Hi CycloneI wo't say I'm an expert on reversal and ranging...

  1. 8,720 Posts.
    lightbulb Created with Sketch. 81
    Hi Cyclone

    I wo't say I'm an expert on reversal and ranging strategies - until recently I stayed away from them because I was wary of this issue of dealing with range breakouts. I've spent a fair bit of time a few ranging/reversal concepts in the last couple of months and as a result I think I at least understand some of the factors better, and to the extent I now have one strategy coded into a fully auto EA that I am happy with in terms of how it deals with this range breakout issue.

    I have so far developed a view on a 3 of factors that I think are important in developing a successful strategy, and they are as follows:

    1. Market selection. Imo not all markets are suitable for range trading. One factor is the total range a market has traded over in the last X years. Some markets have traded over a way bigger range than others. The AUDNZD is a good one because it has traded within a 1200 point range for the last 5 years (no guarantees that will continue of course), and puts in lots of regular swings. A second factor is also the size of the swings when a market starts to trend away from a range. If it breaks a range and starts to trend, but it puts in regular swings on the journey, then that's ok. But sometimes you can see a 500-1000 point move within the space of a few weeks or a month and there are no pullbacks. A ranging strategy can still do ok in a move with swings, but where they break down is when there's a move with no swings, imo. So picking your market(s) makes a big difference, and whereas I am in the habit with my models of trying to apply to as many of a standard set of 20 markets that I like, with my new reversal model I have picked only 3 core ones with ultra low drawdowns, and another 4 with slightly more but still acceptable drawdowns.
    2. The drawdown cutoff point. For quite a while I was focused on leaving lots of room for the range/grid trading to work, basically sizing the cutoff points to be enough to cope with the swings that I measured in each market. The problem is of course when the cutoffs are large, it only takes one or two breakouts to eat up most or all of the profits. I have managed to find large cutoff settings that produced good results, with cutoff values of 500+ points. The cost of hitting the cutoff though, even if trading with 0.01 lots and say a basket of as much as 10 positions, is a breakout can cost $400-500. I'm not comfortable with that. So I decided to try the other way, a much smaller grid with a smaller cutoff (100-200 points), and I have found much better results with this approach, which also seems more robust in testing. As a result there are more cutoff events, but there are some tradeoffs. The cutoff events now only cost less than about $70 (same basket parameters as above) which is a big plus, but as I said they happen a lot more often. But one thing I found was, on the right markets) that if you cut them off more quickly like this, then you get the benefit that the strategy can start trading again and ticking up the profits, whereas for a bigger grid with larger cutoffs, what can happen is trades will sit there open for months, costing interest swaps and also preventing additional trades being taken. I think that is an important factor.
    3. Cutoff method. I began by applying the cutoff range to all trades in the grid, at a distance from the 1st trades, so if a cutoff occurs then the first trade has the biggest loss, and the last one the smallest, etc. But I then tried a staggered exit, essentially applying the cutff value in points to each trade as a stoploss, so the stoplosses cost the same for each trade, however there are times when not all of them get hit, so more trades survive to close out.

    So to compare the effect, with the larger grid approach I was getting results like the following:

    https://hotcopper.com.au/data/attachments/1772/1772252-3038aa5fc0ede66cd229ada767e5f260.jpg

    You can see those breakout events cost around $300-500 each. The above shows about 1600 trades over a 3 year period.

    Whereas using the smaller grid approach the result is more like the following, obviously with a lot more trades as a result:

    https://hotcopper.com.au/data/attachments/1772/1772255-3f09536b9b469d1191cf78cfbf1b0b14.jpg

    Now the above doesn't look like the same type of equity curve behaviour as the 1st one, but if you zoom in (see below, the first couple of hundred trades from this period) you can see it looks the same as the first equity curve, but the values are a lot smaller (more suited to my risk appetite). You can also see the smaller steps where exits are staggered as opposed to the single big drops in the first curve. Open position drawdowns are also a lot smaller as a result.

    https://hotcopper.com.au/data/attachments/1772/1772270-8d5967855f710ad2e9b2eb3a82fbee30.jpg

    Hope the above helps.

    Cheers, Sharks
 
arrow-down-2 Created with Sketch. arrow-down-2 Created with Sketch.