Monthly Archives: April 2014

More projection data

At Rel’s suggestion I took a look at what could possibly be happening after my projection level is reached. Perhaps they tend to extend by a preferred amount? This led to me study the specific data more carefully, and I’ve uncovered a lot more about the levels.

First, the extensions, which are based off of the PP base I’m using.

extensions

about 50% of moves that go beyond the projection extend only up to 2x (projection+projection). Which is not bad, but the data is showing a steady decease as the bins get larger, indicating that the results are not truly ratio driven.

Next, the actual typical PP levels. Close to 45% of the PP levels are within the 40-60 sweet spot area. There’s also a good chuck that’s 30 pips or under, and knowing that the large majority of days move more than 30 pips in one direction, this causes some issues in the projections. For example, these low thresholds will lead to the direction being hit easily, boosting the HBP/LBP hit count from the earlier post.

pp lvl

the under 30 pip projections make up close to 30% of the data as well.

Skewed direction-hit bias..

direciton hit

Clearly most 30 pip days have at least 1 projection hit, along with an increased both direction hit percentage. Also expectedly, the chance that no pivot is hit decreases as the PP level increases.

 

All in all, these are the kind of results you would expect if you were to assign a random number as your projection everyday in my opinion. This suggests to me that the levels are not dynamic, or interacting with the metadata, enough. What I’m really interested in is capturing the 40-60 pip days since they are most common, and in combining data I end up with something like 85% 1 direction hit which is good, but a near 24% probability to hit both bounds; close to 1/4 is too high for me. I’d like to see that probability drop down to under 10% if possible. It looks like I won’t be leaving this idea as early as I thought. Looks like I’ll be building my first “panel” to speed up the process and analyze different projections faster.

 

 

Advertisements

Projection Levels 1.0

My first attempt at pivots(edited-I think Projections are a better term to describe what I’m doing) turned out pretty good I think. I might actually just call it done for now and move on to the next area…

I used a relatively simple formula that I thought of while looking at charts (!) these past few days. I was taking cursory glance at the old SB waves indi and the types of bars that were occurring at ends, but I guess experience helps point the eyes at what’s interesting in the charts; I wanted some way to take the information the candles seemed to be showing and put it into the pivot formula. I’m not sure if it’s mathematically sound, but there’s no hurt in trying anything. 7th did mention something about ‘inventing your own math’.

pivots

HB/LB-Higher bound/Lower bound

Rather than conventional pivots that have multiple lines (PP, R1-3, S1-3) my pivot only has 2 lines, 1 above and 1 below price. Inputting the Pivot is easy, but I don’t particularly see a use for it at the moment.

HBP/LBP-Higher bound picky/Lower bound picky

Being “Picky” has an additional filter that future accounts for the days direction.

The top stat is stating that I can account for at least 1 bound to be hit about 87% of the time. Additionally, only 1 pivot level is hit close to 70% of the time. I was hoping for frequencies of about 90% and 80% respectively, but this seems ok. The next stat is saying that GIVEN the days direction (C>O or O>C) then price hits it’s respective pivot direction close to 80% of the time. Ex if the day is up, the upper limit will be reached 79% of the time. Not bad I think; I will think about it a little more before moving on and noting that this is something to come back to later

Post move bars

Chuggin along. Not sure how to tackle the next obstacle, and in the mean time I’ve been working on a few other things.

Magnitude. Magnitude and Omega fill are about finding and predicting ranges. Future range based on past range, but how helpful is it? Isn’t range more defined by “enough” vs “not enough”? I think the idea of range becomes a lot more useful if I start considering bounds, or projections, but the idea of tackling this seem daunting. Something I’m learning slowly is how well you have to design the thoughts (so meta haha) that lead to research. What I mean is that I have to have a better picture of what kinds of results I want to see, in order to design tests that will search for this. If I want to predict possible H/L points for the next day, how accurate do I want to be able to do so? The degree of accuracy will influence how I want construct the indicator.

 

Here’s a baby statistic that I ran on simple question. What happens after a “big” move? I’m initially defining a big move off of 2 thoughts: 1 being Relativity’s idea that 20 pips is a very manageable thing to do, and the other being 7thSignalTraders idea to not be too greedy about what you can get out of the market, and 50% is pretty safe. Therefore-40 pips is what I’m studying. It’s not very common that my results surprise me, but this one is actually pretty neat. On the left is all the data; The distribution of bar lengths. On the right is the length of the bar that occurs after a 40 pip bar. A length of at least 20 occurs well over 80% of the time. Wow. Could I simply wait for markets to actually move to cue me in on when to trade? That still leaves the monster task of knowing which direction to trade in, but as a standalone, this offers some potential timing ideas.

hmm

 

Pre/post bar Fill%

Right now I’m only working in the scope of one out of four (Magnitude/Omega). Within this scope contains multiple approaches, but also 3-6 time frames. Hourly, daily, weekly, monthly. Also available are H4 or H6 or even H8. Thus, it’s (the ideas) not as simple as Wave versions 1.1-x, because I’m working in different areas.

That being said, here’s what I’m working on atm. I remember my talks with Relativity in the early days, back when I was ready to relearn everything I knew about how trading fx worked. When we were discussing waves and the criteria to have a “complete” trading system, one of the things that was mentioned was the need to be self adapting in nature. I wondered and wondered how this could be possible or accomplished. I’m currently pondering the conclusion that a complete system using pure meta data absolutely needs to use averages.

I’m also currently in the process of figuring out if there’s a real difference between EMA and SMA. I’m only 1 or 2 tests in and while there are certainly differences, they are not noticeable just yet.

 

I took a quick detour and looked at the kinds of bars that form between mover bars on the H1 TF. My assumption was that before a trade-able range occurred (I’m currently interested in bars that are >24 pips), the bar would be in an underfill state. In other words, an underfill would occur in the Fill%, leading the market to “make up” lost Fill % and we would end up seeing a bar that would overfill. I think there is a way to make things less.. discretionary, but for now:

Fill% predictive

Interesting no? Over 60% of the bars occurring before bars that have H-L > 24 are in a fill state of 70%-120% of the average. While that doesn’t quite answer my question about underfill leading to overfill, it is providing some hot spots. I would think that the average fill is 100%, but again I am wrong (swapped on the conditional formatting, oops)

Capture

 

More or less the same results. The 70-100% fill is just very common. I did 1 last study looking at what kind of bar was likely to occur after a fill% between 70-120%, the range of interest from earlier.

Capture1

 

As it turns out, the majority is actually under the expected range of 24. Smart statistics! Correlation is not causation, and fills of 70-120% do not cause Omegas of 24+, even though Omegas of 24+ tend to have fills of 70-120%.

 

Also worth noting, bars that overfill are more likely to overfill again more than anything else, overwhelmingly. Some clear implications of this, at least to me, will be doing a lot more research.

Capture

Fill % Basic

D1 H-L. I wanted a basic test of how effective the Fill % is, now that I’ve gotten the understanding of it and understand that the term Fill % actually makes sense to me.

I was told that MA that fills are measured against is actually an EMA. I’m not sure what EMA though. I do know that SB seemed to have a thing for 20-21 periods, but I decided to give 10 a try: 2 weeks.

Left-EMA

Mid-SMA

Right-that days H-L

It can be noted that it doesn’t really make that big of a difference (between EMA vs SMA) as far as overr 100%/under 100% is concerned, but there may be a difference should I do something else with it. As mentioned before, it’s possible this can be used to actually project the next range like Saiduso was apparently doing (either with fills or some other type of TCD) but I’m not that far yet.

Image

It does look like the Fill has some bearing on predicting the next days range. Minor usefulness atm.

Pip lengths of H1 bars based on D1 ranges

Some times I just have a lot to say and other days I don’t. I only had 3 ideas between the end of December until March and this post makes 7 for April alone. Quality, not quantity, but I think the more ideas I can think of, ultimately the better.

I’m trying to approach the SB theories from multiple angles, both simple and complex. My mind naturally works in a very simple fashion; Apart from the wave crafting, most of the statistics that I’ve created are very simple to understand and create excel wise. I consider them to be statistics about the market that most traders have thought of, but have been too lazy to discover for themselves. The other part is the Rel/SB/BS side of things. Statistics about the market that most traders haven’t, and won’t think about because they’re very well hidden. I hope accumulating the small edges will end up leading me to think about tackling the secret edges. That being said, here’s one of the simple statistics (haha).

Question: Are the larger moving days in the market created from overall higher volatility hour to hour, or is it made from just an hour or two? e.g is the only difference between the 40 pip day and the 80 pip day just that 1 news related bar that pushed the daily range?

On the left side, the daily ranges. On top, the ranges of H1 bars. The data contains average counts for each of the H1 categories.

Another note I wanted to make is that the small details in execution matters! Compare 1 and 2

pip

pip

Much cleaner in the second.

Roadmap

Here’s a quick rundown of what I’m trying to do in following footsteps:

4 keys to success:

Timing

Direction

Magnitude

Probability

Perhaps the wave work I’ve done up until now can help in the probability and timing areas. If all the work I did in the past year combines to give even a slight indication/edge in that department, I’ll actually be quite relieved. I think if I have a better idea of the things I’m looking for, I can improve some the findings I’ve made in the past.

I’m setting out to create an “indicator” for each of these, and per the forum posts I read, each on a set of Tfs. All the work with the DV and things that extrapolate from that will hopefully give me some aid in the Magnitude aspect.

D1-W1-M1 Distinct Vega Fill% [timing]

D1-W1-M1 TCD Fill % [direction/projection]

D1-W1-M1 Omega Fill % [magnitude]

D1-W1-M1 LocBind [probability]

I don’t understand these much yet, but I aim to try to see how they fit eventually. I saw a lot about it posted in dailyfx; each one had a net signal, either long or short, and from the looks of it, simply trade the majority. Although I wonder, how can magnitude in itself have a long/short bias without any influence from direction? Maybe magnitude contains momentum?