Risk/reward ratio

[andrew100]

Hi,

Happy new everyone.

Can I get comments for the values below:

Risk / Reward ratio 100:16
Win / Loss ratio 71%:29%

Is there anything wrong with this system? lots of small wins and less large loses.

Is there anything I should check before I trade this system?

Regards,

Andrew

 

[brty]

Re: risk / reward ratio

Risk / Reward ratio 100:16
Win / Loss ratio 71%:29%

Is there anything wrong with this system? lots of small wins and less large loses.

Andrew, the way you have written this sounds like that for 29 losses of 100, say 2900, you have 71 winners of 16, say 1136. If making 1136 means losing 2900 to get there is your idea of a good system, then can I suggest going back to the drawing board.

brty

 

[Wysiwyg]

I'm wondering how the results of one test are indicative of a tradable system. I too can get a similar test result in a bull market.

 

[andrew100]

brty: hehehe, yeah you're right.

The back testing shows that most of my loses are not full loses (below 100) and also because of the win/loss ratio, the result shows I make a profit.

As long as I don't have lots of 100 loses in a row, I should be right???

Wysiwyg: I have tested with historical data on the market that I am planning to trade, is there other test I can perform on this system to confirm?

Andrew

 

[Wysiwyg]

brty: hehehe, yeah you're right.

Wysiwyg: I have tested with historical data on the market that I am planning to trade, is there other test I can perform on this system to confirm?

Andrew

Okay, is your trade commission accurate to your broker?

Does your test enter/exit the market "next day open"?

Is your test starting capital accurate to you?

Are the positions open at any one time what you will do?

* Have you back and forward tested the system on another market? (e.g. S&P500)

Are your stop loss or exit rules true to be used?

What is the maximum drawdown on the posted figures?

And finally but not only ...

Try testing these dates (16/07/2007 to 2/3/2009) and post the win/loss %, R/R ratio and MDD.

Thanks.

 

[brty]

The back testing shows that most of my loses are not full loses (below 100) and also because of the win/loss ratio, the result shows I make a profit.

OK, that means that your original results given are not accurate.

As long as I don't have lots of 100 loses in a row, I should be right???

Your system should tell you the likelyhood of this, what are the odds of it happening??

brty

 

[andrew100]

I have back tested the system on one year's historical data (2009) and it looks good, but when I tried on the dates that Wysiwyg suggested, it didn't work out too well. More tweaking to do I guess.

Thanks for your comments.

Andrew

 

[nomore4s]

I have back tested the system on one year's historical data (2009) and it looks good, but when I tried on the dates that Wysiwyg suggested, it didn't work out too well. More tweaking to do I guess.

Thanks for your comments.

Andrew

Be careful with this. Market conditions play a huge part in how profitable a system is.

All you need to do is look at a chart of the XAO for those two different date ranges to see the difference in the market conditions, not many systems would work well in both time periods.

Also 2009 from March onwards has been an exceptional year really, a strong rally after a massive decline. How often do we see market conditions like that? Maybe test it between May 2005 to May 2006, May 2006 - Oct 2006. Oct 2006 - May 2007 and see how the results vary in the different market conditions.

IMO with the results you posted in the first post you would need a strong bull market to make this system work and your draw-down periods would be quite deep at certain periods which will make this system very hard to trade practically. You would only need a small period where your win/loss ratio dropped to be in trouble.

 

[Boggo]

Be careful with this. Market conditions play a huge part in how profitable a system is.

All you need to do is look at a chart of the XAO for those two different date ranges to see the difference in the market conditions, not many systems would work well in both time periods.

There are some very valid points raised by both Wysiwyg and nomore4s, in particular the entry, exit and market conditions.

The entry and exit, does your system backtest enter the next day regardless (EOD assumption) or does it wait until a condition or price is met which => the initial buy signal or does it just enter regardless.
Wrong procedure here may have you entering against the trend.

Similar issues for the exit, on low volume stocks you will get slippage (min vol filter ?) and is your exit on the next bar (next day) after your stop is hit or do you have a clearly defined stop to exit the trade immediately if it goes wrong.

How does your system take profits and does it cater for an adjustment of the the stop to create a "free" trade.

In a sense I am probably complicating the backtesting process but I am also trying to highlight some of the many variables that are hard to test but will have the greatest influence.

I am not an expert on by any means but I have gone full circle and have now come back to the basics....
1. Don't trade against the trend
2. Know your entry
3. Know your stop
4. Have a potential price target
5. Use items 2, 3 and 4 to calculate R/R
6. Always have a $ figure that you are willing to put at risk
7. Item 6 divided by the difference between items 2 and 3 determine how many shares you buy

Below is an example pic of a current trade on TOL simulating a $20000 account with 2% ($400) at risk. Note that it does not buy the full $20000 value due to the $ at risk.
My stop is currently just above entry price to cater for brokerage. Barring some major gap downwards then this is now a free trade.

There is a great article on a free document (v6.5 Introduction PDF) on the left side of the page in this link, go to page 52 and read from there, it simply tells it how it is and is essential knowledge when considering any system testinghttp://www.mtpredictor.com/

I am not pushing MTPredictor, it is great software that does all of the above for you but you can do it all yourself with a basic charting program and a $2 calculator once you understand the process.
The process is to be in the trade with (known) minimum risk and the rest will take care of itself, there is too much emphasis on how much can I make instead of handling the reality that you will have more losing trades than winning trades.

This is my worth while I am on call and hoping that the phone doesn't ring.

(click to expand)

 

[DaveMac]

Hey andrew100, when you've got results like your risk:reward and win:loss, you should run it through Van Tharp's expectancy calculation:

(average win * reward) - (average loss * risk)

I can't be sure but I think yours looks like this:

(.71 * 1) - (.29 * 6) = -1.03

Again, I can't be sure because it looks like your average risk is 6 times your average reward...?

What you're aiming for is a positive number, of course.