Moving Average with ATR filter

trendie

Legendary member
Messages
6,875
Likes
1,433
I've been going through my indicators, and deciding why I don't like (most of) them!
I revisited Hull MAs, and AMAs (adaptive moving averages).

There seems to be some logic whereby some people want faster averages, such as HMAs, and some want slower ones. I think this is because we want the indicator to keep us in trends, and not to spook us out after price takes a breather, but want it to be quick in identifying a trend-change. Never gonna happen!

However, this got me to thinking. Are there any moving average indicators that have an ATR filter on them?
In that, if the ATR falls below a certain value, the MA flatlines?
This is to replicate that if price starts to go sideways, you don't want to be taken out of a trade as the MA catches up to the congestion.
This would also allow the MA to not whipsaw in tight ranges as much.
Once price starts to move, and the ATR increases beyond its certain value, the average would stop flatlining, and start to move as per normal.

Is there anything like this or similar? I am trying to find a way to remove choppy action, and small-range congestion, ie, whipsaws, out of the equation. I know that manual trendlines would do this, but I cant automate them.
Yes, I have looked at Point and Figure.

thanks.
 
Try a Variable moving average. Sometimes its knows as a Variable Index Dynamic Average. Apparently it adjusts it's smoothing to how volatile the instrument is behaving. You might find a use for it?

Rossini
 
Rossini: thanks for input. I have looked at Vidya a while ago.

NR: hhmmmmm.
 
Yes, I have looked at Point and Figure.
Hi trendie,
As I was reading your post, I thought I know exactly what you want - and then I came to your final sentence! It sounds as if you've looked at P&F and dismissed it - why? I think it will do exactly what you want. You can still add any indicators you want to P&F charts - just as you would bar or candlestick charts. True, some work better than others - but you'll have no trouble with MAs.
Tim.
 
Hey,

What you can try is perhaps color-code your MA. As in, once you have establish your ATR filter rule, you can have your MA change color so that you can visually dismiss it.

Also, trading system-wise, you can just establish a rule aggregation system which initiates an order only if both MA and ATR rules are satisfied.

Other thing is about the ATR itself. Based on what you want to achieve here you probably need to use a variable ATR as well as opposed to a fixed-length period ATR so as to adapt to market cycles as efficiently as possible.

:)
 
I've been going through my indicators, and deciding why I don't like (most of) them!
I revisited Hull MAs, and AMAs (adaptive moving averages).

There seems to be some logic whereby some people want faster averages, such as HMAs, and some want slower ones. I think this is because we want the indicator to keep us in trends, and not to spook us out after price takes a breather, but want it to be quick in identifying a trend-change. Never gonna happen!

However, this got me to thinking. Are there any moving average indicators that have an ATR filter on them?
In that, if the ATR falls below a certain value, the MA flatlines?
This is to replicate that if price starts to go sideways, you don't want to be taken out of a trade as the MA catches up to the congestion.
This would also allow the MA to not whipsaw in tight ranges as much.
Once price starts to move, and the ATR increases beyond its certain value, the average would stop flatlining, and start to move as per normal.

Is there anything like this or similar? I am trying to find a way to remove choppy action, and small-range congestion, ie, whipsaws, out of the equation. I know that manual trendlines would do this, but I cant automate them.
Yes, I have looked at Point and Figure.

thanks.

This trick may only work with 30 minute bars, but here goes: what I do is take a 6 period MA of the 30 minute closes. I like 6 because the NYSE is open from 9:30 to 4, giving you 13 30 minute periods in a day, so this divides the day in two, or as close to that as you can get. Do this, and you will have a distinct morning and afternoon session in your data.
Then take volatility for 6 periods, and throw out any turn that takes place when this is rising, which it usually does in the morning. Wait for it to calm down, and if the signal still looks good, take it. If not, not. Goes with the wisdom that the dumb money piles in in the morning, while the smart money waits for the afternoon. Not always true, but it is often enough to be useful.
There's other interesting ways of using volatility. I used it for years to do all kinds of things before I turned to doing options exclusively, which of course depend on knowing volatility very well. A very underutilized and underrated statistic.
 
I like your idea Trendie. But instead of trying to find an indicator that will keep you out of the choppiness, why not figure out why the choppiness occurs in the first place, and then you'll better be able to avoid it without any indicator at all. On the short time frames for currencies, I can very often tell exactly where the choppiness will be.
 
thanks for all suggestions:

TheArtist: not sure what you mean by colour-change to dismiss it. If the MA is pointing up, its one colour, and if its pointing down, its another. Don't understand under what circumstances the dismissal occurs. Variable ATRs? That would require another set of back-testing within the bigger project. You then have two variables: testing for an optimal MA, then an optimal ATR within that. Complexity arises when the overall optimal condition could be a sub-optimal MA mixed with a sub-optimal ATR leading to an overall optimal trading combo. This would worry me, as it may smack of curve-fitting.

BDS: your views of recognising different action zones during the day are noted. My current settings attempt to catch the mood/trend of the european session, but also to be sensitive enough to react to a different sentiment when the US session starts, so I acknowledge your thoughts.

Shakone: currently, I have a rule-of-thumb to ignore my indicators if the range falls below a certain range, usually about 50/60, and switch to breakout mode, as I know that causes chop. However, I have to take a couple of hits before I realise Im in chop-mode. I do save myself further losses, but at a cost.
I dont know how to assess the cause of choppiness. I have read its about price being at a pivotal point, where sentiment may be about to change from bullish to bearish, and vice-versa. Or, its because of low activity. Dont know where to start with that one.

dcraig1: have looked at Kaufman. rejected it. the results, for me, do not warrant further work on it.

timsk: in truth, I would love to use PnF. I like the granularity, and clarity of the charts. dentist007 seems to be something of a master of them. I would have to apply some different ways of trading, esp in terms of finding how I would apply my pullback style to them. It may require a re-working of all my rules.
 
timsk: in truth, I would love to use PnF. I like the granularity, and clarity of the charts. dentist007 seems to be something of a master of them. I would have to apply some different ways of trading, esp in terms of finding how I would apply my pullback style to them. It may require a re-working of all my rules.
Hi trendie,
I don't know enough about your style of trading to be able to make useful comment but, in principle, there's no reason (that I'm aware of) why P&F charts can't be used in much the same way as any other chart type. All I can say is that for me, they remove an unnecessary barrier between me and price. I'm sure you remember the days when dbphoenix used to be a big hitter on T2W? I remember something he used to say frequently. To para phrase, he said that a price chart is a movie, yet most traders treat as a still picture frozen in the current minute, hour or day. They get too hung up on the meaning and significance of the last candle to print without looking at the wider context of what price is doing. Well, for my money, P&F charts pretty much remove that problem completely and the 'true' patterns are revealed, as opposed to what the last candle might have you believe. So often it is a smoke screen, cul-de-sac or even a trap. I believe firmly that if everyone included P&F charts in their analysis, their P&L would improve dramatically. To conclude, if you made the decision to put aside a few days to seeing how you could include P&F in what you're doing, I'd be extremely surprised if you regretted it and felt it was a waste of time. Needless to say, if I can be of any assistance (not that I'm an expert by any means lol!), don't hesitate to ask.
Tim.
 
timsk,

my biggest problem with PnF is that although they show Sup/Res far more clearly its the box-reversal logic that I feel can fool a lot of people.

For example, say the box size was 20 pips. and you decided on a 3-box for reversal.
what that means, if the price is going up, you add another box for every 20 pips price rises.
But, crucially, because you have a 3-box reversal, price would need to drop 3 boxes, ie, 20x3=60 pips before the chart recorded it.

Looking at the chart you might see the reversal, ie, when the reversal box is printed, ie, seemingly after a 20 pips reversal move, but in reality, it had to move 60 pips for the boxes to have been printed. You could change reversal size, but that doesnt alter the essential weakness, as I see it, in PnF as a trading tool.
So, I feel that whilst PnF is useful for marking S/R levels, I need the conventional charts to visually see congestion points as potential trade points.
 
timsk,

my biggest problem with PnF is that although they show Sup/Res far more clearly its the box-reversal logic that I feel can fool a lot of people.

For example, say the box size was 20 pips. and you decided on a 3-box for reversal.
what that means, if the price is going up, you add another box for every 20 pips price rises.
But, crucially, because you have a 3-box reversal, price would need to drop 3 boxes, ie, 20x3=60 pips before the chart recorded it.

Looking at the chart you might see the reversal, ie, when the reversal box is printed, ie, seemingly after a 20 pips reversal move, but in reality, it had to move 60 pips for the boxes to have been printed. You could change reversal size, but that doesnt alter the essential weakness, as I see it, in PnF as a trading tool.
So, I feel that whilst PnF is useful for marking S/R levels, I need the conventional charts to visually see congestion points as potential trade points.

Hey Trendie

I might just be thinking out loud here...but isn't a Bollinger Band tied in with volatility? Could you fiddle with the coding on a BB to get what you want?
 
I like your idea Trendie. But instead of trying to find an indicator that will keep you out of the choppiness, why not figure out why the choppiness occurs in the first place, and then you'll better be able to avoid it without any indicator at all. On the short time frames for currencies, I can very often tell exactly where the choppiness will be.

Do you mean, try to divine the fundamental reason why the choppiness occurs? There must be a range of reasons, surely. You reckon you can identify them? As a mechanical system builder, I would love to know how to predict choppiness, but it would have to be from the price action, not an external fundamental analysis type reason.
 
I agree with Timsk about P&F.
I would also consider learning more support and resistance (a la Pozzie if you don't like P&F). Then you have areas where you are looking for reversals and can use lower time frames or faster MA's if you want to identify it earlier.
 
Gumping: have trialled bolly-bands. my problem with them is that the break of them usually signals an extended move. the nature of bolly-bands are that they lull you into believing price will remain within the bands. the other component is of evaluating whether its a bolly-squeeze, leading to a breakout.

hi Nic,

since with pozzyp has been the first time I ever really "got it", with regards to Sup/Res, that would be the one possible avenue to pursue.
(actually, we are also big fans of Beachtraders threads, and his breakout methods)

a merry band of us exchange ideas via skype, and we have come to agree upon a creed of "If it aint broke, dont fix it", referring to the endless chatter about making things better, improving, tweaking, etc. Probably the talk has been brought about the past several weeks of choppy markets, and us having experienced difficulty in trading well since beginning of Feb.

our creed has a logo: its a set of knuckles, with a big fat ruler about to descend upon them. :) :)

hope you had a great week, and look forward to the next one. (the intra-day ranges appear to have returned)

to all intents and purposes, this thread, for me, has reached its conclusion.
 
Top