Determining Swing Reversal Points

[Chorlton]

Hello All,

Can anyone offer any suggestions on how one can identify major swing reversals points?

When I used to trade discretionary I used to use the Zig Zag indicator and found it extremely useful. However, due to the way it is calculated I cannot use it when backtesting as the results obtained can be unreliable due to the fact that it can reference future data.

Any Suggestions?

Thanks in advance,

Chorlton

 

[Chorlton]

Hello Again,

Just wanted to "bump" this thread up again, in case it was initially missed by someone who may be able to help.

Basically, I'm interested in approaches that others might take to define a swing/ reversal point on a chart. The approach suggested must be able to be used reliably in backtesting.

As mentioned in my first post, I have found the Zig Zag indicator to be a good choice but unfortunately it can't be used for backtesting purposes, due to its reference to future data.

All Ideas Welcome.....

Thanks again,

Chorlton

 

[Vaco]

Chorlton,


When you say major what do you mean? which time frame? major for the day, week month or year?

 

[Chorlton]

Chorlton,


When you say major what do you mean? which time frame? major for the day, week month or year?

Hello Elefteros,

Thanks for the reply.

By "major reversal" I am simply referring to a point which signifies a change in the direction of the trend, on whichever timeframe I choose.

Basically, I want to determine reversal / "change of trend" points in a very similar way to what is produced using the Zig Zag indicator. eg. Remove the general noise from the chart.

Hope this clarifies....

 

[Vaco]

pivot points are a good start i use the weekly ones as there is less noise, add a moving ave on a higher timeframe to ensure your moving with the greater trend and your pretty much good to go.

books i've read recommend the monthy pivots as well as a 200ma for positional traders.

 

[tomorton]

Hello Chorlton -

For really long-term see Coppock. It doesn't always give a timely signal, and can lag the market as badly as a 200EMA cross-over, but occasionally gives a bang-on or leading signal. There are plenty of good examples in long-term charts, sometimes the Coppock is the only signal . Change of direction of Coppock is as important as crossing-over the zero.

Its not something I have done much research on because I trade shorter time frames but occasionally the swing points are stunning - e.g. Glaxo: made a significant 3+ year high price 10/03/06: the Coppock made its own significant 6+ year high just 10 sessions later on 24/03/06 and turned downwards. Neither level has been challenged since, price has fallen 36% at worst.

 

[the blades]

Hello Again,

Just wanted to "bump" this thread up again, in case it was initially missed by someone who may be able to help.

Basically, I'm interested in approaches that others might take to define a swing/ reversal point on a chart. The approach suggested must be able to be used reliably in backtesting.

As mentioned in my first post, I have found the Zig Zag indicator to be a good choice but unfortunately it can't be used for backtesting purposes, due to its reference to future data.

All Ideas Welcome.....

Thanks again,

Chorlton

Chorlton,

If the stock is on a pullback from an up-trend - why not fibs? I believe there's an add on tool in the AFL from Amibroker so you may be able to define it for backtesting purposes?

Cheers,
UTB

 

[Chorlton]

pivot points are a good start i use the weekly ones as there is less noise, add a moving ave on a higher timeframe to ensure your moving with the greater trend and your pretty much good to go.

books i've read recommend the monthy pivots as well as a 200ma for positional traders.

Hi Elefteros,

Thanks for the suggestion. I explored Pivots in the past but never for this application. I'll take a look...

Cheers....

 

[Chorlton]

Hello Chorlton -

For really long-term see Coppock. It doesn't always give a timely signal, and can lag the market as badly as a 200EMA cross-over, but occasionally gives a bang-on or leading signal. There are plenty of good examples in long-term charts, sometimes the Coppock is the only signal . Change of direction of Coppock is as important as crossing-over the zero.

Its not something I have done much research on because I trade shorter time frames but occasionally the swing points are stunning - e.g. Glaxo: made a significant 3+ year high price 10/03/06: the Coppock made its own significant 6+ year high just 10 sessions later on 24/03/06 and turned downwards. Neither level has been challenged since, price has fallen 36% at worst.

I'm never heard of Coppock til now and given the fact that I would be trading a shorter timeframe, I'm not sure whether this signal would be that helpful. But I'm always willing to look at different avenues so I'll do some research on Coppock.

Thanks for the suggestion........

 

[Chorlton]

Chorlton,

If the stock is on a pullback from an up-trend - why not fibs? I believe there's an add on tool in the AFL from Amibroker so you may be able to define it for backtesting purposes?

Cheers,
UTB

Hi Blades,

Thanks for the reply.

Although I can understand why you've suggested fibs, I'm not actually looking for a method to enter into a pullback which I assume you are suggesting?

In my case, I just need to be able to define how many bars have past since the point at which there was a recent change in trend occurred. The Zig Zag (as mentioned in post #1) achieves what I need perfectly, although as stated looks forward in time to calculate the last leg.

 

[swandro]

Chorlton

You CAN use the zig-zag for back-testing but you just have to bear in mind that a swing high or low is only confirmed when the price has moved the minimum size of the zig-zag in the other direction. But this is equally true of using a zig-zag in real-time.

I presume you use a zig-zag based on a number of points. Another way to do it is the Gann method - he would simply use a bar count, for example a 2 or 3 bar swing. So a top would be in when it was followed by 2 or 3 lower bars (lower high / lower lows).

This Gann approach gives very similar results to the point based style. If you take your reading on the bar close, you will have removed the problem of unconfirmed swings.

Also, this style is very easy to code.

Hope this helps.

 

[rathcoole_exile]

Can anyone offer any suggestions on how one can identify major swing reversals points?

a simple method is to identify the most recent high.
if it is the highest bar for the last 20 bars, wait til the low of the high bar is broken (on a Close) and go short.

obviously with a new low, lower than the preceding 20 bars, you wait for a break of the high point of that bar and go long .....

simple but effective

 

[BSD]

Good stuff Exile, interesting approach.

Way I do it is I'm in an uptrend as long as higher lows are being made, in a downtrend when lower highs are being printed.

Or one could use a moving average where price above signifies the uptrend, or a slope of a moving average, dual MA's, MACD, etc etc.

Ed Seykota'd just pull up a chart and eyeball it from across the room ;-)

 

[Chorlton]

Chorlton

You CAN use the zig-zag for back-testing but you just have to bear in mind that a swing high or low is only confirmed when the price has moved the minimum size of the zig-zag in the other direction. But this is equally true of using a zig-zag in real-time.

I presume you use a zig-zag based on a number of points. Another way to do it is the Gann method - he would simply use a bar count, for example a 2 or 3 bar swing. So a top would be in when it was followed by 2 or 3 lower bars (lower high / lower lows).

This Gann approach gives very similar results to the point based style. If you take your reading on the bar close, you will have removed the problem of unconfirmed swings.

Also, this style is very easy to code.

Hope this helps.

'
Hello Swandro,

Interestingly, this is the current approach that I am looking into. I basically search for a 5 bar pattern which has bars 1, 2 & 3 with Higher Highs and Higher Lows and bars 4 & 5 with Lower Highs and Lower Lows. The centre of the pattern then defines the swing point.

Overall, results in defining these reversals are ok but specific reversals (which are clearly noticable when eyeballing the chart) are missed simply due to one of the lows/highs not meeting the pattern criteria.

Consequently, I'm still playing around with other ideas in an attempt to capture more of these reversal points.


Regards,

Chorlton

 

[Chorlton]

a simple method is to identify the most recent high.
if it is the highest bar for the last 20 bars, wait til the low of the high bar is broken (on a Close) and go short.

obviously with a new low, lower than the preceding 20 bars, you wait for a break of the high point of that bar and go long .....

simple but effective

Hi Rathcoole,

Thanks for your comments.

Nice and simply approach. I like it as it is basically a micro pattern so in a lot of cases a tight stop can be used increasing the R:R potential. This is assuming the stop is placed just above the high.

However, in my situation all I want to do is identify the point at which the trend changes and more specifically how many bars have occurred since the change in trend.

IMO The approach you suggest wouldn't really be a suitable solution as price could easily move into a wide range creating numerous swing points although the larger trend hasn't really changed ie. it has continued to move sideways.

If you think of the Swings Legs associated with Elliott Wave Theory, these are the kind of larger reversal points I wish to capture.

All the best,

Chorlton

 

[rathcoole_exile]

my point is, you NEVER know in advance when a trend has changed direction or gone to/from flat.

there is NOTHING, imho, that can signal such an occurence.
The only way is trade each and every set up (whichever technique you use) in the expectation and knowledge that one of the trades will eventually be "the one"

all that elliott wave count theory BS is just that - theory.

we have another name for it where i come from - shyte ....

 

[Chorlton]

my point is, you NEVER know in advance when a trend has changed direction or gone to/from flat.

there is NOTHING, imho, that can signal such an occurence.
The only way is trade each and every set up (whichever technique you use) in the expectation and knowledge that one of the trades will eventually be "the one"

all that elliott wave count theory BS is just that - theory.

we have another name for it where i come from - shyte ....

Rathcoole,

You are missing the point. Obviously, I don't expect to know in advance of a reversal.

I simply want the ability to identify the last one that has already occurred. Visually, looking at a chart this is very simple but the challenge for me is in trying to look into ways of finding a reliable method which can be hard-coded, for backtesting purposes.

Basically, as already stated I just want to identify how many bars have passed since the last reversal, as I need this information for further calculations.

My reference to Elliott Wave was used purely to highlight the kind of swings that I am looking for ie. Trend Reversals / swings of a particular magnitude.... I could have easily mentioned the results achieved using the Zig Zag indicator.

Apologies if you misunderstood my intentions....

 

[TheBramble]

Can anyone offer any suggestions on how one can identify major swing reversals points?

Ahead of time or at the time, no, not really.

You can validly anticipate potential reversal points at 'likely' levels and times (S/R, Fibs, Time Cycles) where a major reversal is 'more' likely to occur and construct your play accordingly. You can also use some of the more classic chart patterns (climax, blowoff etc), but again, they’re to be taken in context and only on on-going performance. Needs luck and skill. (Doesn’t it all…)

The caveat, as always, is making sure you have go/no go criteria clearly delineated, a get out, and a general idea of where you expect it to go and what it should look like getting there.

I know of no formulaic definition of any level of useful probability which can be reduced to a N bars / X bars scenario that you’re seeking. Doesn’t mean it doesn’t exist of course…

 

[Chorlton]

Ahead of time or at the time, no, not really.

You can validly anticipate potential reversal points at 'likely' levels and times (S/R, Fibs, Time Cycles) where a major reversal is 'more' likely to occur and construct your play accordingly. You can also use some of the more classic chart patterns (climax, blowoff etc), but again, they’re to be taken in context and only on on-going performance. Needs luck and skill. (Doesn’t it all…)

The caveat, as always, is making sure you have go/no go criteria clearly delineated, a get out, and a general idea of where you expect it to go and what it should look like getting there.

I know of no formulaic definition of any level of useful probability which can be reduced to a N bars / X bars scenario that you’re seeking. Doesn’t mean it doesn’t exist of course…

Hi Bramble,

Thanks for your contribution although it seems clear that my original question has been misunderstood.

Please refer to post #17 for hopefully a more clearer explanation of what I am trying to achieve.

In essence, I am not interested in trying to anticipate reversals (or anything else for that matter). Unless someone has a crystal ball for sale Instead, my trading philosophy has always been to "react" (with the best solution) to what has just occurred & trade accordingly. I'll leave trying to picks tops and bottoms to others.

In my case, as mentioned above, I'm just trying to identify how many bars have passed since the last trend reversal. To clarify "trend reversal" I basically am referring to a x% change in the trend, in an identical way to that of the Zig Zag indicator. However, the approach I need is one which will not get "affected" when backtesting.

Hope this clarifies...

Chorlton

 

[idunno]

volume, trend lines will be very helpful for the turning point detection.