Predicted trends - are they useful?

This is a forum to discuss trading tools and approaches to timing the market.
Post Reply
gezza
Posts: 17
Joined: Wed Jul 22, 2015 1:46 am
Contact:

Predicted trends - are they useful?

Post by gezza » Mon Jul 27, 2020 1:46 am

Dear Group
As I sit in my covid cockpit, I've been wondering about the worth of a predicted trend (or channel).
By that I mean a trend (or channel) that is predicted after a pivot point is confirmed (see pictures below).
I saw something advertised for W59 users which would do that (for a large price), so I decided to create my own tool which did not require the user to find and set a 'correct' scale.
I was surprised in two ways.
The first was how amazingly accurate the predicted trend from my tool invariably was.
But I was more surprised when I realised that a typical trader may not benefit from even such reliable predictions.
I'd appreciate comments from the group since I seem to be missing something.
I've set out some example screen shots to illustrate my point.
1. Pivot confirmed
2. Apply the prediction tool (ie see the trend channel forecast)
3. Typical end result - surprisingly good.
4. But why bother with prediction?
I guess that some traders might like a confirmer, or maybe use it to set closer stop losses? Any comments?
Regards
Gezza
1 Pivot confirmed.jpg
1 Pivot confirmed.jpg (21.67 KiB) Viewed 613 times
2 Apply prediction.jpg
2 Apply prediction.jpg (38.27 KiB) Viewed 613 times
3 Typical result.jpg
3 Typical result.jpg (39.83 KiB) Viewed 613 times
4. Why bother.jpg
4. Why bother.jpg (43.95 KiB) Viewed 613 times

abacaba
Posts: 110
Joined: Sat Aug 15, 2015 5:02 pm
Contact:

Re: Predicted trends - are they useful?

Post by abacaba » Tue Jul 28, 2020 12:39 pm

Hi gezza,

Charts have to be correctly scaled to reveal anything actionable/relevant regarding price/time structure. That's why your channels don't really show anything useful. If you don't have a scaled chart, you have to do cumbersome calculations with a spreadsheet. Using a scaled chart is much quicker and intuitive.

For starters, try using a 45 degree angle on a scaled chart and drawing channels from relevant highs and lows. If you regard the 45 degree angle as a 1x1 as in Gann terminology, know that the 1x2, 2x1, 4x1, etc. are all really manifestations of that same 45 degree angle under different scale transforms . Setting a Gann fan at a particular x/y point will thereby provide a window into market structure at ratios which model the expansion and contraction of the time series fractals without having to rescale the chart. Fan lines only make visually explicit the price domain while implying the time domain. And they are just one class of predicted trend models. I personally don't use them anymore. There are better ways of modeling the vibration of a single price/time data point if that's what you want to do. But as an illustration, the underlying theory is important, and not understanding that is why most peoples fan lines don't work.

Todd
Last edited by abacaba on Tue Jul 28, 2020 7:53 pm, edited 1 time in total.

User avatar
ForJL
Posts: 142
Joined: Tue Jul 21, 2015 11:06 pm
Contact:

Re: Predicted trends - are they useful?

Post by ForJL » Tue Jul 28, 2020 1:09 pm

Hi Gezza,


The thing about trends, and I’m really not trying to be flip here, is that they are good for as long as they last. As for trendlines themselves there are countless ways to draw them, use them and you could write a book on what is actually happening when the trend is finally broken. Actually Edwards and Magee wrote a classic on it … “Technical Analysis of Stock Trends.” It's dated but still relevant in my opinion. In any case to answer your question directly about the typical trader I would simply say that like any method it all depends on the trader. Some will find merit in the approach and others wont. That is of course if I understand your question correctly.



Some time ago I developed a very simple approach to trend lines that I dubbed “True Value Trendlines.” The idea is simple … draw a trendline through a range of prices that has as many touch points as possible … both high and low. Mind you, it may look like a regression channel but it is not. There are a few ways to use the technique and I posted one of them below. All you need do is plot three True Value Trendlines, measure the distance in price between points A and B and then plot retracement levels from point C. If you want you can also plot High/Low parallel lines to the TVT also. Here we are using fib as the harmonic but GS and GPi are also possibilities.



Best,
Joe
Attachments
GTVT.jpg
GTVT.jpg (136.41 KiB) Viewed 584 times

User avatar
ForJL
Posts: 142
Joined: Tue Jul 21, 2015 11:06 pm
Contact:

Re: Predicted trends - are they useful?

Post by ForJL » Tue Jul 28, 2020 4:39 pm

Taking a quick lunch break and decided to follow up on my previous post to give another example of what I was talking about. This is what would have been the next True Value Trendline in today’s market action. It should be noted while this is being done in hindsight nothing would be done differently if we were to use current data.



The same procedure is followed as above to draw the next set of retracement/expansion targets. As you can see we get solid S/R levels at all fib levels and catch the eventual swing high at point E. Will leave it at that because I don't want to hijack Gezza's thread.


Back to work for me.

Joe
Attachments
GTVT_b.jpg
GTVT_b.jpg (152.71 KiB) Viewed 576 times

abacaba
Posts: 110
Joined: Sat Aug 15, 2015 5:02 pm
Contact:

Re: Predicted trends - are they useful?

Post by abacaba » Tue Jul 28, 2020 4:42 pm

Hi all,

Here's an example from today illustrating what I was writing about in my last post here. The chart is ES 4min, scale 0.25 x 2/bar. All the trend lines are "Gann" lines, just abstracted from the fan for clarity. These lines can create channels or single instance trend lines, which if broken as in Joe's example suggest a change in trend. Placement follows simple rules. I have used these in the past and this approach can be very effective in establishing trend and trend termination. But I prefer channels similar to the ones gezza describes now.
Gann tunnels.jpg
Gann tunnels.jpg (76.93 KiB) Viewed 576 times

gezza
Posts: 17
Joined: Wed Jul 22, 2015 1:46 am
Contact:

Re: Predicted trends - are they useful?

Post by gezza » Sun Aug 02, 2020 3:20 am

Thanks Joe and Todd for the great comments.

If I may paraphrase:

1. Todd notes that accuracy is required if predicted lines are to be of any use (and correct scaling by user is normally required for some predicted trend lines like Gann's fan etc).

2. Joe illustrated an approach which tends to confirm the hypothesis that predicted trend lines (however accurate) may not be useful since a user can draw trend lines (and channels) in sufficient time in real time to get the likely exit point.

In a private message, another member suggested that I do some back tests to see what the actual results would be.

My comments at this stage are that

a) it seems that the tool I'm using provides accuracy without the user needing to find the right scale. So to the extent that a predicted trend is any use, getting those predictions is not an issue.

b) It would probably help to combine the predicted direction with some SR levels like the ones shown by Joe. I'm still researching a way to get predicted levels that are consistent. The two problems I've encountered are that there can be way too many levels (eg Joe refers to possibly adding GS and GPi to the fibs) and also that fib levels often show fake levels (eg even on Joe's example we can see a fib level seemingly respected with a minor reversal only to be catch users out with a reversion to the main trend).

Pending me finding the time to get some numbers, I've set out below a continuous sequence on daily Gold up until now to show the kind of consistency we can get with predicted trends. All the lines are drawn at time of pivot confirmation. Since there is some AI built in, sometimes channels cannot be predicted, and sometimes two alternative harmonics are shown when a user can easily work out which one is working. The data is XAU-USD from Investing.com. The last few show the same predicted line at different scales to confirm that the user does not have to worry about scale. - Looks like a limit of five screen shots per post - so I'll send the rest in another post.

Happy trading to all.
Gold1.jpg
Gold1.jpg (73.65 KiB) Viewed 512 times
Gold2.jpg
Gold2.jpg (93.51 KiB) Viewed 512 times
Gold3.jpg
Gold3.jpg (78.56 KiB) Viewed 512 times
Gold4.jpg
Gold4.jpg (83.34 KiB) Viewed 512 times
Gold5.jpg
Gold5.jpg (78.74 KiB) Viewed 512 times
Gold5.jpg
Gold5.jpg (78.74 KiB) Viewed 512 times

gezza
Posts: 17
Joined: Wed Jul 22, 2015 1:46 am
Contact:

Re: Predicted trends - are they useful?

Post by gezza » Sun Aug 02, 2020 3:30 am

Here is a continuation of the Gold sequence ...
Gold6.jpg
Gold6.jpg (80.66 KiB) Viewed 512 times
Gold7.jpg
Gold7.jpg (83.25 KiB) Viewed 512 times
Gold8.jpg
Gold8.jpg (80.79 KiB) Viewed 512 times
Gold9.jpg
Gold9.jpg (84.33 KiB) Viewed 512 times
Gold10a.jpg
Gold10a.jpg (72.41 KiB) Viewed 512 times

gezza
Posts: 17
Joined: Wed Jul 22, 2015 1:46 am
Contact:

Re: Predicted trends - are they useful?

Post by gezza » Sun Aug 02, 2020 3:32 am

Same as last picture but using different chart scale to show relative direction of predicted line does not require user to set the 'right' scale.
Gold10c.jpg
Gold10c.jpg (19.21 KiB) Viewed 512 times

gezza
Posts: 17
Joined: Wed Jul 22, 2015 1:46 am
Contact:

Re: Predicted trends - are they useful?

Post by gezza » Sun Aug 02, 2020 3:43 am

Also, just to illustrate, its too early to determine if the 28 July high was a pivot, but IF it turns out to be a pivot, then the trend and channel is shown in the below screen shot using the same tool as all the prior examples.

This also highlights another factor with predicted trend lines and that is we don't know what level of trend is being predicted. As Joe well said, the trend is the trend until it isn't. Eg, will it be a temporary reversal or point to a new secular market?

In case anyone gets confused, I should make clear that nothing in this forum topic is to be taken as financial advice, but merely a discussion on the merits or otherwise of trading concepts.

Cheers
Gezza
Gold11.jpg
Gold11.jpg (53.01 KiB) Viewed 512 times

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest