Lose/Win Percentage for a Good Trading System

[pkfryer]

I'm evaluating various trading systems and experimenting with a few approaches. I'm going through historic data and trying them out one bar at a time (end-of-day).

I know that some systems work best in trading ranges and others in trends, so taking that into account, what would you say is a good score for successful signals coming from a system. I had a nose around some of these 'black box' services like (Goldline and SpiderTrade etc) and was shocked to discover that there signals were only slightly better than 50%. I always assumed a good system would have to have 70+% correct signals. So is this wrong?

With tight money/risk management can you succeed with a system with less than 60% correct calls. With the systems you guys use or have developed, what is the win/lose percentage?

Is the most important factor the profit/loss percentage per trade rather than the win/lose? So if I gain 3 times more profit than I lose I could succeed with a trading system that gives less than 1/3 correct signals. What are you experiences?

 

[TheBramble]

With tight money/risk management can you succeed with a system with less than 60% correct calls.

[...]

Is the most important factor the profit/loss percentage per trade rather than the win/lose?

PKF - you have already pinpointed it - money management. Some traders claim anything less than a 60% win/loss ratio is not worth trading. Others point to the importance of how much you win and how much you lose. Both have their point.

However. Even with a 50/50 win/loss ratio - 'tight money/risk management' will keep you in the game.

 

[JonnyT]

I have yet to find a system that is sustaintainable with a win ration above 40%

You really do have to have large winners and small losers IMHO

JonnyT

 

[Mayfly]

PKF,

Take a look at the attached spreadsheet, it should help to explain the points about money management the others are making?

HTH

Cheers

Mayfly

 

[elliottmillion]

a 50/50 trading system really is the best you can hope for. anyone who claims to have invented a system 70+ is fibbing!

 

[Two Bars]

anyone who claims to have invented a system 70+ is fibbing!

Not true my friend. Those who have a 70%+ method just keep it quiet. Wouldn't you?

 

[elliottmillion]

exactly true Two Bars, thats why anyone who claims to have invented a system 70% is fibbing!

 

[pkfryer]

Thanks mayfly for the figures on the spreadsheet, not 100% sure what I'm looking at, so just want to clarify. 20 trades 50/50 win/lose (10 winning trades, 10 losing trades). Value on the left (per row) is the losing percentage per trade, value along the top (per column) is the winning percentage? What does the LOSS (percent) along the top mean?

Anyway, it is very fascinating and thought provoking. If we are right only half the time we could be in profit just as long as the gain/lose ratio was in our favour, and how can we guarantee that favourable ratio? That is the value of a good system, yes?

I wonder if anyone has tried an experiment of just flipping a coin each day as to whether to go short or long, and see what the results are? I think it would be an interesting exercise to try. Combined with money management, could you still come out a winner?

 

[elliottmillion]

with a fifty fifty system, your winning bets should average out to at least 1.5 of your losses, so you make .5 profit. Pair the trades off with each other. Take only minimum risk trades, with a movement of 150% of what your stop.

The hardest part of this is to get a stop and stick to it.

 

[TheBramble]

Anyway, it is very fascinating and thought provoking. If we are right only half the time we could be in profit just as long as the gain/lose ratio was in our favour, and how can we guarantee that favourable ratio? That is the value of a good system, yes?

One such system might be to cut your losses, quickly (a small stoploss - rigidly adhered to) and let your profits run (a trailing stoploss - equally rigidly adhered to).

I wonder if anyone has tried an experiment of just flipping a coin each day as to whether to go short or long, and see what the results are? I think it would be an interesting exercise to try. Combined with money management, could you still come out a winner?

This has been covered in other threads and is effectively what the above spreadsheet is illustrating. It doesn't cover the 1:3 rather than 1:2 win/lose ratio (trend can go with you, against you or sideways [which is also against you]) which you would experience in that system, but it's close.

 

[Matt321]

There is a formula for testing your system out on past trades or test runs,Expected Net Profit/ Loss Formula
(E pl) = Pw(Aw) - Pl (A l )

Pw = Probability for win
Pl = Probability for loss
Aw = Average win
Al = Average loss

 

[theknifemac]

There is a formula for testing your system out on past trades or test runs,Expected Net Profit/ Loss Formula
(E pl) = Pw(Aw) - Pl (A l )

Pw = Probability for win
Aw = Probability for win
Pl = Average win
Al = Average loss

This is also known as the expectancy of the system - the amount you can expect in average to make on each trade. It gets round the problem of only looking at the percentage of wins which is meaningless without knowing the profit made on winners and losses made on losing trades. I think this came from Elder but am sure it appears in many other books too.

Stew

 

[TheBramble]

There is a formula for testing your system out on past trades or test runs,Expected Net Profit/ Loss Formula
(E pl) = Pw(Aw) - Pl (A l )

Pw = Probability for win
Aw = Probability for win
Pl = Average win
Al = Average loss

I think 'Aw' should be 'Probability for loss'.

 

[theknifemac]

I think 'Aw' should be 'Probability for loss'.

I think you mean Pl should be Probability for loss....

 

[pkfryer]

Is it
Pw = probability for win
Aw = Average Win
Pl = probability for loss
Al = Average loss
?

Very interesting. Do you use this formula when testing out your systems? How do you know what the probability for a win is? Is this based on the sum of all your win/lose trades when backtesting? e.g. if you win 4 times out of 10, then your probability is 4/10? So the equation will give you information on how reliable/profitable your system is after you have aquired statistics through backtesting.

Thats a very handy equation! Gives you some confidence in your system even when you have a small string of losses.

 

[TheBramble]

I think you mean Pl should be Probability for loss....

Ha....! Lord, yes.

There's nothing quite like correcting someone else and getting it even wronger yourself in the process.

So, Matt321's (adjusted) formula is:-

(E pl) = Pw(Aw) - Pl (A l )

Pw = Probability for win
Aw = Average win
Pl = Probability for win
Al = Average loss

{Slinks away into a dark corner to lick wounds and hope this passes without too much attention...}

 

[Matt321]

Have posted an example of the workings below



Trades
1 -4
2 4
3 6
4 -3
5 -2
6 4
7 5
8 3
9 -5
10 4
11 6
12 -5
13 -4
14 -5
15 4
16 -2
17 4
18 6
19 4
20 -5


Wins 11
Losses 9
Ratio 55


Average win 4.55
Average loss 3.88

Pw (Aw) - Pl (Al)

=0.55 x (4.55) - 0.45(3.88)
=2.50 - 1.75 = 0.75

Update:
If the last number turns out to be(ie -0.75) a minus then your system is losing,although the one above is positive.
There is no backtesting system I know of, you just get a calculator and your broker's statement.
The ratio is really a %,The way I did it was divide 20 trades into a 100= 5 then 11wins X 5 = 55 then 9 x 5 = 45 so thats your ratio or % of winners55%/losers45%.
You can also juggle figures around and play the "what if game" ie: if I had a closer stop
losses it would mean more losses but smaller ones etc the different combinations are endless,but it can help your tweak your system up.Try swopping the losers for winners
so you have 45% winners 55% losers the answer should be if i haven't got it wrong -2.15
bad news.

 

[Matt321]

Pw = probability win
Pl = probability loss
Aw = ave win
Al = ave loss

 

[pkfryer]

Okay.. that makes a lot of sense. Is there a backtesting program or module that calculates that for your systems?

Also is there some relation to the sharpe score?

The P and the A stands for Probability and Average, the w and l stand for win and lose. It has to.. that means your correction of your correction is still wrong :OD

 

[Matt321]

Yer its actually printed wrong/here is the final corrected version
Pw = prob win
Pl = prob loss
Aw = Average win
Al = Average loss
I stopped trading after applying this,I now looking for a good trading system
I might try the that trick of watching for the breakout on the currencies between 07.00 and 10.00.