To understand how to build own filters better read first Filters and their settings >>
After reading that page you understand the information a filter shows in the chart.
To design a new filter you first make a backtest of Active Traders main strategy. The ‘Report’ of that test open you click with the right mouse button on it and select: Save as report. After saving it your browser should open automaticly and show the report. This is your reference you want to improve with a new filter.
To build a new filter you look for loss trade the chart. It can be any loss trade. After building the filter you can test if this loss is a unique event or not.
The loss made on 29. of June has 2 obvious properties that can be seen on first look: candle 2 is falling and the days low is above the high of day 2 (horizontal line in olive color)
A falling candle alone doesn’t mean that the price should fall then. But its an indication. A
days low above the high of day 2 means that the price has been rising well on day 1, a indication that the price has moved a lot recently. There’s a third indication that can’t be seen here in the chart: high of day 2 is above the high of the week before. Another
indication that the price has moved a lot recently. So we have 3 condition to set as own filter:
- candle 2 falling: o2>c2
- days low above the high of day 2: lld>hhd2
- high of day 2 is above the high of the week before: hhd2>hhw1
We put the 3 conditions in the first 3 settings of OwnFilter1. Now we can do a backtest with the Strategytester.
The test shows the result has improved a little. The loss buy trade on 29. of June is changed to a sell win. But the amount of loss trades is not less, its equal as before. If a filter reverses the direction of a trade there should be 1 loss less and one win trade more. But we still have the same amount of losses. That means, that the new filter also has triggered a loss. To see which trades our own filter triggers we set TestFilterBuy to 201 to test own filter 1 alone .
A test from 2008 – 2018 shows that 24 profit trades and 8 loss trades were made. We take a look at the loss trades. The first ones were made in 2008 when our volatility indicator xl shows value way above 300 (shown by mouseover the arrow below the trade). The loss trade we wanted to elminate has a xl of 185. So we add to own filter1, in OwnFilterB1d the condition: xl<va300. To make clear that a variable is following we must write va before the variable 300.
Now the test shows 4 loss trades less. There’s also some win trades less, but that doesn’t matter for 2 reasons: 1. It is probable that some of these win trades are triggered by other filters, as filters often overlap. 2. The value of 300 for xl is nowadays a very high value, something that happened more often at the beginning of the financial crisis. But in the last years xl was always under 300.
There’s 4 loss trades left. One of them is under llm2, the low of 2 month ago. So the price had been falling und is then recovering. The trade on 29. of June started way above the low of 2 month ago, so its ok to set as further condition in OwnFilterB1e: c1>llm2
Now the test shows 1 loss trades less, 3 are left, we take a look at them.
They have something in common: The falling candle 2 and candle 1 together build out a some hours low. Something that happens often before the market rises strongly. Sometimes that low is even the low of the day, but here that is not the case.
To describe that as code we can write ll2==ll5. That describes that the 2 candles made a 5 hours low. In our code where we don’t want the filter to act when that occurs, as additional condition we write: ll2!=ll5.
With this condition added the filter is perfect, it just produces win trades. A test with all filters shows now nearly 200 pip improvement of the result. What here looks easy and fast mabe work for hours. The more experience you have, the faster …