Australian (ASX) Stock Market Forum

Joined
Feb 13, 2014
Posts
6
Reactions
0
If you had to choose one of the below three trend following systems to trade, which one would you pick and why?

System A:
CAGR 24%
maxDD 31%
MAR 0.77
W/L 2.18
Win% 43%

System B:
CAGR 31%
maxDD 39%
MAR 0.79
W/L 2.30
Win% 47%

System C:
CAGR 28%
maxDD 24%
MAR 1.17
W/L 1.73
Win% 51%
 
Joined
Oct 17, 2012
Posts
631
Reactions
1,159
Yikes, MaxDD>30% would be very hard for most mere mortals to survive and trade honestly. I'd be looking at number 3 for decent return and much lower DD.
 
Joined
May 24, 2008
Posts
500
Reactions
7
System C looks the best to me too. Lowest drawdown, highest win rate, and only slightly less profit than System B. I'd take a slight drop in profit for a major drop in drawdown any day!

And if System B's drawdown was acceptable to you (although seems a bit high to me), System C could be leveraged up to produce a 45% CAGR with the same drawdown as System B.

What do the equity curves look like?
 
Joined
Jul 10, 2007
Posts
77
Reactions
18
You have omitted one of the most important metrics in assessing system performance and reliability - trade frequency.
 
Joined
Feb 13, 2014
Posts
6
Reactions
0
You have omitted one of the most important metrics in assessing system performance and reliability - trade frequency.

The trade frequency for all three systems fall between 70 to 80 trades per year.
 
Joined
Feb 13, 2014
Posts
6
Reactions
0
System C looks the best to me too. Lowest drawdown, highest win rate, and only slightly less profit than System B. I'd take a slight drop in profit for a major drop in drawdown any day!

And if System B's drawdown was acceptable to you (although seems a bit high to me), System C could be leveraged up to produce a 45% CAGR with the same drawdown as System B.

What do the equity curves look like?

When back tested over 17 years, the equity curve goes from $100k in 1995 to $just above $10mil in 2012.
I will probably go with System C, which is based on a weekly time frame and System A & B are based on a daily timeframe.
 

Wysiwyg

Everyone wants money
Joined
Aug 8, 2006
Posts
8,424
Reactions
265
When back tested over 17 years, the equity curve goes from $100k in 1995 to $just above $10mil in 2012.
I will probably go with System C, which is based on a weekly time frame and System A & B are based on a daily timeframe.

My experience is with Amibroker back testing. I would choose none of these systems because they don't show the whole picture.
1) The cost per trade.
2) The entry and exit prices you get on back test will not be the same as in real time.

Number two is the unknown factor because there has to be your volume of shares available to either buy or sell and adding your volume of shares will also change the open and close price therefore changing history. In other words altering your back test results dramatically.
 

CanOz

Home runs feel good, but base hits pay bills!
Joined
Jul 11, 2006
Posts
11,534
Reactions
495
My experience is with Amibroker back testing. I would choose none of these systems because they don't show the whole picture.
1) The cost per trade.
2) The entry and exit prices you get on back test will not be the same as in real time.

Number two is the unknown factor because there has to be your volume of shares available to either buy or sell and adding your volume of shares will also change the open and close price therefore changing history. In other words altering your back test results dramatically.

Sounds like you saying you need more information Wys? How about if we had brokerage, slippage, the universe that it was tested on...basically the entire Amibroker back test report...which i assume was used to test this.:)
 
Joined
May 24, 2008
Posts
500
Reactions
7
When back tested over 17 years, the equity curve goes from $100k in 1995 to $just above $10mil in 2012.
I will probably go with System C, which is based on a weekly time frame and System A & B are based on a daily timeframe.

So is System C drawdown based on the weekly close, when the drawdown of the other systems are based on the daily close? If so, that could be why the drawdown appears to be less on System C, when it may not be any less at all in reality.
 
Top