The previous trend categorization work is being put on hold. I’ve spent a little bit too long on this and there is so much to look into in FX, I don’t want to get bogged down on a particular idea. Overall I think the trend categorization idea is great, because it will allow a strategy to retune itself to each trend and hopefully maximize wins and minimize losses. However, I think it will be laggy as it determines what type of trend it is in, and during this time it may be prone to more losses. I’m a big fan of keeping things simple as well, so I’ll take my current trend research learnings with me and see where else I can apply them.
I’ve spent a week on traveling and since I haven’t had the time to sit down for extended periods and work, I’ve simply been getting through some research articles and links that have been on my to-read list, mostly in search of new ideas.
One thing I haven’t looked into much is breakout strategies. Most breakout strategies that I encountered early on in my research mostly involved the breach of some kind of support or resistance line, either defined by points being constantly retested by the market, or long moving averages. For some reason they didn’t really sit well with me, maybe because you can always find a number of exceptions where the line was breached but a breakout did not occur, or because as you look at different timeframes, the support/resistance line would be different.
However, I’ve recently stumbled upon breakout strategies based off ranges of time durations instead. Just think about this for a minute. For some reason over a fixed period of time, the price has moved in a significant one-sided direction in the market. The move has been significant enough to breach the normal range of that time period. This can happen either because the market has been trending so much that the new direction is obvious, or there has been a considerable spike that blasted out of a consolidation period.
For example, OneStepRemoved mentions a strategy called Double-7s. The idea is you enter when the market reaches a new high or low compared to the past 7 days (read more here: http://www.onestepremoved.com/double-sevens/).
These kind of time-based ranges make sense. There could have been some kind of fundamental change in the market like QE being announced, or new interest rate or inflation targets (like Japan recently) that will lead to a new long-term view on the currency. To me these strategies make sense in theory, which is a key requirement before I implement any strategy, at least for longer timeframes.
If you tried to apply this on shorter timeframes it probably wouldn’t work, due to the normal intraday fluctuations, but a daily or weekly chart for example makes sense. My logic here is it takes time for significant fundamental change in an economy to drive prices in a new direction.
So, I implemented the strategy – not Double-7s specifically as I don’t know where the 7 came from, but a more flexible approach which allows me to configure the time period. The results have been promising, but interesting and worrying as well. Check out the AUD/JPY example below.
The interesting/worrying point is the implementations usually do not have stop losses – something that indeed worries me. The idea is an opposite signal (may be from a shorter time period) exits positions. So for example, let’s enter when the price reaches a new high or low compared to the past 5 days, and exit when the opposite occurs over the past 3 days. The weakness with no stops was immediately obvious to me – we have to wait for the day to close before analyzing the exit. So if the price closed 10% against us that day, it would be too late! However, 10% days don’t happen very often but as we know from the black-swan experience, it doesn’t mean they don’t happen.
So, I added some stops and the results sometimes were better, sometimes were worse. So I optimized the hell out of the strategy over a long period of time. The strategy was making heaps of money, especially during periods like the 2008 GFC. How can this be? Well, here’s the equity curve for AUD/JPY for that period with a tight 20-pip stop.
The most profitable strategy according to the optimization had, for the period of Jan-2006 to Nov-2013, only 12 profit trades and 157 loss trades. The account balance grew by over 150% though. You can see though, that this kind of strategy is simply just waiting for the ‘big one’ to make all the small losses worthwhile.
Well, that might work for some but isn’t what I’m looking for. I think I’ll let the strategy close itself out, but place large 150+pip trailing stops for those 10% down days.