Optimization: Intermarket Two

Optimize Lenl= 3 to 10. Subtract Commissions = $25, Slippage = $50

Source: Printed using TradeStation by Omega Research Version 4.02.15-Jul < 09 1996.

Adding another Filter to Improve the system

In system Intermarket Three, shown in Figure 14.13, we add another momentum filter to Intermarket Two. We will buy the S&P only if T-bond momentum is positive, while the S&P momentum is negative. This creates a situation whereby the T-bond momentum actually leads a change in the S&P. We look for a divergence in the momentum indicators and trade in the direction of T-bonds.

By adding a second input, we are making the system more complex. However, it is still a relatively simple system. If we optimize the two input values, we can see if performance holds up over wide ranges of inputs.

Figure 14.14 is an optimization table, sorted by net profit, showing all combinations of input values between 2 and 10. It is seen that all combinations are profitable, and a great many are very profitable.

We need to stop at this point to make sure we are heading toward a robust system, and to decide what input values we would consider using. We can then consider adding an artificial stop, such as a money management stop or a trailing stop.

Building a robust system

A robust system is one that will stand up over time. It is designed based on data collected in the past, and expected to continue to generate profits into the future. Since the future will not be exactly like the past, a systems robustness is a measure of how different the future can be from the past and still yield good results.

Page 1

Type : System

Name : Intermarket Three

Notes . Uses two momentum indicators.

Last Update : 02/07/97 11:12am Printed on : 02/07/97 11:13am Verified : YES

{Datal is S&P. Data2 is T-Bonds. This system mathematically transforms T-Bond prices to S&P prices. It generates a trading signal when the Transformed T-Bond line crosses the S&P line, but only when the momentum of T-Bonds is positive while the momentum of S&P is negative.. It is a simple system with two inputs, and is shown for development purposes.}

Van Y(0), b(0), begSP(O), begTR(0), g(0), j(0), m(0), n(0); Input: Lenl(3), Len2(10);

If CurrentBar = 1 then begin

begSP = Close of Datal,

begTR = Close of Data2, End;

If CurrentBar > 1 then begin

m = AbsValue(Close of Datal - begSP); n = AbsValue(Close of Data2 - begTR), lfn<>0thenb = m/n; g = b* begTR; j = begSP - g;

Y = j + b * Close[ 1 ] of Data2; {This is the transformed T-Bond line.} End,

If Y crosses above Close of Datal and Momentum(Close, Lenl) of Data2 > 0 and Momentum(Close, Len2) of Datal < 0 {This is the additional filter} then Buy {the S&P} on Open;

If Y crosses below Close of Datal and Momentum(Close, Lenl) of Data2 < 0 and Momentum(Close, Len2) of Datal > 0 {This is the additional filter} then Sell {the S&P}on Open;

Source: Printed using TradeStation PowerEditor by Omega Research Version 4.02.15-Jul < 09 1996.

In building a trading model, we must first strive to design our system in a way that is likely to be robust. This involves principles previously discussed, such as minimizing system complexity and assuring data adequacy.

However, all the factors that lead to robustness cannot be specifically verbalized or quantified. A necessary part of system development is to test each system for robustness. This can be done by carefully and exhaustively evaluating the performance that each system achieves through backtesting, before beginning to trade it.

Figure 14.13 System: intermarket three.

Optimization: Intermarket Three

Optimize Lenl = 2 to 10, Len2 = 2 to 10. Subtract Commissions=$25, Slippage=$50