Strange Currencies II: A Donchian Entry With A Time Based Exit
The Trading System,
This system is a variation of my first Forex trend following system 'Strange Currencies'. The system uses a Donchian entry entering long when the closing price closes higher than it has ever closed in the last 80 days, and enters short when the price closes lower than it has ever closed in the last 80 days. I used the day's closing price instead of the high and low extreme prices for three reasons. Firstly, using the closing price is a way to increase the system's chances of avoiding false breakouts as it avoids price 'spikes'. Secondly, because it's easier to follow - using the closing price means you only have to check the market once at the end of each day. And thirdly, and most importantly, because my mechanical testing shows that using the closing price tends to produce better results than using the extremes, particularly on shorter time frames. If this isn't what you experienced with your system back testing, please write and let me know.
Exits: Each trade must be automatically set to expire in 8 trading days if not stopped out. This means that during a strong up trend when the price is consistently closing higher, or during a strong down trend when the price is consistently closing lower, you will have lots of open positions on as new positions will be being taken before existing positions have been closed. This 'pyramiding' is an important part of the system as it increases the stake (and therefore your winnings) during stronger trends. Obviously, you can only have a maximum of 8 positions open at any one time per currency pair, and you won't be entering into another position on the same currency pair unless the position you entered into previously is in profit.
Stop Losses: During my testing I will be using no stop losses, adding stop losses to any system usually degrades the system's performance unless the stop loss is fairly large. It is far better not to 'over trade' by avoiding using excessive leverage than it is to try and trade with a very tight stop loss. Over leveraging and hence having to use to tight a stop loss is one of the most common mistakes novice traders make. However, it is also true that good trades rarely move to far into the red, so using a stop loss is important. I will include a column containing the stop loss that would have been needed to avoid being stopped out in my testing to help the reader to decide the degree of risk they wish to take when they 'tune' this system to make it their own - with a system like this I would suggest using a stop loss of somewhere around 2%-4% of the trade's opening price, depending on the markets volatility.
Testing: I am going to test this system on 10 currency pairs from early 2006 to around the present time in 2009. The full results include: The date and the closing price on that day, the maximum closing price in the last 80 days, the minimum closing price in the last 80 days, the closing price in 8 days time, the maximum high in the next 8 days, the minimum low in the next 8 days, the result of the trade and the stop loss you would have needed in order for the trade to not have been stopped out.
The Results,
| Currency Pair |
Number Of Winning Trades |
Number Of Losing Trades |
Number Of Winning Pips |
Number Of Losing Pips |
Profit To Loss Ratio |
| AUD/USD |
64 |
53 |
9910.7 |
8682.1 |
1.14 |
| EUR/GBP |
47 |
63 |
7625.2 |
6882.2 |
1.11 |
| EUR/JPY |
68 |
65 |
14080.3 |
14088.1 |
1.00 |
| EUR/USD |
72 |
36 |
15265.4 |
6298.9 |
2.42 |
| GBP/USD |
68 |
42 |
23393.7 |
15097.6 |
1.55 |
| NZD/USD |
70 |
70 |
9017.4 |
9298.4 |
0.97 |
| USD/CAD |
78 |
45 |
16332.3 |
10922.5 |
1.49 |
| USD/CHF |
65 |
48 |
12011.9 |
7685 |
1.56 |
| USD/JPY |
39 |
50 |
5645 |
7095.7 |
0.80 |
| USD/NOK |
70 |
53 |
84338.4 |
55922.9 |
1.51 |
| Average |
64.1 |
52.5 |
19762.03 |
14197.34 |
1.39 |
On average this system wins around 55% of the time and produces fairly respectable returns. With the exception of the USD/JPY currency pair it performs far better on the larger, more liquid currencies pairs than on the smaller less liquid ones. I believe this is because the large currency pairs with the highest liquidity are less volatile and tend to trend better than the smaller ones. On average, the system performed quite badly on the smaller less liquid currency pairs, except for the USD/NOK. I believe the reason the Norwegian Kroner vs. the US Dollar worked well is because the Norwegian Kroner's value, like the economy of Norway, is linked to oil prices and the oil market tends to trend well; I have no explanation for the Japanese Yen vs. the US Dollar's dismal performance though.
The full result details are available here.
Suggested Improvements,
Several of the trades are held whilst they move 800 or more pips into the red, usually stop losses degrade the performance of any technical trading system, however this is one system where the introduction of a suitable stop loss may not only limit the risk it might actually improve the expected returns. Different time frames to the 80-day break, 8-day hold could also be used (especially shorter time frames to increase the number of trading opportunities). The system could also be tailored to the larger currency pairs with the most liquidity, but beware of curve fitting. If the reader thinks they might be interested in using this system I suggest they experiment with these parameters, thoroughly test it, and make the system their own.
|
Site Navigation -
On Site Articles -
Offsite Links -
|
