I’ve got a strategy that has been backtesting strongly on EUR/USD, on trending type years, but crap on other test material. For the two years I’ve tried it, it sucks on AUD/USD (2007), and EUR/JPY (2015).
I’m trying to incorporate an ADX filter to my strategy, to help keep me out of ranging markets. I’m finding though that the lag on it is causing me issues, getting me into trends too late. I suppose I also am still struggling to identify myself when a market is switching from range to trend.
Is there a better mechanical way of filtering for this kind of thing? Should I just shorten the window on the ADX?
I would like a strategy that is not over optimised, but robust across different market conditions. Is this realistic?
While I’m throwing my newb questions out there, I have one more.
Is there a measure or indicator of some kind that keeps a useful track of how “wicky” the candles are? I know from babypips school that crosses can be more spiky, but I am hoping that once this strategy is built, it will be robust enough to work successfully across all the majors and minors. At the moment it is pretty much calling for an initial SL of 1/2 ATR. I was hoping to maybe modify this somehow depending on the spikiness of the pair, if I could come up with a useful rolling measure of that.
Finding a way to get you into real trends and out of failing trends is clearly important to make the most of your strategy. But it seems like you’ve decided ADX is the the tool to do that and are now seeking a way to make that tool work. If you find a way to optimise ADX for the under-performing years, it seems very possible you will erode its performance in the other years.
Firstly, if ADX’s under-performance is so isolated, why not just live with it? If it performs well in 80% of the years tested, that should be enough for a profitable margin. If not, you could optimise performance in the winning years, rather than improving the poor years: this has far more potential.
Re: getting out of failing trends, after your replies to me the other day, I’ve modified my trailing stop exit system. Previously the trail was set to 2xATR, but I’ve now adapted it to begin at 2xATR, and after three days I am allowed at my discretion to tighten it to a minimum of .75 ATR. I need to do alot more testing on this though, but conceptually it makes sense that it would be an improvement.
Its a bit of a mish mash of your manual exit system, and a bit of Nick Radge (from the latest episode of chat with traders podcast) where he talks about (equities) using a “dynamic” trailing stop loss. This is what he calls a “regime” filter where if the overall stock market is moving against the trend of the trade, the stoploss starts to tighten up, while still leaving just a little bit of room in case the outcome changes.
What alternative tool would you use in place of ADX? I tried bollinger bandwidth ratio for a while with mixed results. Maybe just a simple RSI, is that kind of what you mean?
I take your point re: optimisation. Maybe I will give up the idea of having a strategy that works on “everything” and just go for one that works more strongly on one or two things. I suppose it was a bit of a silver bullet mindset. Thanks again for your reply, I always get alot out of what you write.
Actually I wouldn’t advocate any off-chart indicator or Bollinger Bands for any purpose - I think they’ve been sucking traders down rabbit-holes for years. But if you find your strategy absolutely demands an indicator to improve its performance (no indicator can make a strategy) it must be the right one, it must earn its place on the screen and there must be no more direct way to resolve the issue. By direct way I mean by studying price.
But I like the idea of a dynamic SL. And I think its perfectly good to start with a wide initial SL and then narrow it down according to price action after opening the trade.
I’ve been thinking alot about this comment. I understand that all indicators are derived from price action and by their nature are usually either lagging, or leading and so more prone to fakeouts.
I wanted an indicator which would show me when a market was trending or ranging. I chose ADX because of the babypips lessons. It lagged too much that I was missing trends so I halved the time window input. While all this had been going on I’ve been trying to make an effort to start reading and predicting price action based off what I’ve been learning, trying to look through the indicators a bit.
My previous reasoning had been: If the ADX keeps me out of x% of the ranging market trades I would have been lured into, then great, its done its job. This seems like an all upside decision. But today I start to see that actually it also keeps me out of trades I should have gotten into, so perhaps there is a trade off there that might not be worth it, forming a crutch.
I suppose I could backtest a year on a pair with my current system, then go and do something else for a week or so, then come back and see if I can beat the system myself reading purely price action. Then maybe we might see.
Do you mind me asking how much weight you give to reading volume also?
Edit: nevermind I just found your thread, I will read that.