I have several strategies based on momentum and though these are not tradable yet I think they deserve some second thought. I'm used to set up a trailing stop based on some moving average but I made some test and the ATR seem to work better... in some cases. On strong trends moving averages seem to have better performance.
On the other side ATR look to be more in tune with the current conditions of the market.
I'm going deeper on further analysis but at this point I wonder if you use ATR (or a percentage of it) as a general rule to trail your stops? Also, if the Stops are dynamics why not the TakeProfit level? (in case we include a Take Profit on the strategy). If so, do you use a pre-defined set of variables on ATR based on Timeframe, market,...?
I recognize I don't master the ATR concept beyond its formula neither what did drive Wilder to buid it, that's one of my pending topics long list. Probably its knowledge would point me for the right use, but I'm also interested in your experience/advice so your thoughts will be very appreciated.
Well you sort of answered your own question, sometimes one works better than the other - this is always the case with all arbitrary trailing stop methods, so just pick one and run with it, neither is better because neither is good.
I use to really like ATR, but because it's lagging and takes an average, when you transition from high to low volatility, your stops are too wide, then when you transition from low to high volatility, your stops are too small - this transition happens all the time and will happen when momentum comes in and then leaves the market- which is exactly where you are trying to get in and then get out. This means you always end up with a cruddy stop at the wrong time.
The same issue lies with MA's.
It's better to use a stop that is based on price action, so for a momentum uptrend break, you're looking at placing the stop a tick below the lowest low, and move it up as the market swings and makes new lows. However, you also need to understand you can get gamed at the start of a trend by people running the stops before driving the price higher, so you need to test it.
Your next question is of course, what happens when the bar is really big, you would give up a lot of the profit by using swing points as a trail? The answer is, exit on spikes - they are usually a sign of exhaustion. Your stop should work in tandem with an exit signal/target. Don't rely solely on a trailing stop because you will always miss the big spike at the end.
You can also test and find the average number of swings a market will make before it exhausts itself, so if for example 3 swings is 2 standard deviations, then it will make sense to trail and then exit on the spike in the 3rd swing.
The following 3 users say Thank You to PeakGrowth for this post:
An issue i have encountered, when changing the stop and target dynamically in the strategy
is 'over-fill', means the original order (target or stop) gets filled and your changed new limit also
so besides the 'what is best', if you make if dynamically in the trade, you will bump into that
issue on top of all the rest...
The following 2 users say Thank You to rleplae for this post:
I have found that most of the time, ATR stop and/or profit targets are much worse than just regular dollar stops and targets. That sounds strange, and doesn't always make sense, but it holds true for most of the swing type strategies I create.
If you have any questions please send me a Private Message or use the futures.io "Ask Me Anything" thread
The following 7 users say Thank You to kevinkdog for this post:
The magic of ATR is that it allows us to normalise volatility across a wide range of instruments, for instance, I can position my ZN and my ES position so that the volatility of both instruments is the same in a portfolio. As a rule of thumb, ZN with its lower volatility would have more contracts and ES of course would have less contracts. Using ATR we can get a pretty accurate picture of the volatility of both and now we can calculate the sizes of each accurately, rather than relying on the rule of thumb I mentioned earlier. Also adjusting for changes in volatility is a breeze using ATR.
Whenever you use ATR as a stop, your stop will be closer during periods of low volatility and wider during periods of high volatility. If you size your positions according to your stops, i.e. risk 1% of your capital per trade then position sizes will become smaller when volatility increases - the effect on your position sizing is exactly the same as using ATR to size positions. Of course, if you increase your stop, your profit targets should also increase. What happens now, is that by giving the trade more room, your positions have become much smaller. Also, it becomes more difficult for either your stop or target to be reached and in my own trading, positions would fluctuate a lot more without reaching either their stop or target. Thus, volatility of the portfolio lower, but positions open for longer and the trade needs to move more before you can get the same profit / loss as during low volatility periods.
Using $-stops and targets during high-volatility periods increases the odds that either your stop or target will be reached. Position sizing remains constant, but trade frequency can increase as both stops and targets have the potential to trigger sooner. If you go through a win-streak, then using $-stops will lead to greater profits than ATR-stops. However, if you hit a losing period, then you will lose more.
The following 4 users say Thank You to grausch for this post:
That's the other element I was considering but didn't look as "fancy" (just kidding here) as my other two options; MA's and ATR. I had the feeling that ATR could be more reliable on adjusting the Stop as a mesure of how excited the market is.
Also the Profit. If the market spikes and enter on its exhaustion zone you better take the Profit and run.
That opens a new perspective to me. In fact sounds quite strange but if it works well worth a try. Though I'm quite averse to explore things against my rational mind.
Regarding trailing stops, review this thread on the TradingBlox forum - Dead Turtles - Traders' Roundtable. You will note that trailing stops degraded system performance in all cases. Best results were had once they started scaling out of trades. The logic was to take the full position on entry, either get stopped out on the full position, or gradually scale back as the trades start working. I also tested various trailing stops and found that while they make me feel better, i.e. I get more winning trades, as a whole they degrade my performance.
Of course, the Turtles used long term strategies on daily bars, which may not be applicable to day- or swing-trading - nonetheless, in my own development I learnt a lot from Curtis Faith's books and some of these principles I still apply today.
The following 2 users say Thank You to grausch for this post:
I don't, at the moment, myself, but in his book Trade Your Way to Financial Freedom Van K. Tharp details several examples of trading methods (mostly for commodities, as I remember) for which an ATR multiple proved a stable, consistent and productive method of trailing a stop. It's a perfectly viable method, in principle. The question, of course, is whether it's better overall - for any individual method - than any other, which is typically far from easy to research and prove.
No great surprise, there? The reality is that those key words "in some cases" are going to apply whatever method you use for trailing a stop, if trailing stops are your approach. Whatever method you eventually use, there will always be occasions when something else, with hindsight, would have been better. This is just one of the inescapable market realities, but it by no means excludes ATR-multiples/fractions as a valid method.
The following user says Thank You to Tymbeline for this post:
Having an open mind will help you find success a lot faster, don't let your ego get in the way.
It may not be rational to you now, but once you accumulate more knowledge you may look back and think how irrational it is now. Statistics don't lie, but your mind loves to lie to you by connecting dots and rationale that aren't necessarily there.