Check out the TrendSensor Blog ....

 MarketBrief

Trading System Review - Feb 2008

  

 

This update: Feb 18, 2008

 

ISSUES: The recent dramatic declines in stock markets and increased volatility in markets generally, have resulted in poor performance for the TrendSensor Systems (Forex and Index systems). Specific areas of concern are:

 

  • Both systems have performed well with a select range of markets, but when these markets change (as they have), it may take some time before a new range of markets emerge as those best suited to the TrendSensor systems.

 

  • The Index system is set up to favor long trades and a bullish market, and doesn't perform well in a bearish market.

 

  • Neither system has performed well during highly volatile market periods.

 

APPROACH: Aside from improving system performance, my main focus has been on greater system robustness: with the goal of achieving consistent results across a full range of markets and market environments (bullish, bearish, volatile, trending, trendless, etc).

 

RESULTS: In the table below, all 4 backtests are done on exactly the same basis, with the results on the left being for an updated version of the current TrendSensor System, while those on the right are for a new TrendSensor System that evolves from some components of the current system, but leaves out the
worst-performing components in the context of the issues outlined above. Since the new system represents an evolution, but a fairly major divergence, from the current system I have called it "TrendSensor II".

TrendSensor

TrendSensor II

Index Markets:

 

Yearly Return: 

Yearly Return: 

Comments: When applied across a full range of index and index ETF markets, the current TrendSensor system does not perform well, but TrendSensor II appears to be much more robust, at an aggregate level.

  

Security Profit Distribution


Comment:  Drilling down to the security level, shows that TrendSensor II is indeed far more robust across a full range of markets than the "tweaked" version of the current TrendSensor System.

Backtest Results:  Index Markets

 TrendSensor TrendSensor II

Average Yearly Return (% p.a.)

 -2.12% +118.25%

Profit Factor

0.9597

2.274

% Winning Trades

30.9%

49.3%

Win/Loss Ratio

2.1397

2.323

Avg/Max Win Run

2.23 / 14

3.25 / 36

Avg/Max Loss Run

4.97 / 32

3.30 / 26

Avg Duration: Winning Trades

69 days

61 days

Avg Duration: Losing Trades

15 days

20 days

Avg Profit: Long Trades

+$6.46

+$3179

Avg Profit: Short Trades

-$15.85

+$4569

System Expectation

-$0.03

+$0.30

Max Peak-Valley Drawdown

57.22%

37.62%

 

 

 


TrendSensor

TrendSensor II

Forex Markets:

 

Yearly Return: 

Yearly Return: 

Comments: Forex returns from TrendSensor II are less impressive than those for Index markets, but still a big improvement on TrendSensor in terms of size of returns and consistency over the years.

Security Profit Distribution


Comments:   Once again, TrendSensor II proves to be much more robust across a wide range of markets, than TrendSensor is.

 

 

Backtest Results:  Forex Markets

TrendSensor

 TrendSensor II

Average Yearly Return (% p.a.)

-1.0%   

+28.37%

Profit Factor

0.9825

1.637

% Winning Trades

36.36%

47.0%

Win/Loss Ratio

1.843

1.856

Avg/Max Win Run

3.06 / 20

2.75 / 24

Avg/Max Loss Run

3.45 / 21

4.90 / 30

Avg Duration: Winning Trades

97 days

80 days

Avg Duration: Losing Trades

44 days

39 days

Avg Profit: Long Trades

-$27.03

+$56.10

Avg Profit: Short Trades

+$26.46

+$154.00

System Expectation

-$0.01

+$0.15

Max Peak-Valley Drawdown

52.98%

12.88%

 

 

 

The characteristics of both TrendSensor II systems are much improved over the original TrendSensor systems, both in terms of return and risk profile - and especially in the area of robustness across different markets and market environments.  While the performance of the TrendSensor II Index results clearly exceed those of the TrendSensor II Forex systems, the Forex system has a much more modest and consistent risk profile, so is likely to be a good complement from a diversification perspective.

I am beginning to trade the TrendSensor II systems immediately, and ceasing trading the original systems.  The robustness of the new TrendSensor II systems means they will be traded as automated systems with limited discretion on my part (only trade size and stop-loss system approach in most cases).

Finally, the period covered in the above backtests was constrained by the fact I have less data history for Forex markets.  The following chart shows over 12 years of yearly returns for The TrendSensor II Golden Index system:


Parameters And Assumptions For Backtests:

Risk per trade: 1.5% of balance

Max capital in trade: 1.5% of balance (max trade size = 12% of balance)

Slippage built into every transaction (buy and sell):

I normally enter at the open and exit at the open, a stop or target, as I make no attempt to time entries on an intra-day basis.  For backtesting purposes I have assumed the following conservative entry and exit prices, thus allowing significant "slippage" in each transaction:

Long Trades:

Buy Price = [(2 * O) + H] / 3  ie: 1/3rd of the distance from the open to the high for the day.

Sell Price = [(2 * O) + L] / 3  ie: 1/3rd of the distance from the open to the low for the day.

Short Trades:

Buy Price = [(2 * O) + L] / 3  ie: 1/3rd of the distance from the open to the low for the day.

Sell Price = [(2 * O) + H] / 3  ie: 1/3rd of the distance from the open to the high for the day.

Max Leverage:  8:1 (avg leverage under 5:1)

Margin Rates:    Long trade debit rate = 4.5% p.a.

Short trade credit rate = 2.5% p.a.

Compound and re-invest gains (and reduce trade size for losses):  Yes

Max Portfolio Heat:  55%

Pyramid Trades? Yes, from trading system triggers only (to a max of 5 levels).