Trading strategy "Butterfly Tie"
In issue 4 of ForTrader.org we will look at a trading strategy based on bow tie patterndescribed by Dave Londry. The strategy is based on three moving averages (Moving Average): a simple moving average with a period of 10, and two exponential averages with periods of 20 and 30.
Trading strategy algorithm
1. Signal conditioning
A buy signal is formed when the moving average with a period of 10 becomes higher than the average with a period of 20, and the average with a period of 20 is higher than the average with a period of 30. The price should form a minimum. In order to formalize and determine the minimum, we took the Stochastic Oscillator with parameters 1,3,3. When such a combination of indicators is formed, a buy order is set at the current maximum. The sell signal is formed by analogy with the sell signal, in this case the rules should be reversed.
2. Position closing
We did not find a description of how to close a trade or what to do with an unfinished order in the strategy description. So we decided to delete the pending order from the previous pattern when the next signal is generated. We will close the open position when the price crosses EMA20 from top to bottom when buying, and sell when it crosses it from bottom to top.
Strategy testing
Let's test the developed rules and consider the results. The author recommends to look for patterns on the daily charts, which we have done.
Having tested the system with standard parameters on various financial instruments, we obtained positive results on a number of them. Now let's try to apply this method to intraday charts.
The results show that trading on intraday periods by the standard parameters of the strategy is unprofitable. Let's try to choose the most optimal parameters (periods) of moving averages for intraday charts in the period from 2007.01.01 till 2008.01.01 and then check them on the future period up to the current moment.
Having tested different combinations of parameters for EURUSD H1, we settled on the following parameters:
SMA with period 10, EMA with period 70 and 275. Let's see what results the optimized strategy produces.
As we can see, the result is positive, for this period the profit was - $817. Now let's check how these parameters work for the future period 08.01.01 to the present.
Testing showed that the parameters turned out to be workable not only on the historical data, where the original study was conducted, but also on other data. The result for two months was just over $200.
We will conduct a similar test for EURUSD H4. We will select the parameters from 2006.01.01 to 2007.06.01. Timeframes are moved apart, as the selected timeframe is higher and the period for selecting parameters needs to be longer to take into account the strategy operation at various price movements.
After obtaining the results of parameter matching, we took the average result, because the best parameters showed a small number of deals.
Result: SMA with a period of 25, EMA with a period of 20 and 120.
Such settings of averages (second EMA less than SMA) generate a different kind of trading algorithm, because the standard described pattern is totally inoperable on 4H.
Let's check these parameters on the section from 2007.06.01 to 2008.03.15.
As you can see, the parameters we have chosen for H4 are also workable for future periods.
Bottom line
In general, the system showed good results with 100% mechanization rules, it shows the possibility of its more effective use for non-mechanical trading. The advantage of this system is that when the conditions of our pattern appear, the discretionary trader is able to assess the overall environment to open a trading position. The mechanical approach, on the other hand, gives the opportunity to choose the most successful parameters to track a good set-up.