I’ve been working on a strategy for breakouts that looks promising in back testing. It works best on pairs that move a lot or pairs that are in a trend. Ranging markets don’t tend to break out properly, so they’re best avoided. The entry rule I used is simple and as follows:
When the current candle closes above the high of the last 50 candles you buy, when the current candle closes below the low of the last 50 candles you sell. I’ve tested this with selling in an uptrend and vice versa, it still works as you catch retracements too.
Some other things that I think will improve results.
D1 time frame. It’s not a fast action strategy this way. My testing suggests about 20 trades per year on GBPJPY.
I’ve found best results on GBPJPY, it seems to work well for most JPY pairs. I’m sure it’ll work for anything that’s not usually in a range, probably avoid things like USDCHF although I haven’t tested that specific pair.
Follow the rules even if you already have trades open for that pair. You might make buys 3 days in a row if the breakout is strong, ride your winners.
And that’s about it. I’ll show some charts below of the backtest. The data is good and it’s long term trading, so the backtest will be accurate. Additional rules for the back test are:
Starting balance £10 000
£5 per pip on every trade
SL = TP = 1.5% of pair. So at 100 the TP would be 101.5 and SL 98.5, at 200 the TP would be 203 and SL 197
I suppose a fixed lots system doesn’t take advantage of the ability to compound returns, I’ve got a more complex calculation to deal with position sizing that allows the backtest to run but assume some withdrawals as the profit has run.
For that test, you’d have ended with the account balance shown and withdrawn approximately £150k.
Please try this strategy on different time frames and pairs and share your results with me in here. I will try to add this strategy to my trading and post results on here. I may even set my EA loose on a VPS so it enters at the same time as the test would
I’m sure there are other ways to manage the SL and TP than a fixed percentage, I think there’s lots of opportunity to expand the system. Possibly using structure from the daily/weekly charts.
The problem with that, is that a system is only as good as your ability to follow the rules. Adding something that creates a more qualitative approach means that back testing is harder and no longer valid unless you manually go through the charts. The rules I used worked well and I think are good enough to warrant a forward test.
Coincidence. I have also started looking at break-outs recently but have only skimmed the surface, focusing on Donchian levels. I don’t have a full strategy on this yet. I’m always a bit wary of strategies that don’t feature an obvious TA-based stop-loss level and I need to work more on this.
But it did lead me into another line of thought which is daily break-outs. Again, this needs a highly active volatile pair and I also selected GBP/JPY. The idea is to set a buy order at today’s high (SL at today’s low) and a sell order at today’s low (SL at today’s high). Close the open trade if there is one at tomorrow’s close. Win rate is low but the losers are limited. The winners however can be multiple times larger, which raises the long-term r:r. Testing with small positions from tonight.
Interesting. That strategy is something also used in scalping. I have no idea how successful it is, but I’ve seen somebody apply that to the 5M chart. I suspect it doesn’t work great in lower time frames, very little does as a mechanical system that I know of. I think that system sounds like it would have a high level of success in getting moves that go in the right direction early on, although where to set TP would be the big question. Let me know how your testing goes. I can try and write it into an EA when I’m not so busy (laugh at the ambition) if it shows promise
As for Donchian levels, I don’t know what they are. Seen them mentioned a few times, should probably take a read but I don’t like to get bogged down too much in theory. I don’t think more knowledge always means better performance in this game.
Cheers. I think you’ll find great similarities between your approach and Donchian’s. Richard Dennis’s Turtles used his approach too and were amazingly successful - though the profits relied on risk control and pyramiding, not a great entry signal.
The next obvious development of the daily high/low break-out would be to shrink the order trigger distance to sit within the daily range - this could be done rationally using ATR for example. So instead of using the high and low as the trigger range between long and short entries, shrink the trigger range down to a multiple of ATR. That work will have to wait but it might make sense when I have some baseline statistics to compare against.
I’ve set up an indicator on the 28 crosses of the major currencies to track this. I will not be taking all trades, but will try to post them on here so that we can track them. I’ll only be adding ones now that have only just made the signal.
AUDCHF sell at 0.66256 on 24/6
CADCHF sell at 0.73945 on 24/6
EURCHF sell at 1.00978 on 24/6
GBPCHF sell at 1.17676 on 25/6 (should be taken immediately on market open tonight)
NZDCHF sell at 0.60230 on 24/6
Unsurprisingly they all line up at once. CHF had an unexpected interest hike of 0.5%, so might remain strong for a while as markets are risk off currently. Not trading these unless something else tempts me, but I will track progress in here.
Edit:
I quite like the look of AUDCHF. Friday had a retracement, but that’s fairly normal after big moves for a Friday, so could give a good opportunity if Monday shows more risk off. The 1.5% SL and TP line up roughly with structure on the daily, so it seems possible that this is a good signal. I suspect the others are probably similar although I’d stay well away from CAD because that’s been bullish too.
I’ve read the turtle approach, it’s a little different but not too far away. They traded if a price went above the high of some period for 1 tick, I’m waiting for the close on an arbitrary period that I plucked out of the air, which will give fewer signals, but hopefully better ones. They used ATR of some sort to set levels, although I have no memory whatsoever of what they did with SL and TP, I seem to think they didn’t actually put those levels through with the brokers to stop people figuring out what they did. I also think they set the TP quite a long way ahead, because back then the markets trended much more than they do now.
This AUDCHF signal looks good, we’re about halfway to the 1.5% TP.
However I’ve backtested lots of pairs with this strategy, and most pairs don’t work well over a long time period, so I will only concentrate on GBPJPY in here. But it does seem that this strategy combined with an overall view of the market condition can work on other pairs too, just don’t take every signal. The backtesting does show other pairs that work, just GBPJPY is the best.
GBP/JPY is the most volatile major pair, also the most volatile GBP pair. Its the 6th most volatile of the big 28 pairs. Its been the 6th biggest mover in the last month.
Its also known as The Dragon. It can turn around and easily burn the trader.
That’s why I chose it to run. I thought it would be easy to add an indicator that demonstrates this for all pairs, but the back testing wasn’t great for many of them. This thread will track GBPJPY signals only, might be a slow thread…
I’m really busy at work so not been trading much at all. I’ll pick back up in a few months, but right now my life is too hectic to do anything more than buying shares that are under valued
