readonly StrategyParam<int> _williamsRPeriod;
	private readonly StrategyParam<int> _averagePeriod;
	private readonly StrategyParam<decimal> _deviationMultiplier;
	private readonly StrategyParam<DataType> _candleType;
	private readonly StrategyParam<decimal> _stopLossPercent;

	private decimal _prevWilliamsR;
	private decimal _avgWilliamsR;
	private decimal _stdDevWilliamsR;
	private decimal _sumWilliamsR;
	private decimal _sumSquaresWilliamsR;
	private int _count;
	private readonly Queue<decimal> _williamsRValues = [];

	/// <summary>
	/// Williams %R Period.
	/// </summary>
	public int WilliamsRPeriod
	{
		get => _williamsRPeriod.Value;
		set => _williamsRPeriod.Value = value;
	}

	/// <summary>
	/// Period for calculating mean and standard deviation of Williams %R.
	/// </summary>
	public int AveragePeriod
	{
		get => _averagePeriod.Value;
		set => _averagePeriod.Value = value;
	}

	/// <summary>
	/// Deviation multiplier for entry signals.
	/// </summary>
	public decimal DeviationMultiplier
	{
		get => _deviationMultiplier.Value;
		set => _deviationMultiplier.Value = value;
	}

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

	/// <summary>
	/// Stop-loss percentage.
	/// </summary>
	public decimal StopLossPercent
	{
		get => _stopLossPercent.Value;
		set => _stopLossPercent.Value = value;
	}

	/// <summary>
	/// Constructor.
	/// </summary>
	public WilliamsRMeanReversionStrategy()
	{
		_williamsRPeriod = Param(nameof(WilliamsRPeriod), 14)
			.SetGreaterThanZero()
			
			.SetOptimize(7, 21, 7)
			.SetDisplay("Williams %R Period", "Period for Williams %R indicator", "Indicators");

		_averagePeriod = Param(nameof(AveragePeriod), 20)
			.SetGreaterThanZero()
			
			.SetOptimize(10, 50, 10)
			.SetDisplay("Average Period", "Period for calculating Williams %R average and standard deviation", "Settings");

		_deviationMultiplier = Param(nameof(DeviationMultiplier), 2m)
			.SetGreaterThanZero()
			
			.SetOptimize(1.5m, 3m, 0.5m)
			.SetDisplay("Deviation Multiplier", "Multiplier for standard deviation", "Settings");

		_candleType = Param(nameof(CandleType), TimeSpan.FromMinutes(5).TimeFrame())
			.SetDisplay("Candle Type", "Type of candles to use", "General");

		_stopLossPercent = Param(nameof(StopLossPercent), 2m)
			.SetGreaterThanZero()
			
			.SetOptimize(1m, 3m, 0.5m)
			.SetDisplay("Stop Loss %", "Stop loss as percentage of entry price", "Risk Management");
	}

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

	/// <inheritdoc />
	protected override void OnReseted()
	{
		base.OnReseted();
		_prevWilliamsR = 0;
		_avgWilliamsR = 0;
		_stdDevWilliamsR = 0;
		_sumWilliamsR = 0;
		_sumSquaresWilliamsR = 0;
		_count = 0;
		_williamsRValues.Clear();
	}

	/// <inheritdoc />
	protected override void OnStarted2(DateTime time)
	{
		// Reset variables

		// Create Williams %R indicator
		var williamsR = new WilliamsR { Length = WilliamsRPeriod };

		// Create subscription and bind indicator
		var subscription = SubscribeCandles(CandleType);
		subscription
			.BindEx(williamsR, ProcessCandle)
			.Start();

		// Setup chart visualization
		var area = CreateChartArea();
		if (area != null)
		{
			DrawCandles(area, subscription);
			DrawIndicator(area, williamsR);
			DrawOwnTrades(area);
		}

		// Enable position protection
		StartProtection(
			takeProfit: new Unit(0m), // We'll manage exits ourselves based on Williams %R
			stopLoss: new Unit(StopLossPercent, UnitTypes.Percent)
		);

		base.OnStarted2(time);
	}

	private void ProcessCandle(ICandleMessage candle, IIndicatorValue williamsRValue)
	{
		// Skip unfinished candles
		if (candle.State != CandleStates.Finished)
			return;

		// Check if strategy is ready to trade
		if (!IsFormedAndOnlineAndAllowTrading())
			return;

		// Extract Williams %R value
		var currentWilliamsR = williamsRValue.ToDecimal();

		// Update Williams %R statistics
		UpdateWilliamsRStatistics(currentWilliamsR);

		// Save current Williams %R for next iteration
		_prevWilliamsR = currentWilliamsR;

		// If we don't have enough data yet for statistics
		if (_count < AveragePeriod)
			return;

		// Check for entry conditions
		if (Position == 0)
		{
			// Long entry - Williams %R is significantly below its average
			if (currentWilliamsR < _avgWilliamsR - DeviationMultiplier * _stdDevWilliamsR)
			{
				BuyMarket(Volume);
				LogInfo($"Long entry: Williams %R = {currentWilliamsR}, Avg = {_avgWilliamsR}, StdDev = {_stdDevWilliamsR}");
			}
			// Short entry - Williams %R is significantly above its average
			else if (currentWilliamsR > _avgWilliamsR + DeviationMultiplier * _stdDevWilliamsR)
			{
				SellMarket(Volume);
				LogInfo($"Short entry: Williams %R = {currentWilliamsR}, Avg = {_avgWilliamsR}, StdDev = {_stdDevWilliamsR}");
			}
		}
		// Check for exit conditions
		else if (Position > 0) // Long position
		{
			if (currentWilliamsR > _avgWilliamsR)
			{
				ClosePosition();
				LogInfo($"Long exit: Williams %R = {currentWilliamsR}, Avg = {_avgWilliamsR}");
			}
		}
		else if (Position < 0) // Short position
		{
			if (currentWilliamsR < _avgWilliamsR)
			{
				ClosePosition();
				LogInfo($"Short exit: Williams %R = {currentWilliamsR}, Avg = {_avgWilliamsR}");
			}
		}
	}

	private void UpdateWilliamsRStatistics(decimal currentWilliamsR)
	{
		// Add current value to the queue
		_williamsRValues.Enqueue(currentWilliamsR);
		_sumWilliamsR += currentWilliamsR;
		_sumSquaresWilliamsR += currentWilliamsR * currentWilliamsR;
		_count++;

		// If queue is larger than period, remove oldest value
		if (_williamsRValues.Count > AveragePeriod)
		{
			var oldestWilliamsR = _williamsRValues.Dequeue();
			_sumWilliamsR -= oldestWilliamsR;
			_sumSquaresWilliamsR -= oldestWilliamsR * oldestWilliamsR;
			_count--;
		}

		// Calculate average and standard deviation
		if (_count > 0)
		{
			_avgWilliamsR = _sumWilliamsR / _count;
			
			if (_count > 1)
			{
				var variance = (_sumSquaresWilliamsR - (_sumWilliamsR * _sumWilliamsR) / _count) / (_count - 1);
				_stdDevWilliamsR = variance <= 0 ? 0 : (decimal)Math.Sqrt((double)variance);
			}
			else
			{
				_stdDevWilliamsR = 0;
			}
		}
	}
}