I have to say I am a bit of a loss as to what you’re pointing out?
Are you just showing the 1-5 movement in vol currently, and awaiting the abc correction? It seems while that may be there, that price can still do whatever it wants? If what I’m saying makes any sense?
hi rel,
please suggest me about tro’work.Is it going to be useful to go through all of those 100s of pages specially when i am already following your work.
So I was thinking about this idea, and how I may go about visualizing, or seeing it with the data possibly. So this has nothing to due with direction, but I was thinking about simply how far an H1 might move, or what its range might be given it’s relationship to volume.
Also, related to YOUR findings you have mentioned many times, the range or distance price moves on H1 and lower has no where near the same UNIFORMITY that the VOLUME has on H1.
Also, with the post you made above, about the H1 bar, how it actually MOVES with INSIDE that bar… haha I have actually tried to model that with various types of home-made zz indicators… Just to see if I could find some type of repeatable patterns that may exist.
But I digress… attached are two pictures, of the H1 range (orange), as compared to it’s respective volume on H1(blue). And there is a bit of a correlation possibly, as you mentioned the volume on H1 you suspect drives price. Which in my current studies as shown, surely price does move more typically as volume accompanies it. Again this has nothing to do with direction, just pure range with respect to volume. Again this is a rolling 4 week average of the range on EACH INDIVIDUAL DAY, for each hour. Lemme know if the pictures are not clear, but just a general look at the shape should be sufficient I think.
I dunno if this is anything of value practically at the moment, so I’ll just keep posting my thoughts as long as it doesn’t clutter up the thread to much, as I’m sure I’ll fire a hundred blanks before posting something useful
I’ve actually made quite a few of them, for different purposes. I’m not very good at writing scripts, so if I create a new one, it will probably take me some time.
I even created one, with a purpose of looking to see how one could play continuous reversals throughout the day. And this question could go to anyone reading this thread… But have you ever looked at a chart in hindsight, and thought to yourself… “Hmmm, if I had just bought and sold/reversed position every time price moved x tics off a high or low, i’d be rich!!!”
Well, I was looking to see if there was a pattern to the optimal, or close-to-optimal length for each day. I.e. maybe some days that length would be 10 tics, others it could be 30, maybe it was dependent on volatility, time of year, day of week, etc… Idea was to look historically, at how various lengths would work on any given day, but that was an idea all together different. Imagine the script, for each day would check the length of 1-30 tics, and report statistics on how all those various lengths performed… that was the idea at least.
I’ve reformatted, but the problem now is that NT is weird with its sessions, and does not allow one to set the time scale to GMT, unless you change your PC clock all together to match it. So the sessions are somewhat mixed, making data collection a PITA… I can export raw data into excel, but my skills in excel are even worse than my skills for using NT, so scripting a zz in that is even more of a challenge. I’ll have to work on a solution to this.
But with the zz, I’ll have to look into what I think I can find useful with them again with regards to data collection. Most of the zz’s I’ve created work simply based on length, so if price moves in the counter direction greater than a certain # of tics, it would start a new leg. But I think this is limited. As there are just some days that have too many swings, to the point where any patterns that exist get muddled in too much data, and too many ways to try and classify PA.
I think your concept of simply classifying swings based on 4 simple patterns might be more useful. At least that way it’s an easy way to organize your data. I will have to give that idea a bit more study.
Hehe, I’m not sure what I’ve scripted is that intelligent yet to be useful?!?!?! You can tell what that is right? What I’ve shown in the picture? I think it’s not too obvious to someone just passing by, but if YOU know what it is, then maybe it may be something useful.
But about you’re waves, you are talking about your U1/D1 - U4/D4 correct?
These aren’t much yet… but I’m getting some ideas based on them. Kinda look like measured moves, wouldn’t you agree? They seem to share very similar traits with regards to duration, overall angle and distance traveled.
Will continue to give it further study.
Pics are same chart, one is just zoomed in a bit, in the area I’ve highlighted.
Something else I was working on was creating some stats. These are primarily weekly stats. Interesting enough, similar to some data Rel grabbed over a year ago, is where trends start, or how they develop through the week. The weekly trends definetely do like to start on Monday. Also, primarly starting in the earlier hours… end day typically in the evening.
Going to add some more things, as well as drill into some daily stats next. See if some obvious things might line up with them in concert.
One problem I have now is that I would like to get some more data to look at. I’m using a demoCQG feed for FX right now. I’m not getting much back minute data. Might invest in some data, or maybe just a temporary feed to grab some data.
I have quite a few thoughts on things I want to look at, but organizing it is something I will have to think further on. I.e. for example, do I want to capture aggregated data, or do I want to look at the total raw data without organizing it into subsets… So still thinking about that.
For example, I could look at the min/median/mean/max of a group of data, or just present it all out there? Just some things I will have to consider.
Some other ideas I’m thinking about… for example after the end of a day, you can clearly see what the main move is… I’m going to be gathering the OR on that day for sure… but also thinking about measuring the pullbacks that occur while that main move was being made. Then the idea is, is that data significant or useful? Also with regards to practically applying that data, are the average useful, is the min/max useful… etc… so these are just some things I am considering. A lot of things to consider… sometimes it seems easy to get lost with all the things there are to think about (I’m sure we’ve all experienced this:) ).
Primarily my focus for the moment is going to be on weekly and daily data. I may include other tfs… and naturually I am sure I will do so. BUt initially this is where my thoughts are taking me, to just try to come up with some useful things to monitor on the weekly/daily tf… and also what is actually useful, and what is not.
Initially, I’d be ok just figuring out a way to capture the main move of each day. Of course eventually the goal is to be able to capture whatever the maximum each day will allow. But I figure that will be easier once one figures out how to capture a MAIN move of each day regularly.
By that I mean “Opening Range.” But not just one, but several to consider the approach. I don’t have that modeled yet, but the idea is:
W1 = First Day of week
D1 = First 4 hours after 0 GMT
M240 = First Hour
Some other notes for the thread, I’ve got a large portion of my engine complete to capture data, last TF is in. There are still many, many more things I need to do. But here on the chart you can pretty much see 3 separate TFs captured here. I still have no idea if this engine is of any use, but I should figure that out eventually.
First pic is the last smaller TF I just finished. 2 pic is a composite of them all.
Collecting some more numbers. Still early on in this type of analysis with regards to looking how the wave, or the day is shaping up… but it looks like simply wether or not the main move of the day being up or down doesn’t seem to have much impact on the next day’s direction.
For example on this pic attached, sorted for ease of view ability… you can see the current days direction, and the two days prior. What I noticed is that the Up/Down directions are more or less even… but more importantly, if the prior two days are both up, seems to have no statistical impact on the third day being up. Same for the down.
Now with that, you’ll see on the image, that there does seem to be more of a propensity for price to move up vs. down… as there are more instances of “Up” occurring, but I would attribute this more so to price having been spending more time in an uptrend during the timespan my data was collected.
So now what I’m looking at is isolating what happened on those prior two days… i.e. HOW the waves of those days formed. Going to look at some ratios, magnitude and time… and see if I can come up with some type of way to classify this behavior. Two sessions (in this case two days) seems to be a reasonable delimiter to use for finding data.
If need be, I may build an engine that can do this on a sub-day level to have access to a larger sample size. But first have to try to find something useful and practical.
The second pic, is really zoomed out, but the point is just to pay attention to the colors, so you can see.
So far, looks like if two Up days occur, 58% of the third days were up as well.
Similarly:
2xDn = 60% of 3rd days up
up/dn = 50% of 3rd days up
dn/Up = 57% of 3rd days up
So… this is just data collected from a big aggregate of data, about 3 months worth. I don’t have more data, so I am going to do a few things. Firstly start by seeing how these percentages change as a function of time throughout my data. This is just to be able see if there is any consistency of any type there.
Next I would like to see if there are any distinguishable behaviors that exist when a day went up or down. I’m not sure which characteristics I am looking to quantify just yet, but I’ll start just by eyeballing the various outcomes on the chart.
Eventually I’d like to use a delimiter smaller than 2 days. This would likely make more sense to do IF I can find any semi-repeatable price behavior.