Tuesday, September 16, 2008

KOP 10 and risk/reward

KOP10 is a streamlined version of KOP16. I've peeled off 6 of the original conditions since, after more testing, I've found their equity value to be minimal. In fact, by eliminating some of the conditions we find that they were acting as constraints to the equity curve rather than contributors. . . so in this case less is better.
However, the point of today's post is not to look at the performance of the KOP10 per se, but to look at what happens when we take a fairly robust system (IMHO) and apply pyramiding to the entries.
The KOP16 post included the middle pyramid option. . multiple different entry signals.
In today's KOP10 study we'll look at the risk reward of all 3 options. . . top to bottom.
While the % profitable is almost identical across the models, what does vary significantly is the max intraday drawdown. It should also be noted that pyramiding also requires an increased capital commitment since multiple positions are held before exiting on a common condition.
The difference in max consecutive losers, both long and short, is only 1 between the first two models, while the equity gain is 50% over the original model using 25% more frequent trades.
Max intraday drawdown, however, also increases 50% and this is the added risk.
Where things really get exciting is with the third pyramiding option turned on.
This model squeezes 5 times the gain out of the system where compared to the no pyramiding model and yields a whooping $62K net gain over the same test period on 719 trades (a bit active).
But. . .and this is a big BUT, there are few traders who would put their own money into such a system and for good reason.
First is the max drawdown. . .close to 10K, which is a lot to watch slip away and trust that things will turn out for the better in the end.
Second in the max consecutive losers versus winners which run 1:3 for the longs and 1:5 for the shorts. These aren't great odds in my book and ,although the allure of a $62K gain is enticing, this is a risky way to trade as the model tends to add to losing positions with different entries until the first exit signal appears.
If we just focus on the short side of the lower model, the results look more attractive as they almost mirror the total return of the middle model using the same number of trades.
However, the max intraday drawdown (2.5K) and the consecutive winners/losers ratio (1:5) still favor the middle model for my money.
That being said, the tactic to increase overall equity gain would be to increase position sizing in the middle model rather than buying the risk inherent in the last model.
TS2000i code for KOP10 shown below:

Condition1 = PercentR(7) Crosses Below 95 and RSI(Close,2)Crosses Below 87;
Condition2 = PercentR(7) Crosses Above 20 and RSI(Close,2) Crosses Above 18;
Condition3 = RSI(Close,2) Crosses Above 20;
Condition4 = RSI(Close,2) Crosses Below 95;
Condition5 = RSI(Close,2) > 90 and CCI(8) > 98;
Condition6 = RSI(Close,2) < 32 and CCI(8) < 32;
Condition7 = RSI(Close,2) > 78 and CCI(8) > 86;
Condition8 = RSI(Close,2) Crosses Below 86 and CCI(8)Crosses Below 86;
Condition9 = RSI(Close,2) < 28 and CCI(8) < 20;
Condition10 = RSI(Close,2) Crosses Above 21 and CCI(8)Crosses Above 23;

If Condition2 OR Condition10 Then Buy at Market;
If Condition1 OR Condition5 OR Condition8 Then Sell This Bar on Close;
If Condition6 Then Buy This Bar on Close;
If Condition3 OR Condition9 Then Exitshort This Bar at Close;
If Condition4 OR Condition7 Then Exitlong This Bar on Close;

No comments: