Signal Bender

free & uncensored discussion arena for TheRumpledOne

Moderator: moderators

User avatar
forexjake80
rank: 50+ posts
rank: 50+ posts
Posts: 137
Joined: Sun Nov 15, 2009 12:20 pm
Reputation: 0
Gender: Male

Postby forexjake80 » Thu May 20, 2010 11:29 pm

ah... good your awake sb! I am struggling with the following a bit:

1. OR MORE
2. NON-CONTIGOUS DELTA

..."while a Trajectory is a Contiguous Delta of two adjoining Bars of data between two OR MORE data points. The other category is called the NON_CONTIGOUS DELTA, which relates one or more data points across multiple (disconnected or non-adjoining) Bars of data."

I do understand everything else pretty good, which makes me wonder if i should get a life :-) Could you elaborate those 2 points again?

btw, pretty sure mql4 is not oop.
It's time to kick ass and chew bubble gum. And I'm all out of gum.

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.

SmokyByers
rank: <50 posts
rank: <50 posts
Posts: 16
Joined: Wed Sep 12, 2007 5:03 pm
Reputation: 0
Location: Scottsdale
Gender: Male

Postby SmokyByers » Fri May 21, 2010 12:36 am

SignalBender wrote:Now, I'd like to push my system into the intra-day realm for increased profitability. There's nothing wrong with my current system. I've done quite well with it over the years in its various cycles of development, but I've always wanted to do something intra-day, as long as I can maintain the same level of consistency. Same system, just used in lower time-frames.



Tried it, TCD's and the gang generate a lot of noise at lower time frames. I had it all automated in C#, was able to get consistent break even, but no better. But you know your system better than I (duh) so you can likely make it work.

Smoke

SignalBender
rank: 50+ posts
rank: 50+ posts
Posts: 108
Joined: Mon May 10, 2010 9:36 pm
Reputation: 0
Gender: None specified

Postby SignalBender » Fri May 21, 2010 12:43 am

Ok, TRO - when you get back from the South, I've got more.

TCD Extrapolations

Let's make sense out of the whole TCD thing-a-ma-bob. There are actually three (3) of them:

Strategic TCD (MovingAverage)
Tactical TCD (AbsoluteValue)
Real-Time TCD (NextBarProjection)

Let's draw a picture, as this may help to visualize how it was first designed:

Image

This is the Signal Engine Core. On the left is the 21 Day InputStream from the Real-Time Database.

This part of the Signal Engine uses a maximum of 25 rows of data. You can think of the OHLC format similarly to the way you would think of a Table in a Relational Database (not exactly - but similar in concept). You are looking at the Daily Engine, thus InputStream contains is Daily bars of data. Therefore, row1:colum1 (labeled "Today") = Current Bar. Likewise, row2:column2 (labeled "2") = One Bar Ago; row3:column3 (labeled "3") = Two Bars Ago, and so on, through 25 Bars of Daily data (it takes 25 bars to output 21 values used elsewhere inside the engine). That should be clear as crystal.

The next module is called the "Enhanced TCD Synthetic Double Helix."

Inside this module, you see eight [8] indicators: Omega-D, SAC Long, SAC Short, TAC PD, TAC Long, TAC Short, RT Long and RT Short. The White, Green and Red numbers are real-time inputs used to calculate the Yellow numbers at the bottom of the module. This module uses the data in the InputStream to feed its calculations and near the bottom of the Enhanced TCD Synthetic Double Helix module, are the Indicator's output values (Yellow number). Thus, each Indicator in this particular module outputs a 21 day MovingAverage for each indicator type. When I plot the input values that make up the output values, I get a MovingAverage Line in a chart. Clear as crystal.

The format used here was the standard four (4) digit pip format. So, the Yellow number near the bottom for Omega-D (0.0131) is showing that for this particular range of data, the Omega-D MovingAverage = 131 pips on that particular date. 90 Pips for SAC Long, 136 Pips for SAC Short, 120 Pips for TAC PD, 116 Pips for TAC Long, etc.

Ok, so that's how I generate my primary TCD values. Now, let's talk about the meaning behind the number. First, the math.

Omega-D (Omega mentioned earlier), is nothing more than a straight MA of the Delta between the High - Low for X periods (in my case 21 periods), output as a MovingAverage.

The SAC Long is nothing more than the Delta between the Current High - Low of One Bar Ago for X periods, output as an Average of the MovingAverage. The SAC Short is the Delta between the High of One Bar Ago - Current Low for X periods, output as an Average of the MovingAverage.

The TAC Long, is the Absolute Value Delta between the Current High - Low of One Bar Ago for X periods, output as a straight MovingAverage and the TAC Short, is the Absolute Value Delta between the High of One Bar Ago - Current Low for X periods, output as a straight MovingAverage.

The RT Long, is the Absolute Value Delta between the Current Close - Low of 1 Bar Ago for X periods, output as a MovingAverage. And, the RT Short, is the Absolute Value Delta between the High of 1 Bar Ago - Current Close for X periods, output as a MovingAverage.

From the math you can see that SAC-TCD-Long and SAC-TCD-Short move slower than TAC-TCD-Long and TAC-TCD-Short, which in turn moves slower than RT-TCD-Long and RT-TCD-Short.

The Meaning:

Omega = Magnitude
SAC = Slow speed
TAC = Nominal speed
RT = Fast speed

Strategic, Tactical and Real-Time, help to synchronize the Timing aspect of this particular Signal Engine Core Module. So, the first thing to do, is to plot all four (4) on screen in MT4. Note here that the direction of the calculations in the Engine all produce positive numbers for the TCDs.

RT TCD is interesting because its calculation connects you to the "future" TCD as it unfolds. Remember, the concept written about under "TCD Fundamentals" earlier in the thread. The "contiguous" language in Transequential Contiguous Delta (TCD), means "connected." Thus, the TCD always connects the past, present and the future. SAC and TAC both use the Absolute Highs and Absolute Lows of the previous bar and the current bar. However, RT connects the Previous High and Previous Low to the Current Close. Well, what is the current close, if not the "Next Open." Thus, if you "know" the "next open" value AND you know something about the historicity of the TCD's themselves (their moving averages), then you can extrapolate to "some degree" into the future, or "next bar's" activity to determine what it should do in order to remain intact with its own history.

This is not intuitive at first glance, so re-read if necessary. The RT TCD is always connected to the Present AND the Future price action of the "next bar," by logical definition. So, RT TCD gives you an instant preview of what the limits or possible outer regions of the "next bar" might look like, based on the historical TCD averaging maintaining its structure and place in history. This is therefore, information the Trader can use when they look at the TAC TCD, when it is reaching its extreme historical limits. The bigger the bar, the bigger the pop - but also, the bigger the probable draw against the position before the "next bar" starts to fill-in the RT TCD expected range.

Something to think about, ponder and observe in real-time, before attempting to trade with it. I offer NO trading advice to anyone - merely some of the tools of the "trade" that I use.


Notes:

The only reason I switched the order of calculations previously, was to see what the Histogram might look like, remember. In the system, most of the time, I use a positive value and then invert that value whenever I need to using *-1 as a sort of toggle switch. So, do both. This way, we can look at the pos/neg Histogram as well as the above-zero graphs. Which is better will depend on the user.

Not to worry about TAC PD. That is nothing more than the Tactical version of Omega and I only use it for internal system checks. The other Yellow values are storage values set by the macro at engine update time, so don't worry about those either.

Next Module?

Enhanced DirectionFinder:
Image

Everything up until now has been fairly straight forward in terms of Excel to MQL. However, things start to steepen with DirectionFinder. It should be a good test of how MQL handles more complex system variables.

Enjoy the South!
Last edited by SignalBender on Fri May 21, 2010 3:03 am, edited 1 time in total.

SignalBender
rank: 50+ posts
rank: 50+ posts
Posts: 108
Joined: Mon May 10, 2010 9:36 pm
Reputation: 0
Gender: None specified

Postby SignalBender » Fri May 21, 2010 1:02 am

Re-read. I just made some important edits under "Meaning" about RT TCDs.

SignalBender
rank: 50+ posts
rank: 50+ posts
Posts: 108
Joined: Mon May 10, 2010 9:36 pm
Reputation: 0
Gender: None specified

Postby SignalBender » Fri May 21, 2010 1:42 am

forexjake80 wrote:btw, pretty sure mql4 is not oop.


If I posted elsewhere, it was a clear typo. MQL, EL, PTL, MS, etc., are all Procedural Languages. Any reference from me about OOP should have been in the context of deciding how best to continue using Excel as the Engine, or translating the Engine to something like MQL, or doing a full-blown conversion into OOP, where there are few roadblocks to integration with just about any real trading platform these days.


forexjake80 wrote:I do understand everything else pretty good, which makes me wonder if i should get a life :-) Could you elaborate those 2 points again?


Bar 1: Open High Low Close
Bar 2: Open High Low Close
Bar 3: Open High Low Close
Bar 4: Open High Low Close
Bar 5: Open High Low Close

1 Bar is Contiguous with 2 Bar
2 Bar is Contiguous with 3 Bar
3 Bar is Contiguous with 4 Bar
etc...

Any data point (OHLC) that connects 1 Bar with 2 Bar is a Contiguous connection. Any data point (OHLC) that connects 2 Bar with 3 Bar is a Contiguous connection, etc., etc., etc.

However:

1 Bar is Non-Contiguous with 3 Bar
3 Bar is Non-Contiguous with 5 Bar
etc...

Any data point (OHLC) that connects 1 Bar with 3 Bar is a Non-Contiguous connection. Any data point (OHLC) that connects 3 Bar with 5 Bar is a Non-Contiguous connection.

Any connection is valid. Meaning, if you can find a new Delta between any set of data points (OHLC) that had not yet been discovered as having trade value, more power to you, as you've just done some outstanding Delta Research. 50% of my time as a trading systems developer, is spent on new Delta discovery. Finding the Deltas that no one else sees, knows about, or cares about, is like finding diamonds in the rough. The difference between a Contiguous and Non-Contiguous Delta has everything to do with categorization, ordering of calculations and my established protocol for building the system.

When you are designing a new language or a new system based upon a new language, the earlier you identify structural differentiations between concepts and ideas, the better formed the language and/or system will be in the future, because you will have meaningful points of linguistic reference to lean on when trying to build one idea on top of another. I'm not a linguistic expert, but I learned this the hard way, when developing my system.

Thus, Contiguous and Non-Contiguous help you understand the Type of Delta calculation or pattern you are dealing with, or may discover at some point in the future. It is merely part of the system's taxonomy born out of desperate need to organize and categorize thoughts, ideas and dissimilar concepts.

Hope this helps! :)

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.

SignalBender
rank: 50+ posts
rank: 50+ posts
Posts: 108
Joined: Mon May 10, 2010 9:36 pm
Reputation: 0
Gender: None specified

Postby SignalBender » Fri May 21, 2010 2:00 am

General Question Folks:

Is it possible to plot Multiple PSAR Indicators (where the user can set individual Acceleration and Maximum values for each PSAR Indicator) on the same price chart in MT4? If so, how?

I want to start comparing some PSAR waves to some TCD waves to see what I come up with and hopefully come up with a new PSAR-TCD MetaDelta. MetaDeltas are part of the advanced components of my system that I have not talked about in public. A MetaDelta is simply the Delta between two existing Deltas (similar or dissimilar). Just like Metadata is basically Data about primary data, for you RDBMS freaks. In other words, there can be good information locked inside the Deltas themselves - hidden from plain view. Thus, you have to sometimes pry them open like so many Oysters at sea.

Thanks. :)

SignalBender
rank: 50+ posts
rank: 50+ posts
Posts: 108
Joined: Mon May 10, 2010 9:36 pm
Reputation: 0
Gender: None specified

Postby SignalBender » Fri May 21, 2010 3:59 am

SmokyByers wrote:Tried it, TCD's and the gang generate a lot of noise at lower time frames.

I had it all automated in C#, was able to get consistent break even, but no better. But you know your system better than I (duh) so you can likely make it work.


You most likely did not have it "all automated" as I've never released more than the 4th generation information about the system and even then, not all of the components were released.

The system is current in its 7th generation, thus the system name: 7thSignalTrader. Gen 4 was decent, if you used it correctly, to between 67% to 71% accuracy to a specified, system designated target. You cannot achieve good accuracy with any system unless you learn how to optimize targeting. Gen 5 through 7, pushed the system to a benchmark 90% to a system designated target with the past 2 years ranging in the 93% realm an over time, introduced the Intelligent Targeting Package Algorithms, an entire system of nothing but targeting schemes base on system performance.

The wrong targeting scheme can ruin a perfectly fine system - any system. The Intelligent Targeting Package was designed to dynamically assign Limit Levels for each trade signal, based on a host of system performance inputs. Vanilla (drone) targets were remove from the system in Gen 4 and accounted for more than 10% to 15% of failed trades! That's how important your targeting package can be when done incorrectly. 10% overall performance is a huge number in any system design, especially when you run super high cost basis trades the way I do.

Most people don't trade more than 5% of their equity on a single trade. The Intelligent Targeting Package was also integrated into the Revenue Model (Money Management System) and would dial-up cost basis trades as high as 50% total equity. That's how I grow my capital geometrically. The package "knows" when to scale-up the cost basis and when to scale-down, such that losses are not devastating or catastrophic. This helps to keep the equity curve non-linear, most of the time.

Gen 7 produces all pure Day trades - single shot - with variable targets ranging from a low of 7 pips to a high of 124 pips, coupled to a dynamically assigned Cost Basis (percent of equity used). So, a 50/50 Gen 4 performance, meant that either the targets were not optimized (which makes sense because I never discussed target algorithms in public) and/or the underlying principles were somehow missed (easily done because I never held classes on the subject). I basically put forth TCDs, Distinct Vega and Stealth Indicators, which when combined with LocBind and targets that make sense, should produce an accuracy rating of near 67+% as a baseline.

Watch those combined Fill% ratios and be patient!!! (I used to always tell people that)

Gen 5, introduced the Alpha4 and Alpha5 class of Indicators. Gen 6 enhanced and expanded the entire class of MetaIndicators and Gen 7 introduced the entire Meridian Matrix and Meridian Projections, along with the enhancement and expansion of the MetaBrain, to include the MetaFractal module, which is now the Neural Net "like" component of the Signal Engine that causes the system to query itself before throwing a trade signal.

These are a fraction of the MetaBrain Inputs and the Preliminary Trade Profile seen in upper left with associated Intelligent Targets:
Image

This is the MetaBrain in its final filtering stage:
Image


Not quite sure you had "all" of this in C-Sharp. What you had, unless you took the basic ideas and designed something of your own (which I highly encouraged people to do!) was based on some of the earliest work done to date. Good ideas and concepts that should have expanded the mind and challenged the traditional theories about trading "price."

G7, is so far beyond G4, that I could write an entire book between them.

Cheers. :)

SmokyByers
rank: <50 posts
rank: <50 posts
Posts: 16
Joined: Wed Sep 12, 2007 5:03 pm
Reputation: 0
Location: Scottsdale
Gender: Male

Postby SmokyByers » Fri May 21, 2010 4:16 am

SignalBender wrote:Not quite sure you had "all" of this in C-Sharp. What you had, unless you took the basic ideas and designed something of your own (which I highly encouraged people to do!) was based on some of the earliest work done to date. Good ideas and concepts that should have expanded the mind and challenged the traditional theories about trading "price."


Yes, quite right. I didn't mean to say I had 'all' of it. The point I was trying to make was using the basics, the TCD's and the DV behave much differently at the smaller time frames at least from my vantage point. I didn't mean to come off cocky (I've been accused of that before), just letting you know I had done some work in that area albeit with a limited system.

...and yes, a lot of my own discoveries are in there too.

SignalBender
rank: 50+ posts
rank: 50+ posts
Posts: 108
Joined: Mon May 10, 2010 9:36 pm
Reputation: 0
Gender: None specified

Possible TCD Set-Ups

Postby SignalBender » Fri May 21, 2010 5:27 am

Graphic on possible TCD Set-Ups.

Chart:

M5

Chart Zoom:

Absolute Minimum MT4 will offer

Indicators:

TCD; TCD Average overlay; PSAR

Colors:

Green = TCD-Long-Abs, Red = TCD-Short-Abs, White = TCD Averages (both), Dashed Yellow Line = TCD Nominal Activity Level, Dotted White = TCD Extreme Activity Level.

Notes:

The objective is to locate the high valued, high probability pivot. The idea is to see if the data contains valuable signature for such an event and to answer the question: Is there a correlation between the TCD activity and price behavior. Or, are these just random events with anomaly as fools gold.

The way you view the data on screen, can have just as much impact on your interpretation of it, as the data itself. Using the minimum zoom allowable, helps to reduce the "Nominal Activity" level while highlighting the "Extreme Activity" level.

The next step would be to look for independent and unique signature characteristics within the Nominal and Extreme levels that occur on or before the price pivot event itself. If you can find distinguishable levels AND unique signature characteristics (at least these to stand-out elements) that occur either on or BEFORE the actual price swing event, THEN there is the possibility for true, independent event verification.

That is the standard (protocol) that I use for researching new Signals and Triggers.

The other thing to note is whether or not there is meaningful correlation between the strength of the signature and the Magnitude of the move the follows. In this case, it was between 200 and 100 pips through two moves, but does that describe the historical range for the same Signal/Trigger? That's the kind of question you want to ask yourself when you do the research. If both strength of signature consistently matches magnitude of move over time, then that help to verify that that the signature has "uniqueness" and distinguishable characteristics that can be packaged for use within an overall Signal processing routine/schema.

That's it for today - have at it! :)

Image
Last edited by SignalBender on Fri May 21, 2010 11:26 pm, edited 14 times in total.

SignalBender
rank: 50+ posts
rank: 50+ posts
Posts: 108
Joined: Mon May 10, 2010 9:36 pm
Reputation: 0
Gender: None specified

Postby SignalBender » Fri May 21, 2010 5:39 am

Image is lousy. Let me know if you can make heads or tails of it. If not, I'll have to try something else. I'm trying to capture the entire screen - problem - its a high res screen - that's how I work.

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.


Return to “TheRumpledOne”