Trading by win rate and profit factor

tradercn

#1

3/7/2012 3:11 PM

From Performance Visualizer, just had a sudden thought, but I was stuck at how to get the win rate and profit factor in the past performance of any strategy? For example,based on the strategy, Scalper QQQ & Stocks, I want to trade when average win rate (of past 100 trades) is lower than the average win rate (of past 10 trades), and profit factor is higher than the average profit factor (of past 30 trades). Is there any example strategy for that? Thanks.

Eugene

#2

3/7/2012 3:22 PM

Looks like you and another user (sedelstein) are heading in the same direction:

Access to data/stats normally displayed in the performance tab during a backtest

Basic Question about Position Sizers

Eugene

#3

3/8/2012 4:43 AM

Here's something for you to start with. What this example code does:

1. Loads the pre-built "Moving Average Crossover" Strategy, executing it with default parameters
2. Calculates the win rate of the past 10 closed trades and the win rate of the past 100 closed trades
3. Takes only trades that satisfy this condition: average win rate (of past 100 trades) is lower than the average win rate (of past 10 trades).

Size:

Color:

Here's something for you to start with. What this example code does:

1. Loads the pre-built "Moving Average Crossover" Strategy, executing it with default parameters
2. Calculates the win rate of the past 10 closed trades and the win rate of the past 100 closed trades
3. Takes only trades that satisfy this condition: average win rate (of past 100 trades) is lower than the average win rate (of past 10 trades).

[CODE]using System;
using System.Collections.Generic;
using System.Text;
using System.Drawing;
using WealthLab;
using WealthLab.Indicators;
using Community.Components;

namespace WealthLab.Strategies
{
	public class WinRateTrader : WealthScript
	{
		protected override void Execute()
		{
			// Specify Strategy name:
			string myStrategyName = "Moving Average Crossover";
			
			#region Build equity curve and other performance data
			
			Utility u = new Utility(this);
			string firstRun = ""; string results = "";
			SetGlobal( firstRun, Bars.Symbol );

if( (string)GetGlobal( firstRun ) == DataSetSymbols[0] ) 
			{
				RemoveGlobal( results + "_PerformanceResults" );
				
				/* Option 2: Load an existing Strategy from its XML file on your disk */
				WealthScript myStrategy = u.LoadStrategyFromDisk( myStrategyName );
				SystemPerformance sp = u.runDonor( myStrategy );
				SetGlobal( results + "_PerformanceResults", sp.Results );
			}			
			
			// Retrieve and plot the portfolio equity from Wealth-Lab's global memory
			
			SystemResults sr = (SystemResults)GetGlobal( results + "_PerformanceResults" );
			DataSeries globalEquity = sr.EquityCurve; globalEquity.Description = myStrategyName + " Portfolio Equity";
			globalEquity = Synchronize( globalEquity );
			ChartPane gEqPane = CreatePane( 40, false, true );
			PlotSeries( gEqPane, globalEquity, Color.DarkGreen, LineStyle.Histogram, 2 );
			HideVolume();
			
			#endregion

int NumberOfTrades1 = 10;
			int NumberOfTrades2 = 100;
			
			// Trading
			for(int bar = Bars.FirstActualBar; bar < Bars.Count; bar++)
			{
				// Build a list of closed positions
				int closedTrades = sr.Positions.Count;			
				List<Position> ClosedPositions = new List<Position>();
				foreach( Position p in sr.Positions )
				{
					if( p.EntryDate <= Bars.Date[bar] && p.ExitDate <= Bars.Date[bar] )
						ClosedPositions.Add(p);
				}

if (ClosedPositions.Count >= Math.Max(NumberOfTrades1,NumberOfTrades2))
				{
					int wins1 = 0; int wins2 = 0;
					double winRate1 = 0; double winRate2 = 0;

// Calculcate the win rate
					for (int p = ClosedPositions.Count - 1; p >= ClosedPositions.Count - NumberOfTrades1; p--)
					{
						if (ClosedPositions[p].NetProfitPercent > 0)
							wins1++;
					}

if (wins1 > 0)
						winRate1 = (wins1 * 100) / (double)NumberOfTrades1;
				
					for (int p = ClosedPositions.Count - 1; p >= ClosedPositions.Count - NumberOfTrades2; p--)
					{
						if (ClosedPositions[p].NetProfitPercent > 0)
							wins2++;
					}

if (wins2 > 0)
						winRate2 = (wins2 * 100) / (double)NumberOfTrades2;
						
					foreach (Position p in sr.Positions )
					{	
						// Entries
						if(( p.EntryDate == Bars.Date[bar] ) & ( p.Bars.Symbol == Bars.Symbol ))
						{
							// Insert your win rate / profit factor rules here
							if( winRate1 > winRate2 )						
							{
								Position pos = u.entry( p, bar, p.EntrySignal, p.BasisPrice );
								if( pos != null ) 
								{
									pos.Priority = p.Priority;
									pos.EntrySignal = p.EntrySignal;
								}
							}
						}
						
						// Exits
						if( ActivePositions.Count > 0 )
						{
							for( int pos = ActivePositions.Count - 1; pos > -1 ; pos-- )
							{
								Position ap = ActivePositions[pos];
								if(( p.ExitDate == Bars.Date[bar] ) & ( p.Bars.Symbol == Bars.Symbol ))
									if( p.EntryDate == ap.EntryDate )
										if( p.Priority == ap.Priority )
											u.exit( p, ap, bar, p.ExitSignal, p.ExitPrice ); 
							}
						}
					}
				}				
			}
		}
	}
}[/CODE]

Eugene

#4

3/8/2012 6:43 AM

P.S.

Although not exactly what you're looking for, the closest thing is to use either the built-in "Percent Winners" sizing or the "Percent Winners (% Equity)" PosSizer. Available instantly without any coding.

sedelstein

#5

3/22/2012 7:02 PM

Hi Eugene

As you surmised, this is interesting to me.

Do you see any issue with taking the portion of code that generates the closed positions to date

and generating the win/loss percentage for each symbol in the dataset?

put that section below the if {} statements and adding

to the part that generates the list of trades

So take AAPL's trades if the win rate for AAPL is low but not IBM's if the win rate for IBM was high.

Would it take an extraordinarily long time to run or any other issues?

sedelstein

#6

3/22/2012 9:11 PM

Update: It works for small datasets. Large datasets just take forever (and then some). Any thoughts?

Size:

Color:

Update:  It works for small datasets.  Large datasets just take forever (and then some).  Any thoughts?

[CODE]using System;
using System.Collections.Generic;
using System.Text;
using System.Drawing;
using WealthLab;
using WealthLab.Indicators;
using Community.Components;

namespace WealthLab.Strategies
{
	public class WinRateTrader : WealthScript
	{
		protected override void Execute()
		{
			// Specify Strategy name:
			string myStrategyName = "Moving Average Crossover";
			
			#region Build equity curve and other performance data
			
			Utility u = new Utility(this);
			string firstRun = ""; string results = "";
			SetGlobal( firstRun, Bars.Symbol );

if( (string)GetGlobal( firstRun ) == DataSetSymbols[0] ) 
			{
				RemoveGlobal( results + "_PerformanceResults" );
				
				/* Option 2: Load an existing Strategy from its XML file on your disk */
				WealthScript myStrategy = u.LoadStrategyFromDisk( myStrategyName );
				SystemPerformance sp = u.runDonor( myStrategy );
				SetGlobal( results + "_PerformanceResults", sp.Results );
			}			
			
			// Retrieve and plot the portfolio equity from Wealth-Lab's global memory
			
			SystemResults sr = (SystemResults)GetGlobal( results + "_PerformanceResults" );
			
			#endregion

// Trading

for(int bar = Bars.FirstActualBar; bar < Bars.Count; bar++)
			{
				// Build a list of closed positions for this symbol as of this bar
				int closedTrades = sr.Positions.Count;			
				List<Position> ClosedPositions = new List<Position>();
				foreach( Position p in sr.Positions )
				{
					if( p.EntryDate <= Bars.Date[bar] && p.ExitDate <= Bars.Date[bar] && p.Symbol == Bars.Symbol )
						ClosedPositions.Add(p);
				
				
				}
				//	PrintDebug (Bars.Symbol  + "  Positions as of " + String.Format( "{0:d}", Bars.Date[bar]) + " " + ClosedPositions.Count );

}
				PrintDebug("All done");			
		}
	}
}[/CODE]

Eugene

#7

3/23/2012 12:04 AM

Notes and Limitations

sedelstein

#8

4/5/2012 10:31 PM

I'll probably get in trouble with this but here goes

I want to use this technique above to run a strategy, call it ParentStrategy, which calls ChildStrategy via runDonor to get the equity curve, calculate some statistics based on it to decide whether to take the trades as of a date and perhaps even use SetShareSize to size the trade.

If I want to collect the results from the ParentStrategy, say via the performance engine in MS123 Scorecard, it is possible to create a GrandParent Strategy, and use runDonor to execute the ParentStrategy. Can runDonor in a sense be nested?

thanks for your help

Steve

Eugene

#9

4/6/2012 12:23 AM

I see what you're after but you probably will not get what you looking for. Any visualizer works with a "snapshot" of a Strategy's performance. Consequenntly, the PerformanceEngine class can't be used to calculate some statistics while looping by SystemResults.Positions. It's not going to work. Consider asking further questions on runDonor in other threads started by you, please.

In essence, you're trying to push the limits of Combination Strategies (even though you're not using them for this task). But it would be great if support for conditional execution existed. So a better idea is to call your Fidelity rep and vote for an enhancement like that as a Combination Strategy feature:

Combination Strategy Impressions, Suggestions thread, Eugene 6/22/2011 4:45 AM - it's #3 on my list

sedelstein

#10

4/6/2012 5:17 AM

Thanks Eugene

Conditional Execution would be great to have and not just in combination strategy.

I appreciate your input. I'll probably try coding it and see if it blows up. Worst case is a reboot I guess. ;->)

Eugene

#11

6/20/2012 5:28 AM

Important note to the users of the PerformanceEngine class (internal, unsupported part of the MS123 Visualizers library):

The information regarding PerformanceEngine calling conventions provided across the forum before is actual as of the library version 2012.05.1 and earlier. In future versions, the syntax will most likely undergo a breaking change due to a complete rewrite of the PerformanceEngine class.

innertrader

#12

2/25/2020 1:25 AM

Hi Eugene,

I have been working to repurpose the code from Post #3 above, so I trying to make sure I understand how it works. One thing is confusing me. In the code snippet

What's the reason you put the foreach inside of the for loop? It seems that you are recreating the identical ClosedPositions list with each iteration. Shouldn't this code go before the for loop? Is there something I don't understand about the necessity for it to be where it is?

Robert

Eugene

#13

2/25/2020 5:26 AM

Right.

No. The list of closed positions must be recreated on every bar by its nature.

superticker

#14

2/25/2020 6:14 PM

That's only because the code you cite in Post# 12 doesn't match the original code in Post# 6. Let's look at the code in Post# 6 again.

Notice the if statement includes p.Symbol == Bars.Symbol, which is omitted in Post# 12. That condition means only closed positions relevant to the current stock selected by the for loop are getting saved into ClosedPositions. If you omitted that condition, then you could place the foreach block outside the for loop.

If your goal is to create a sorted listing of ClosedPositions by stock symbol, you could certainly sort the sr.Positions object instead to avoid the for loop. My "wild guess" is that the default order of the sr.Positions object is already sorted by symbol since they "may be" created that way, but I could be wrong. If that's the case, then you can just skip the sort step and for loop altogether.

Eugene

#15

2/25/2020 6:58 PM

The topic deals with trading by {win rate, profit factor etc.} exclusively. This task implies calculating a rolling ratio based on closed positions on a bar by bar basis. For this reason, nested loops are required in post #3. Much like the Closed equity curve, it's not possible to avoid the for loop in this context.

UPDATE 02/26/2020

Loosely related discussion from this and other topics has been moved into a new thread:

Export a summary of performance stats by symbol programmatically