Advanced Structured Forex Trading

[7thSignalTrader]

...The question for me is how best to derive the probabilities...

Bingo! Yep - that's the key in what I do. I collect data just like any researcher would, no different. My data is extracted from the real market data. So, I go one step further by creating the metadata. So, for example: I might take a measure between today's high and yesterday's low. Ok, that is now called a "Dimension". So, I now have at lest one dimension in which to do further exploration.

See, the "theory" came from the "idea" that there might be something worth me spending my time studying between today's high and yesterday's low. Then my job is to find out whether or not there is any "there", there. Is there any repeatable "substance" there. The base math on that one was very simple: (Today's High - Yesterday's Low). That becomes the baseline math. Now, I could either find enough empirical evidence to go ahead and build an Indicator out of that for a certain number of bars, or I can make the decision to make that baseline extensible by wrapping it with either more math from more dimensions and/or logical statements that connect its Input or Output to yet another Indicator and or Signal.

The mathematics becomes more sophisticated when you begin to link one Indicator to another to form a Signal. But, at the core, the root math is pretty simple. Things only get more complex when you reach the point of having to write systems integration logic. That is when things can get real interesting, then frustrating, then apparently useless and then all of a sudden awe inspiring because you hung in there long enough to “cracked the code” that was cracking YOU – and now you have something that is simply beautiful to behold – because you were patient. The combination and permutations are endless. That is why building any non-conventional trading system from pure ground-up research is so exciting, time consuming, complex, demanding and taxing on your creative ability.

You have to create that which never before existed. And, that is the hard part, but it is also the rewarding part once your ideas begin to "gel" into something tangible.

Most good raw OHLC patterns did not just jump out at me. Most of the really good ideas took years to really fully understand and apply correctly. They were there all the time, but my eye had not yet been trained enough to notice them at work. Other people stare at charts all day long. I like to look at raw data. A huge difference of view point, but we are all looking at the exact same data.

If you could ride a photon of light at 286,282 miles per second and are being observed by your brother here on earth passing by – you will see things and observe things that your brother simply cannot see and observe. Likewise, your brother will see and observe things you can’t possibly observe and/or see – yet both of you will be sharing the exact same “relative” space. Why? Because you both have two different frames of reference about the exact same event.

So, I can see TCD’s simply by looking at a chart or raw OHLC data, but unless I tell someone what a TCD is, they won’t be able to see it – or, they might see it in their own way (without having a name for it) but have no idea what it means or how significant they are to the Day Trader even though they are absolutely indispensable to all Day Traders on the planet! Why? Because it is logically and physically impossible to trade against the dominant “real-time” TCD and not lose money 100% of the time! Yet, most people don’t even see them – or know that they even exist.

Your probability calculations/results are going to be no better than the underlying "base" of components that you are analyzing. So, Today's high minus Yesterday's low will produce a delta value. That delta value is metadata. That metadata comes from the market data, but it is not market data in and of itself. It is "data" about "data". That's what meta means. Meta is: Instead of; representing; in lieu of; replacing the original; substitution for; etc. This system is now 100% meta driven, aside from the actual real-time market OHLC inputs - everything else is all meta.

There are many dimensions to explore in just the OHLC category. I call these Primary Dimensions because they use raw/actual market data to generate a value, typically a delta of some sort. But, once I get to the Metadata, the number of Dimensions becomes infinite and the exploration possibilities are endless. All you need is to locate your set of Dimensions at the Meta or Primary level that work for you. I don't need to explore ever single Dimension before I find enough to build a solid trading system. If I want to make the system more sophisticated and more precise, then I can decide to explore more Meta Dimensions and do more integration work, which is precisely what I have decided to do with the new Alpha-5P Dimensions.

Alpha-5P will consist of previous Metadata (Alpha-4 ), new Metadata and just five (5) pieces of real market data with two of them being historical market data points and three of them being real-time market data points (the current Open, current High, current Low, previous High and previous Low). The Alpha-5P's will be calculated using Metadata alone consisting of forty four (44) meta components (numeric Inputs) all calculated from Alpha-4. It is simply taking a more in-depth look at the Alpha-4 output and creating a whole new system module from a new/different vantage point.

Again, once you get to the meta level after having created enough solid Primary Dimensions form baseline OHLC inputs, the possibilities for finding new Indicators and new Signals becomes endless.

It is a brave new "unconventional" world out there! The really cool part is that it is all Open Source Code! The entire world has access to raw OHLC data and the ability to see the same (or better) patterns that I do. It does not get any better than that, IMO.

 

[7thSignalTrader]

Yes they are mine, still at very early stage and too rigid

SD

The will find their sweet spot - keep cracking the code!

 

[7thSignalTrader]

Feb 15-16 Dashboard Panel Update at 11:12pm East, 8:12pm Pac: (pic below)

About 4hrs and 12min after the Open of this session and not much change in the Pane, yet.

 

[jiacopel]

How far back do you go with your data? Are you only analyzing today's HLOC in relation to yesterday's HLOC?

Also, you mentioned the difference between today's High and yesterday's Low as being one dimension. I'm assuming there would be 15 other dimensions as well: today's High and yesterday's High, Open, and Close. Today's Low and yesterday's High, Low, Open, and Close. Today's Open and yesterdays High, Low, Open, and Close. Today's Close and yesterday's High, Low, Open, and Close. Are these the other 15 pieces of data that you study as well as today's High and yesterday's Low?

Do you analyze other time frames as well like 4hr, 8hr, etc?

TIA.

Bingo! Yep - that's the key in what I do. I collect data just like any researcher would, no different. My data is extracted from the real market data. So, I go one step further by creating the metadata. So, for example: I might take a measure between today's high and yesterday's low. Ok, that is now called a "Dimension". So, I now have at lest one dimension in which to do further exploration.

See, the "theory" came from the "idea" that there might be something worth me spending my time studying between today's high and yesterday's low. Then my job is to find out whether or not there is any "there", there. Is there any repeatable "substance" there. The base math on that one was very simple: (Today's High - Yesterday's Low). That becomes the baseline math. Now, I could either find enough empirical evidence to go ahead and build an Indicator out of that for a certain number of bars, or I can make the decision to make that baseline extensible by wrapping it with either more math from more dimensions and/or logical statements that connect its Input or Output to yet another Indicator and or Signal.

 

[7thSignalTrader]

How far back do you go with your data? Are you only analyzing today's HLOC in relation to yesterday's HLOC?

Also, you mentioned the difference between today's High and yesterday's Low as being one dimension. I'm assuming there would be 15 other dimensions as well: today's High and yesterday's High, Open, and Close. Today's Low and yesterday's High, Low, Open, and Close. Today's Open and yesterdays High, Low, Open, and Close. Today's Close and yesterday's High, Low, Open, and Close. Are these the other 15 pieces of data that you study as well as today's High and yesterday's Low?

Do you analyze other time frames as well like 4hr, 8hr, etc?

TIA.

Yes to all points of connection, very good. I cover daily, week and monthly data and I store between 21 to 24 rows of data for each. But, there are many more points of connection than that. For example: I may take a measurement between row 1 (today's data - realtime) and row 2 (yesterday's data) and then compare that to the same measurements between row 2 and row 3, or a partial measurement between row 2 and row 3, or an aggregated measure between row 1 to rows 3 through 24, or any myriad of combinations in-between until I locate repeatable patterns. Then I back test them and forward test them and build an indicator and/or signal from the best in class output.

These types of delta connections (no pun) are but one type of Primary Dimension, there are many others in the system but these would keep one busy for a year at least doing full time research. Then there are Cross Dimensions, where I perform similar analysis against a range of OHLC points between time-frames (between day, week and month), always searching for synergy between each time-frame. Once you get a good set of Primaries and Cross Primaries, then you can start to put together some really good metadata that leads to a predictive model, or at least provides fuel for a predictive model system.

It is the metadata that points you in the direction of where things "should" go "next" with the pair. This system is balanced between a combination of Trend components and Predictive components. The trick is in finding the right balance between the two and knowing when to allow the system to auto-adjust into a more Trend posture, or when to auto-adjust into a more Predictive posture. If done right, the system will give the appearance that it is "thinking" its way through trade scenarios each day. In reality, it is to some extent. The system should roll with the natural flow of the market, if done correctly.

That's the hat trick - getting the system to flow with the natural movements of the market using a combination of both Trend and Predictive inputs taken from the Primary Dimensional Data, Cross Dimensional Data and of course the Metadata. All three types of data make up the core input for this system's primary Signal Engine. The Engine itself is comprised of several layers, but that's a bit beyond the point of this discussion and it really is an architectural/design issue - it can be designed anyway you want. I decided to have several Engines focusing on special tasks and then I integrated all engines into one Signal Engine that generates decision support output for the Dashboard Panel.

 

[7thSignalTrader]

Calling it a night. I'll try to wrap up the rest of the new concept coding before Monday, but not sure if I'll make it by then.

It looks like the IM might have been 2 pips off this time - will have to wait until morning to find out. Small adjustments like that are far easier to correct than big ones. Looks like the longs might be on their way now. We'll see. Be back tomorrow.

P.S. I'm going to clone the current Day Trade by using the existing Long trade from $1.1888 on last session by using this Day Trade's Stop and Limit prices.

So, my previous Long is not set to Stop: $1.1844 and Limit: $1.1931 per the last Panel update. So, this effectively makes this Day Trade's Entry off by about 19 pips. In other words, if the IM misses its Limit on this session, the effective Day Trade Entry would have been 19 pips above the Panel's display value of $1.1869.

 

[chuck5803]

Yes to all points of connection, very good. I cover daily, week and monthly data and I store between 21 to 24 rows of data for each. But, there are many more points of connection than that. For example: I may take a measurement between row 1 (today's data - realtime) and row 2 (yesterday's data) and then compare that to the same measurements between row 2 and row 3, or a partial measurement between row 2 and row 3, or an aggregated measure between row 1 to rows 3 through 24, or any myriad of combinations in-between until I locate repeatable patterns. Then I back test them and forward test them and build an indicator and/or signal from the best in class output.

7th,

With the power of computers nowadays, such as your new super fast PC we have ample processing power, so would your TCD/pattern analysis be more precise if you were to use much more than 21-24 rows of data? In other words, why do you use "only" 21-24 rows?

On another subject, I have decided to start programming my own trading system from scratch. Since you already have experience regarding this matter, if you were to start all over knowing what you know now, would you still use EXCEL or maybe some other program? If so, which one.

Thanks,

Chuck

 

[Punch]

Awesome, Punch! Thank you very much. I find both your Traditional and Custom levels to be rather interesting thus far. Now, do you ever blend these two together and then actually use them in your trading?

I noticed that your actual Pivot turned out to be my standard 10 pip Entry Error at $1.1903! How ironic that turned out to be.

Also, do you have a mechanism for making these dynamic - so that they change with the moving market in real-time. Or, are these fixed for the 24 hour period? I would assume fixed, but I really don't like assuming many things.

Your Support levels were below my Day Trade signal Stop and I can't help but wonder what would have happened had their been no BenSpeak today?

Thanks and keep it coming!

This comparison has been interesting so far and I'm starting to look at additional possibilities. 7ST, at this time, I can't say that I 'blend' the three sets of levels together, but I do take note of where levels from the three different models overlap, and I weigh any overlapping levels extra heavily. Blending is an interesting idea that I will explore. My levels are set at midnight eastern and remain static for the next 24 hours. I suppose I could recalculate them on a "rolling" prior 24 hour basis...hmm....

Here are the traditional and custom levels for today. Sorry guys, I got tired last night and was in bed before midnight Eastern time. For EUR/USD:

Traditional levels:
Resistance: 1.1985 and 1.1934
Pivot point: 1.1901
Support: 1.1850 and 1.1817

Custom levels
Resistance: 1.1930
Support: 1.1838

Predicted:
High: 1.1915
Low: 1.1807

As I mentioned earlier, these three sets of levels are each calculated with a completely different method. Therefore, they can and sometimes do conflict.

 

[7thSignalTrader]

Feb 15-16 Dashboard Panel Update at 1:30pm east, 10:30am pac: (see pic below)

 

[7thSignalTrader]

Brief Analysis for Week of Feb 12th, 2006

Impact Analysis Report: No. 00-7-123-V6.0

Begin:

Now, I think (a light bulb moment) I understand better the system’s insistence on not turning “On” the Swing Trade signal. Outlook has been Short the entire time an “On” from Feb 1st, 2006. It has now increased its Capture to 297 pips and is now targeting 309 Expected. The Outlook has had a two week head start on this week’s Swing Trade signal and this week’s Swing has been locked down in the “Off” position with a real-time Draw Down of 23 pips (at 1:30pm east Feb 16) and a Max Draw Down for the week of 51 pips – both well within specs for “Swing Trade Draw Downs”. So, the Swing seems to be well balanced and even a bit smarter than it was before the changes of last month.

However, though the system is making the right small tick adjustments in the Limits an Stops for the Initial Move and the Day Trade signals, I’m not seeing the follow-through to the upside on any of these Day Trade profiles. I think there are several reasons for this:

1) The system Bias has been predominantly Short for the entire week, which is a good indication that the market this week is still in a Bearish Flat condition. The Bias light has only flickered to the Long side a couple of times this week, but nothing sustained Long on the Bias Indicator light.

2) The Channel Indicator has likewise been in a Short configuration all week long with “Inside Channel Short” being the dominant indication all week. Looking back at the very first “Inside Channel Short” Indicator on Sunday of this week and then forward through today, the Channel Indicator has been dead-on precise with its message to the Panel.

3) The system’s insistence on suppressing the Long Swing Trade signal and not allowing an Entry into that position. (beautiful)

So, I think details to me some of the causality behind the week Day Trade signal performance this week.

Beyond that, I may have found a second Trigger for the Blade/Apex trade profiles. Seeing for the first time, the Swing Trade get suppressed this much and for this long during the course of a week (it typically goes on and stays on all week long), I think that what it is truly indicating is a Swing Mixed Signal. I think the combination of the Swing Arrow being active (meaning it shows on screen) AND the Trade Status Indicator being “Off”, is the system’s way of basically saying to me:

Hey, look trader – I’d like to go Long now but I can’t given the heavy Short side inputs that I’m still getting from the signal cluster – so I’ll show you what I want to do, but I won’t allow a trigger at this time – please standby.”

I think, clearly, this is what the Swing Trade signal is telling me.

That being the case, it would open the door for an additional Trigger sequence on the Blade/Apex trade profiles. Right now, the Blade/Apex profiles stay in the “Off” position until the Day Trade goes “solo” – meaning the Swing and Outlook match each other and the Day is attempting to run in the opposite direction - like Swing Short, Outlook Short and Day Long – that would be a “solo” Day Trade signal. The only time I allow the Blade/Apex trade profile to turn “On”, is when the Day is solo. However, I think that a Swing Mixed Signal (which is basically what a Swing in the “Off” condition really is) could provide the causality for releasing the Blades/Apex profiles and turning them “On”. This would of course turn the Initial Move and the Day Trade signals, “Off” and only allow the advanced trade profiles to be active for the session.

Had that been done from Sunday through today, the total net pip count for the week would be higher than it is right now even in this slightly sideways, news driven week that we’ve had thus far. I would consider that a total victory. This week has been the perfect storm Blade/Apex set-up given the Swing Trade in the “Of” position which I now believe to be a genuine “Mixed” signal indication.

It is logical – it makes sense – I’m going to code it with conviction and then test it.

The Blade/Apex profiles don’t care which direction the currency pair decides on – as long as it moves into the Apex zone (the in-the-money Blade cut-off point) and then regress back to mean – which is exactly what a CHANNEL is all about! And, from the above I already know that the system has been “Inside Channel Short” with “Short Bias” all week long. So, the Blade/Apex light should be “On” – not “Off”.

So, the adjustment here will be in the additional interpretation of ”when” the Swing is “Mixed” and then using that in lieu of a “solo” Day Trade Signal in order to get the Blade/Apex lights “On” and thus collect more pips throughout a bearish or bullish “channeling” week such as this.

End:


Newbies: This is the kind of development analysis you will need to get good at doing. I have two types of reports that I store in a Reports Database. They are known in some circles as Impact Analysis Reporting. I have reports going back for years and I run key word searches on them whenever I see system anomalies or when there is a failed trade to see if there is some correlation between the “event” and some older component of the system that has been captured inside a report at some time in the past. This will help you zero in on any “problem child” or “maverick” inside your system so that you can make the intelligent decision to withdraw the code, change the code, or keep the code and live with its anomalistic tendencies.

This report will become part of my Impact Analysis database, so that I will be able to track the “casual nature” of the changes being made in the future. For newbies, the way you go about developing your system and the way you track changes to your system is just as important as any successful trade you make. Knowing what changes were made, why they were made, when they were made and the Impact those changes had on the system could actually make you money, or be the cause for you not losing money in the future.

So, doing proper Impact Analysis reporting can be crucial to your long term success. The data you collect in these reports can be just as valuable as the data you use in your trading system. It is minutia, but it is the kind of minutia that can pay huge dividends down the road

 

[Brabed]

7th,

All of that analysis you do on your system must take up a good portion of your time. That, to me, is what being a full time trader is all about; not only trading, but building/maintaining/improving trading systems as well. This is where I want to be someday. A lot of people who try to become full time traders are trying to get by on trading alone. A lot of these people fail. Doing nothing but trading becomes pointless unless a trader is trying to continually improve himself/herself. I think this is a point that a lot of would-be traders don’t get and why the failure rate is so high.

The US military does not just fight wars. The US military spends a great portion of their time and money building tools to fight wars and building tools to protect our country. If the US military concentrated solely on fighting wars, we’d still be fighting with musket rifles and cannonballs. Think about it, and this is the way that traders should think of it too. Anyway, I’m getting way off topic here.

The Alpha 5P concept sounds interesting. I assume it will perform the same function of the Alpha 4, which is to basically predict the current trajectory termination point, but in a much improved manner??


BRABED

 

[7thSignalTrader]

With the power of computers nowadays, such as your new super fast PC we have ample processing power, so would your TCD/pattern analysis be more precise if you were to use much more than 21-24 rows of data? In other words, why do you use "only" 21-24 rows?

Very good question, Chuck.

Yes - and No. Yes, I can get more precise by using more rows of data, but at some point it will end up costing me on the Magnitude side of the equation. In my research, it just so happens that between 21-24 rows of data, I can capture just the right amount of precision coupled to the right amount of Magnitude.

Day Trading (for me) is about four (5) mission critical things:

Timing
Direction
Magnitude
Probability

I’ve been online for years demonstrating this and using it in my trading, but I have not ever seen anybody online, in a book, or in a trading seminar talk about these four (4) 500lb Gorillas and the synergy they represent to anywhere near the level that I rely upon it inside my trading system. The entire trading system can be broken down into these four (4) major components. There are several other components as well, but these are the big four (4) monsters. Get just one of them wrong, and it could spell disaster for your trade. Yet, I’ve never encountered anybody able to clearly articulate why they are so important to the Day Trader, or why utilizing their collective synergy is so important to Day Trading in general. Only through research did I discover the importance of all four (4) working in synergy.

Your question deals with the Magnitude factor and to some extent the Probability factor – they are tied together. If I use more rows (the research just came out this way), I tend to see Magnitudes drop-off to the point where trading the Forex just stops being the grand thing that it is right now given my personal trade style. There is only so much Magnitude (the movement between the highest high of the day and the lowest low of the day) contained in any currency pair. The EURUSD ranges between 85 pips per day to as much as 125 pips per day. But, that is merely a measure of ATR – a conventional TA component that is very well known. What is not very well known is my TCD, or Transequential Contiguous Delta, which is a cross dimensional measure of basic ATR coupled to some additional “modifiers” which I cannot discuss and have never discussed in public.

So, when I extend my row count through the filter of TCD with its special Modifiers, I start seeing the tradable Magnitudes drop-off considerably. This all goes to the type of trader I am – a Day Trader. So, I need a certain degree of reliable Magnitude that has a high enough Probability for existing in each Daily Bar. Without that Magnitude and its associated Probability (degree of occurrence) I’m playing a very dangerous game. I try to tilt the playing field in my favor – so I use the number of bar that I found as optimal to carve out the necessary Magnitude and Probability combination that I need to reach my daily revenue targets.

There are enough data pattern anomalies (this week being a great example) to provide me with enough thrills and chills. I don’t want to compound that by diluting the Magnitude and thus at the same time, the associated Probability. So, for my revenue needs the 21-24 rows of data just happened to be the sweet spot. I don’t need all 85 to 125 pips per day. But, my revenue model does call for a certain number of aggregated pips over the span of a month in order to stay on track to its capital growth projections.

So, what I’m telling you is that I built this system NOT on just merely making money – but rather, making money at a very precise rate of speed. That changes the entire dynamics of my research AND it is the engine the provides the possibility for establishing reliable geometric growth of capital – which is why I am in the trading business in the first place.

TVM – Time Value of Money, is the reason why I don’t use more rows of data. But, just saying that does not explain much, so I elaborated a bit for you.

On another subject, I have decided to start programming my own trading system from scratch. Since you already have experience regarding this matter, if you were to start all over knowing what you know now, would you still use EXCEL or maybe some other program? If so, which one.

No. I can still move much faster in Excel than I ever could in an OOP IDE (Object Oriented Programming Integrated Development Environment). In my response to another in this thread, I indicated that I could not “start out”, building a trading system. That would have been impossible because this system was built from the ground-up from a blank sheet. All I had were ideas, theories and concepts in my head. Nothing existed prior to what you see here now.

Therefore, everything had to begin first with research. No indicator existed, no signal existed, no signal clusters existed, no system architecture existed, nothing – just a blank sheet of paper. In fact, Excel was not even used at the start. At the start, I used a hand-held calculator and had my Wife run DAPD numbers on about 15 to 30 stocks per day! Every morning at 4:30am to 5:30am before the stock market opened up while I did other market research. DAPD was the very first Indicator that I ever built. It is much like ATR, but it does not have the exact same calculations. This was all done by hand and it was very painful to say the least. This was before I fully understood that using a Spreadsheet just might ease things up a bit – LOL! (boy that was funny!)

So, there was no “system” in the beginning – only ideas. Eventually those ideas turned into theories which needed mathematical constructs in order to test their viability. That’s when it became crystal clear that a Spreadsheet would be the ideal environment for doing research. An OOP IDE is a lousy tool for doing this kind of research; just ask any C++, Java, etc., developer. Unless the application was designed to be a Developers Desktop, specifically geared toward this kind of trading system development, then it makes a lousy tool for pure research. To do research, you need tools. I had to build those tools (testing environments) inside Excel.

To do that in C++ would be insanity at best and most time consuming. In Excel I don’t have to compile code to see my changes at work – it is a simple key stroke. I don’t have to create libraries, methods, function calls, etc. The speed with which change takes place in a design/build/research environment like the one necessary to product this system is extremely fast sometimes. Making those kinds of rapid changes in the direction of the design architecture in a C++ world would be devastatingly slow for me. The OOP environment would already need to be a Developers Workbench Application (a full 32bit architected program) specifically set-up with research tools that enable freedom thought and total creativity for making the system a reality. That 32bit application simple did not exist for this specific system – so Excel was the best possible alternative.

Now, things like TradeStation and MetaTrader were looked at by me, but their development environment is set-up for conventional TA. My TA language is different and is not easily built inside of something like TradeStation’s EasyLanguage. Their language structure is not robust enough to handle this systems architecture. So, I had no choice but to build the prototype in Excel. Later on, I might decide to go through the translation process to C++. But, that is only possible today, because the system now has architecture to translate. In the beginning, the architecture kept changing. In a C++ environment, continually altering the core architecture of your application is mindless work. Nobody does that, if they know what they are doing.

Unconventional trading system research and prototype system development (remember I do not use conventional TA – all of my TA is new) makes Excel a prime target. Remember, we are talking about creating that which never before existed. So, the cookie cutter development environments that already exist for the trading business are simply not able to support such an unconventional project.

Again, very good questions!

 

[7thSignalTrader]

Feb 15-16 Dashboard Panel Update at 3:13pm east, 12:13pm pac:

 

[echelon4x]

Hi 7th,

Some great posts since I was last here !

Thanks for that fascinating insight into how you approach your research - I've often wondered if you had some sort of "unusual" ability to read naked OHLC data (in the way that there are some guys around who've worked on developing bar codes such much that they can pick up a can of beans in a supermarket, read the barcode and tell you what it says! ).

It appears that you've made all your discoveries through hard slog, trial and error, looking for patterns, capitalising on good patterns, enhancing your original discoveries by looking at the output in a different way and a host of other activities which any of us could do provided that we had the commitment and dedication to the task in hand - a lot of which can be "dirty work", trawling through the the same old stuff looking for that golden nugget which will make it all worth while.

That is when things can get real interesting, then frustrating, then apparently useless and then all of a sudden awe inspiring because you hung in there long enough to “cracked the code” that was cracking YOU – and now you have something that is simply beautiful to behold – because you were patient.

It sounds as though you went to hell and back with that signal integration lol !

I've been doing a bit of background reading on signal integration as it's not something I've ever needed to use until now. From what I can see my time would be well spent brushing up on Markov Chains and Bayesian Networks, when I get to the signal integration stage (I've still a fair bit down the road to go until then...).

If you could ride a photon of light at 286,282 miles per second and are being observed by your brother here on earth passing by – you will see things and observe things that your brother simply cannot see and observe. Likewise, your brother will see and observe things you can’t possibly observe and/or see – yet both of you will be sharing the exact same “relative” space. Why? Because you both have two different frames of reference about the exact same event.

Yes, the old relativity business again - it seems to keep cropping up everywhere. I've been trying to work out how to literally see data in more than one dimension. I created a rough graphic of daily data overlayed on weekly data (see attachment) - it's slightly out of alignment because of the candle spacing, though you can still see the way the "noisy" daily data makes up the much more flowing (and seemingly so more stable) weekly data.

An talking of flowing and inspiration - a while back you mentioned being inspired by laminar flow formulae and used an equation in your system which was loosely based on one of those equations - well, I never found anything directly about laminar flow - there was a theory which was connected to air pressure over area I seem to recall - but anyway I digress.

Whilst looking up "flow" I found some references to fluid dynamics - which apparently was one of the prime theories which all those bright MIT "Quants" that we were talking about a while ago took with them into the banks to attempt to predict market movement in the early 1990's. Interestingly, I read that most of those projects were pulled - why ? Not because the systems were no good, but because the banks traders weren't used to research taking more than a few weeks ! When they saw the disciplined approach required in creating a highly accurate system - they gave up. There's probably a lot of traders who read these posts on here thinking - "Well, I'm doing ok, I'm getting 60% on average. I'd like 80%-90% accuracy, but this stuff 7th is talking about looks like far too much work for me - after all I went into trading to take it easy lol !".

Fluid dynamics looks as though it could work really well in providing formulas which could describe price movement in large time frames where everything appears to be so much more sedate and "fluid".

The change that you're making to the swing, blade and apex signals sounds like a good plan - there were a good few pips to be had last week, though very strange trading indeed - your comment was interesting about the banks selling off EUR this month causing that odd linear decline - I believe that there a good fundamental reasons why this would be the case - though I don't think about fundamentals anymore - so won't mention it further lol !

I closed my trade from last night for 20 pips - ok for a "hands off" trade I guess. Placing orders with limits is hugely useful because it forces you to think about what your expected target is and your anticipated drawdown and hence stop size required - It'll be great when I've managed to build a system which will predict all this for me.

I'm still in the spreadsheets at the moment looking for those "golden nugget" relationships in data which will lead to some high accuracy predictions. A question just out of interest - a while ago you said that there were no superstars in your system any more - all your indicators and signals worked as a team. That got me wondering if all your indicators when running individually would be expected to reach a high level of performance before inclusion in the system - say at least 70% accuracy in making predictions over a 2 year backtest.

I'm trying to judge whether to look more into merging indicators right now or to concentrate on finding better relationships in the data.

 

[7thSignalTrader]

7th,

All of that analysis you do on your system must take up a good portion of your time…

Just no way around it with a ground-up deal like this.

That, to me, is what being a full time trader is all about; not only trading, but building/maintaining/improving trading systems as well.

No doubt. If I want to get to the point (and I’m very close now) where I have a fire-and-forget trading system, the full-time work is necessary unless you are a genius – and even then, the genius would have to put in some overtime, too. With my goals (a trading system that calls trade signals every single session to better than 93% accuracy to target and almost 99% break-even ability) that will require a ton of fine-tuning and a lot of research and testing. I don’t like losing money because I missed something. I mean, that really tics me off big time. I don’t mind losing money on news, or some out of the blue anomalistic spike (because those things WILL happen), but when I lose money on a design error, that bugs me to no end because it says that I was not paying attention to something that I should have paying attention to. This of course, only applies to those who actually believe that there are repeatable patterns in the market. For those who don’t believe that (or understand that) then what I’m saying now seem like a joke.

This is where I want to be someday. A lot of people who try to become full time traders are trying to get by on trading alone. A lot of these people fail. Doing nothing but trading becomes pointless unless a trader is trying to continually improve himself/herself. I think this is a point that a lot of would-be traders don’t get and why the failure rate is so high.

Amen. You have to find the time to study, research and commit to developing skill regardless of the fact that one might be a Fundamental trader, or a Technical trader. Both have their own skill-sets that need to be developed with time. The don’t come overnight with any degree of long-term consistency. Trading without understanding what happened, why it happened and what the probability for it happening again under the same conditions, is ultimately pointless. Being able to recognize shifts in the probability for trade success given a specific trade set-up is worth more than gold to me. I think they call that, “insight”. Well, “insight” can be trained, learned, acquired over time with research and study.

The US military does not just fight wars. The US military spends a great portion of their time and money building tools to fight wars and building tools to protect our country. If the US military concentrated solely on fighting wars, we’d still be fighting with musket rifles and cannonballs. Think about it, and this is the way that traders should think of it too. Anyway, I’m getting way off topic here.

I know a little bit about fighting wars (just a tad, lol.) and the US military. We build and develop war technology to upon the premise of establishing and maintaining “in theatre” supremacy. In other words, if you find yourself in a war, you don’t want a level playing field on your hands as that only ends up on more casualties than necessary. In air combat for example (which I know a tiny bit about) the primary goal is to establish what’s called air superiority – thus the term, air superiority fighter. The purpose here is to establish and maintain control of the skies above the theatre, free from enemy strikes either air-to-air, or air-to-ground.

I think trading should take on a similar mentality. Those who don’t understand that there is a “war” going on out there for your dollar (whatever international denomination it might be), or focus too much on the “zero sum game” issue, entirely miss the point. Whether the other party wins or loses is moot – whether or not you retain your dollar and grow it, is all that matters. If I’m going to design a trading system, then I want that system to have two global functions: Offensive and Defensive. It is a mentality built into your system, not a physical presence in some country.

So, my system is on the “Offensive” (establishing air-superiority) when it is leaning towards the Predictive side of is signal structure. It is on the “Defensive” (protecting the boarders – air intercept role) when it is leaning towards the Trend side of its signal structure. The war is a 24/7 excluding one day a week. It is not a joke, nor is it a game – it is a “business” and should be treated as such. From the establishment of Corporations, LLC’s, Trusts, and other legal entities to do your trading for you so that you don’t get hit with tax bombshells and missiles in route to your target, to the way you design the trading system itself, everything should be constructed for maximum profit potential at minimal loss.

So, I guess you could use the term: Trade Superiority, to classify what I try to accomplish in the market. It is not always easy, nor do I always win the “battle” – but I do expect to win enough battles to maintain financial “freedom”. Much the same way a nation tries to maintain its sovereignty.

The Alpha 5P concept sounds interesting. I assume it will perform the same function of the Alpha 4, which is to basically predict the current trajectory termination point, but in a much improved manner??

Yep – that’s it. It has a slight delay, however. Today, I’m using it on paper because I don’t yet have it fully coded. I’m still trying to figure out the right way to integrate its code into the current Engine Update VBScript, but once I nail the idea down on paper, I’ll go into the Engine and code it. It points to the Primary Move (Day Trade signal) but that move can happen within 24-48 hours. So, it extends the current Day Trade signal by about 24 hours. It did it yesterday, with the move from the bottom near $1.1860 and it appears to be doing it again today with this delayed move Long from $1.1848 which is directly in the heart of the Long Apex Trade!

So, I like it. I mean I really, really like it. I just need to see it perform in back-testing and then forward testing, the Beta and finally through its Certification period. Then I can include it as a permanent member of the team – until then, it is just a rookie sitting on the bench where it belongs. (a high paid rookie no doubt!)

You know how these high expectation draft selections turn out sometimes – they can fizzle, or they can join your team and take you all the way to the Super Bowl. You never really know until you put them on the field.

 

[7thSignalTrader]

I'm still in the spreadsheets at the moment looking for those "golden nugget" relationships in data which will lead to some high accuracy predictions. A question just out of interest - a while ago you said that there were no superstars in your system any more - all your indicators and signals worked as a team. That got me wondering if all your indicators when running individually would be expected to reach a high level of performance before inclusion in the system - say at least 70% accuracy in making predictions over a 2 year backtest.

I'm trying to judge whether to look more into merging indicators right now or to concentrate on finding better relationships in the data.

Congrats on the 20 pip deal! Many were still on the sidelines, so picking up 20 was a big deal today.

Yes - I have components in the system that are still down in the 80% region, but they have spurts up to 100% for periods of time. I also have those which are pretty stable at 89% and they don’t vary much under any market condition. And then I have those hyper-short term elements that are virtually 99% accurate, but they don’t throw signals every single day.

So, there is a mixture of signals in the system each with its own accuracy rating and its own ability to generate signals. That’s why signal integration work is so important and getting the right balance between trend components and predictive components can take some time to tweak just the way you want it – or need it.

These charts are TCD Performance Charts. The first chart shows the blended TCD performance of the Day, Week and Monthly data. They are synchronized because the daily data always points to its corresponding weekly data, and the weekly data always points to its corresponding monthly data. The other charts show the “independent” Daily, Weekly and Monthly TCD performance in real-time.

Thick Red Lines are the Short TCD’s and the thick Green Lines are the Long TCD’s. The lighter shades of green and red are basic trend lines. Notice the scale on the right and that some TCD’s can go negative.

(One note: Each chart says “Tactical”. That’s wrong, these are not Tactical TCD’s. They are Real-Time TCD’s – I was doing another experiment about two weeks ago and never changed the chart title. Tactical TCD’s cannot go negative, only Real-Time TCD’s have routine negative values.)

The higher the TCD value (scale on the right side of each chart) the stronger that TCD. So, one can easily see just by looking at these analysis charts, where the TCD trend is currently on the EURUSD. But, that’s high school analysis! The Ph.D level analysis comes when you start to analyze the peaks and valleys that make up the Negative Mean for each chart. That is where your Probability resides for the “next” move depending on your time frame.

The blended chart covers the past 22 days, the other charts cover their respective time-frames (weeks, or months). You can see the first two charts have the most noise while the last two charts “flow” better because they are taken over larger time-periods. One can easily see where the “next” big normative move in the EURUSD must come from. It will be LONG, not Short. If you take a look at the Weekly chart, you can see its last negative TCD Long peak down near -75 pips. When you go back 16 weeks, you can see it had a negative TCD Long peak of -98 pips. After that, the TCD Long took off up to 319 pips and crossed the TCD Short on its way down to -220 pips.

The key is in knowing which TCD (daily, weekly, or monthly) is dominant or going to be dominant relative to the Entry point of your trade. Of course, that is a trade secret that I don’t talk about. But, I can show you the data (graphically) that the system uses to make the determination. The rest is writing an algorithm that uses this data to make the “next” move projection. That is behind the scenes code that I cannot show in public. But, graphically you can see it here. These are TCD’s at work.

The double helix pattern can even bee seen in the last three charts, and in the blended chart, you can see very large double helix pattern but you have to visualize it beyond the size of the chart itself – fully understanding that you are looking at a blend of daily, weekly and monthly TCD’s. So, the double helix pattern will therefore be ballooned beyond the size of this chart.

This goes inside the “trend” in a way that conventional TA is simply unable to. So, I can see things that “trend traders” simply cannot see. It is like having night vision built into your HUD on a night strike mission. You can see the enemy but the enemy has difficulty seeing you. That is why one of the system’s prototype names was: StealthTrader. TCD’s provide the stealth.

 

[7thSignalTrader]

BTW - I think your blended chart is a cool way to view the market and TCD's! I've never seen them in this kind of visual before. Very creative and I can see the TCD working! Nice.

Put the Month under the Week and take another look. Then you will have all three to look at visually. :idea:

 

[7thSignalTrader]

Feb 15-16 Dashboard Update
(5:00pm est, 2:00pm pac)

About 45 pips thus far on a pretty nominal day. As you can see on the Panel, the Day Trade signal status has turned "Inbound". This means the underlying core signal strength has increased above where it was during the start of the session. You cannot see that on the Panel, but that is why I have a Signal Status Indicator telling me whether or not the signal strength has gotten weaker (Outbound), stronger (Inbound), or remained the same (Parity).

A good sign – as I have not had much "Inbound" follow-through this week on the Day Trade signal.

Omega is still weak at 67+%. It should get up to 100% on a really good day. It measures the real-time Magnitude of the trade - so I like to see that get beyond 90%. Everything else is fairly nominal.

 

[7thSignalTrader]

Out to lunch. Back for the 0000 GMT update.

 

[Punch]

About 45 pips thus far on a pretty nominal day.

Nice trading 7ST.