How to Build an Automated Trading System

Status
Not open for further replies.
Quote from Aurum:

Hi Wayne,

Perhaps I'm misunderstanding you here - my apologies if so - but are you saying:

"If price is moving from the right trendline to the left trendline, but the volume says 'continue' then we won't reverse or exit yet..." ?

If that's the case, then it sounds like you are mixing up the right and left trendlines. In an up channel, the RTL is the bottom line, the LTL is the top line. In a down channel, the RTL is the top line, and the LTL is the bottom line.

Hoping to clear up some confusion on my part.

Best regards
-Au

You said it the same way I understand it. Hey. I'm no expert but what he said (which I mimicked) seems to fit the charts. Look and see for yourself.

JAck teaches on that site, and I see it on the charts, that sometime the market blows totally through the LTL on big volume.

It would be kind of silly to sell right at the LTL and watch all that money on the table go past you.

Worse yet if you're playing always in, you have sweat being upside down during that big move.

I like his plan better to stick with the trade as long as volume is continuing to grow and "let your profits run", then finally reverse when either the volume slows a little bit or prices reverse some.

Yes, I'm fairly sure on that point.

However, how could you judge that by looking at price alone?

I don't see how. The PV relationship is critical there.

I'm not going to sweat it right now. Please give me a few days to the trend processing perfected and then we can tackle volume.

Because I haven't tested anything I'm saying here it's just from reading Jack and eyeballing the charts.

Sincerely,
Wayne
 
Quote from greaterreturn:

You said it the same way I understand it. Hey. I'm no expert but what he said (which I mimicked) seems to fit the charts. Look and see for yourself.

Apparently the operative word "but" in the sentence is what's causing confusion. By no means am I saying that you should get out of a trade on just volume or price action alone. Simply put, I understood your sentence to mean price was moving against the trend (change) BUT volume was contrary in that it said continue.

Either way, it appears you have things in hand. Best of luck.

-Au
 
Quote from Aurum:

Hi Wayne,

Perhaps I'm misunderstanding you here - my apologies if so - but are you saying:

"If price is moving from the right trendline to the left trendline, but the volume says 'continue' then we won't reverse or exit yet..." ?

If that's the case, then it sounds like you are mixing up the right and left trendlines. In an up channel, the RTL is the bottom line, the LTL is the top line. In a down channel, the RTL is the top line, and the LTL is the bottom line.

Hoping to clear up some confusion on my part.

Best regards
-Au

In "Think" (IBM, 1957) there was a discussion on people and their way of perceiving (sensing). Vertical and horizontal are the choices. To maintain a neutral bias it is best to work in a horizontal orientation.

You can check a person's orientation by reading what they write or how they describe things.

So that is why the lines of boundaries are named the way they are. At some point more differentiation will occur and we will build three levels of price movement boundaries (envelopes that are nested).

Moving to a horizontal orientation there is a mental cohoice of looking from the future into the present or looking from the present into the future.

PA bettors look one way and we who do not bet, look the other way.

Going horizontal to be neutral is a hard step for some. Looking into the present from the future is very hard as you can see from all the betting being done.

We will see that "building" can go in either of two directions: analysis or synthesis. People communicate from these points of view too.

Knowledge and technique are paired for making money. What and why are in the locus of knowledge and the how and when relate to technique. I presume that the B. P. A. will nail the knowledge in a series of steps and the B. P. R. will nail the technique (skill levels) in a series of steps.

There is a tweak or caveat about how certainty and sufficiency connect to each other to have this happen. I am studying up on how confusion can contribute, I havn't gotten involved in that too much.

For both certainty and sufficiency I don't feel it is possible to skip steps when going from the general to the intimate. Just drilling down and just using the intimate doesn't happen to work out. The sweeps chart shows a little about this as it STOPS the drilling by using the MODE blocker (stops the drilling down when sufficiency is there to get certainty) called Continue.

Trading does not deal in opposites but instead it deals with orthogonals. End effects are called Change (which is a descriptor of Sentiment change)(directional flow change).
 
Jack,

Recently you posted the following, although I cannot find reference to it currently?:

How the algorithm begins and is applied.

We begin with the seed and expand it to include the scoping and bounding of the science and TA of the algorithm and its three trading applications.
Seven cases for price trending, Illustration 8, provides the intimate foundational seed for the binary vector based approach that, at all times, operates with certainty. Trading orderliness comes as a consequence of using a four part routine called MADA. MADA stands for Monitoring, Analysis, Decision making and timely Action. The routine is performed repeatedly and frequently throughout the day. Operators are used to determine: trends and sentiment; two parallel lines on three fractals; and price volatility. Illustration 9 shows the seven price cases that are examined by the operators shown in Illustration 8.
Similarly, the mind module is used to examine volume, the second independent variable, as shown in Illustration 10. Here operators determine the volume vector, surge or pause as exhibited by acceleration or deceleration, and pace and pace change. All are related to time rate of change of volume bar information.
Illustration 11 shows the Price volume annotation module. Three fractal levels are annotated to form shells within shells. Level III is coarse, the outer shell, level II is the middle shell and Level I is the most intimate level where the seven cases of Illustration 8 appear. Annotations are built from Level I to Level II to Level III.
Illustration 12, The Channel Formation Module, deals with establishing the geometry of the channels of the three levels. This leads to the ability to apply the P, V relationship, a Boolean expression that is a foundational principle that allows for certainty and eliminates any need for probabilistic mathematics. One additional component is determined with this mind module: the overlap relationship of channels. A Failure to Traverse (FTT) is the signal indicator of the beginning of this event (overlap).
The P, V module provides the remaining data set elements for M of MADA. Finite data sets are taken and they configure the, also finite, subsets of M that form the M finite set. M subsets are paired with elements of the A finite set in one to one relationships. This allows operators to be used to determine the market MODE Continue or Change) as expressed by the P, V relationship; internals detection; and a pace change over ride signal. On Level I the market is not always trending. At these times, internal formations are occurring. They fit into two categories of Level II price traverses of Level II channels: dominant traverses or non dominant traverses. Dominant is a term used to describe when a traverse is going with a channel trend. Non dominant describes when a traverse is going against the channel trend.
Illustrations 14 and 15 introduce how the always certain aspects of the binary vector systems afford the practitioner the ability to anticipate. FTT is important for timing the trading segments of profit taking. Illustration 14 deals with the pre and post aspects of the FTT occurrence. Because the market is cyclic under binary vector certainty based data subsets of a finite set, it is possible to anticipate based upon what must come next: inter-bar volume (Gaussian) shifts; and flaws (the acronym WWT is What Wasn’t That). Illustration 15 shows the graphic regions that are assigned to these anticipatory descriptors.
Illustrations 16 through 18 deal with the details of sequences and internal data arrays. This affords the practitioner the ability to anticipate for the next several bars and the potential price and volume value zones. This provides the complete basis of how and why market operating points (as defined by multi-dimensional arrays) migrate instead of jumping around. Market movement is always based on alternative paths being extinguished prior to the only remaining alternative becoming the path.

Are these reference illustrations publically available? If so, can you point me to their location?

Thanks for any insight.
 
Quote from jack hershey:

In "Think" (IBM, 1957) there was a discussion on people and their way of perceiving (sensing). Vertical and horizontal are the choices. To maintain a neutral bias it is best to work in a horizontal orientation.

You can check a person's orientation by reading what they write or how they describe things.

So that is why the lines of boundaries are named the way they are. At some point more differentiation will occur and we will build three levels of price movement boundaries (envelopes that are nested).


Jack, please help with some B.P.A. related questions. Terminology is critical in this process.

1. What boundaries? Are you referring to channel lines?

2. In referring to price movement boundaries you call them "envelopes that are nested" parenthetically. What are envelopes? Is that also another term for channels?

So I interpret this as you explaining why the channel lines are named left and right rather than top and bottom.

If so, fine. If not, please clarify.


Moving to a horizontal orientation there is a mental cohoice of looking from the future into the present or looking from the present into the future.

PA bettors look one way and we who do not bet, look the other way.

Which way, please? [edit: Okay, got it. You start with the future and work to the present.]


Going horizontal to be neutral is a hard step for some. Looking into the present from the future is very hard as you can see from all the betting being done.
I will attempt to break down what you say. Please correct me as needed.

By vertical, you refer to the mindset of looking at price moving as going up and down, right?

By horizontal, you refer to the idea of looking at price as moving from left to right in time.

Frankly, if that's what you mean, then I agree. It felt like an epiphany finally grasping what you write about how you can plan out several trades in advance. Is that thinking horizontally? Thinking from the future to the present?

Hopefully, I'm making headway and getting into the groove of how you perceive the market. Perception is everything!

It seems that concept relates to the pace of the market. You talk about the human nature of it operating faster and gradually slowing till lunch, then quickening again right after lunch, then decelerating till a big final burst in the afternoon due to automated risk management.

It's clear that if a person or ATS can get into the "rhythm" of the market, then the TIME to reverse can be "felt" or planned irregardless of the price level. In fact, you talk about trading with the price bars completely OFF the screen just by watching pace and volume indicators.

It's absolutely genius. Jack, THAT is the true essence of your trading skill. I think you're the first to even come close to describing the "knack" for trading that people develop with experience.

Many others can't seem to express why they can trade almost automatically at the right moment.

It will be a challenge for YOU and for ME both to reach the goal. That is, of course, for me to understand and automate your perception.

It seems however, that we need to first stick to getting the channel states sorted out, right? Is that the first step?


We will see that "building" can go in either of two directions: analysis or synthesis. People communicate from these points of view too.

Knowledge and technique are paired for making money. What and why are in the locus of knowledge and the how and when relate to technique. I presume that the B. P. A. will nail the knowledge in a series of steps and the B. P. R. will nail the technique (skill levels) in a series of steps.
Yes. Precisely! I won't bore with all the steps.

One critical step is to agree on specific terminology. It helped IMMENSELY that you have a glossary of terms on that website.

One example, in my question earlier, if envelopes and boundaries actually refer to channels and channel lines, then it can help for both of us to remain extremely consistent in the terminology. Your point about "confusion" below, demonstrates this point, I think.

Of course, if you need to use various synonyms because you feel that's necessary to convey accurate meaning. Fine. We simply need to clarify if they're exact synonyms or some technical nuance exists between meanings.

Frankly, it will be less work for everyone if we stick to the same term when we're talking about the same thing.


There is a tweak or caveat about how certainty and sufficiency connect to each other to have this happen. I am studying up on how confusion can contribute, I haven't gotten involved in that too much.
That's funny. I thought confusion was you're idea. Maybe you used a different word. Therein lies the need for both of us to make sure we use the same terminology.

Here's your bit of wisdom I was refering to:

On the site traderuniverse.com you talk about beginners needing to trade during market conditions or setups they understand. You recommend sidelining if you feel you don't understand what's happening. I think you wrote something like, "you must know what's going on at all times".

I call that situation that prompts sidelining "Confused".

For an ATS, it's necessary to handle confusion explicitly since that "gate" to trading for humans is based on intuition which, sadly, computers can't seem to must up even an ounce of that amazing stuff.

Frankly, Jack, a number of points will come up where you do things on pure intuition and experience. For you, it's like driving a car. We think of it as simple, eyes on the road, steering wheel, pedals--but the reality is that there's a myriad details our brains is processing in parallel at all times.

The real challenge will be in finding clear definition for human intuition.

Again, confusion was your idea. We just need to agree on the term to use for it. Have a better suggestion?


For both certainty and sufficiency I don't feel it is possible to skip steps when going from the general to the intimate. Just drilling down and just using the intimate doesn't happen to work out. The sweeps chart shows a little about this as it STOPS the drilling by using the MODE blocker (stops the drilling down when sufficiency is there to get certainty) called Continue.
That makes complete sense. That's the holistic aspect of you trading you discuss.

So it sounds like you agree on getting the channeling working correctly first.

Again, correct anything I misunderstand, please. That's your role--to constantly correct till I get it right.


Trading does not deal in opposites but instead it deals with orthogonals. End effects are called Change (which is a descriptor of Sentiment change)(directional flow change).

Can you please give an example of CW (conventional wisdom) of opposites and contrast it with the related orthogonal concept?

Wow, this is super exciting. I'm glad you are participating in the B.P.A.

One final note is your excellent assumption at the difference in BPA and BPR of knowledge before technique.

BPA and BPR can be done iteratively just as you discuss iterative refinement.

So we get a part of the knowledge and apply that. Then add some more, and apply that, etc.

Everything you said in your discussion applies here about needing to get a "belief" in the channels by making some profits. That reinforces and builds on the belief.

While as a developer of an ATS I won't have to deal with the human issues of live trading pressure, I do have to deal with maintaining enthusiasm for doing all this work. Any amount of financial reward, however small at first will build steam and enthusiasm to stick to it.

In most ways, you should treat me as you do any newbie, avid learner of your methods.

For example, I plan for for the ATS to generate a LOG exactly like the type of log you keep your self of it's MADA decisions. That way I can post the log and the chart each day or whenever we agree and scrutinize it using terms and methods familiar to yourself.

I'll post my chart, but eventually I want to make it look exactly as you're comfortable with seeing.

Of course, only difference is that the ATS chart gets completely generated by computer including all the annotations, lines, notes, etc.

Finally, do you feel comfortable so far with this approach I'm taking after reading that website?

Do we continue with just getting the channels correctly automated to start some small profits on a test account and go from there?

I expect to be at that point of handling channels within a week to 2 weeks at the outside.

Then we move to step two. And that will be logical as it can help resolve the flaws or confusion points of the channeling algorithm.

Sincerely,
Wayne
 
This screenshot shows a channel formed that associates with the log below. The log shows every monitoring event and analysis change in state.

In this format:

Id#:Monitor Event -> New Analysis State Time

9:NearLTL -> RtoL 9/4/2008 7:34:23 AM
10:Inside -> TRight 9/4/2008 7:34:24 AM
11:NearRTL -> LtoR 9/4/2008 7:39:40 AM
12:Inside -> TLeft 9/4/2008 7:39:40 AM
13:NearRTL -> FTT 9/4/2008 7:39:42 AM
14:Inside -> Confused 9/4/2008 7:39:43 AM

Note: For less clutter, only the LtoR, RtoL, FTT get annotated on the charts.

The annotations include the id # as in the log. That's quicker to relate them together.

POSITIVES:

This chart shows how the ATS can correctly identify RtoL and LtoR.

ISSUES:

1. It incorrectly identifies the FTT. I discussed a solution a few posts ago but got stuck fixing a bug and haven't implemented the "center" zone yet.

2. It recognizes the trend late. You can see the 1,2,3 points on the left that it used to recognize the trend. blue for lows and orange for highs.

NEXT STEPS:

1. fix the FTT by introducing the "center" zone.

2. Fix a bug in the short channels.

3. Fix the RTL to give a little room on either side of the line to be considered still "AtRTL" versus outside the channel. That problem doesn't appear on this zoomed in example but on others.

4. Move next to taping (micro channels) to identify new channels earlier.

5. Handle FTTs and BOs to transition from channel to channel.

Please post any questions or advice.

Sincerely,
Wayne


<img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=2070613" width=800 height=600>
 

Attachments

i don't know the method (and don't really care to) but here's an idea that might help. If the trade breaks the channel in your favor then you can hold until price reverses by closing below the previous bar low. If you examine the moves that run hard (assuming uptrend here) you will see that more times than not you will have higher highs and higher lows. Once that stops you either have the end of the trend or the beginning of a pullback.

In other words you can trail a sell order along the low of the previous bar. If the current bar has a low that is higher than the previous bar after it is finished printing then you move the stop up. If not, then it gets hit.

With that being said I don't necessarily think that is the place to reverse (would have to test that). If you are playing always in that's fine, but I would still have no problems going flat when price enters at extreme areas and then starting the states back at the beginning waiting for the next entry signal.

-js

Quote from greaterreturn:

You said it the same way I understand it. Hey. I'm no expert but what he said (which I mimicked) seems to fit the charts. Look and see for yourself.

JAck teaches on that site, and I see it on the charts, that sometime the market blows totally through the LTL on big volume.

It would be kind of silly to sell right at the LTL and watch all that money on the table go past you.

Worse yet if you're playing always in, you have sweat being upside down during that big move.

I like his plan better to stick with the trade as long as volume is continuing to grow and "let your profits run", then finally reverse when either the volume slows a little bit or prices reverse some.

Yes, I'm fairly sure on that point.

However, how could you judge that by looking at price alone?

I don't see how. The PV relationship is critical there.

I'm not going to sweat it right now. Please give me a few days to the trend processing perfected and then we can tackle volume.

Because I haven't tested anything I'm saying here it's just from reading Jack and eyeballing the charts.

Sincerely,
Wayne
 
Quote from ehorn:


The current state of the model extracted ~ 1.4X range today with some really silly actions though. Time to debreif :)

Hi ehorn..

If you get the time, could you talk about some of the actions from your chart? I think you mentioned end of bar actions is all the system can muster at this point.

I am a bit puzzled at some of the entries/reversals. For instance, the reversal to short on Bar 17 and 73 . Could you shed some light on the system's decision to act at these points?
 
Quote from Jander:

Hi ehorn..

If you get the time, could you talk about some of the actions from your chart? I think you mentioned end of bar actions is all the system can muster at this point.

I am a bit puzzled at some of the entries/reversals. For instance, the reversal to short on Bar 17 and 73 . Could you shed some light on the system's decision to act at these points?

With regards to entries, without having any landmarks or other Monitoring inputs (i.e. >= level 2 outpus) for analysis and decision, bar 3 is a fixed entry and decision is purely Jokari with respect to the previous bar. At the time (initial development) it was decided to go this route to get the ATS "in the game" so to speak. As for the 2 bar reversals in question, they are both showing volume surges/spikes. The key (which I am still exploring) to defining and assigning a surge/spike correctly is to NOT confuse it with pace acceleration. Hope that helps clarify what is currently happening internally for entries and the 2 bars in question.

I thought I would post todays print to show lack of maturity of the ATS at this point. This will be the last print of the ATS until I can establish a more mature implementation of Channel Formation Module and Price volume annotation module (i.e. 3 shells and associated components). These are just too critical to the foundation of the model to continue without and debriefs of the ATS results are clearly revealing this. I will be taking the prescribed approach of building up from level 1 through to level 3 (hopefully) :)

I wanted to add that working through this development and asking myself questions about how to describe the algorithm and translate it to computer language has had a wonderful side effect of enriching my personal MADA skills. Seems a bit of a paradox though. From a knowledge perpective, you need to start at the outer shells and work your way down to the "intimate"... as Jack mentions. But from an ATS perspective, it seems to work the other way...?!? Not sure on this (just thinking out loud here) :) Either way, I will continue to learn and continue to teach the computer what I have learned.


<img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=2072019" width=800 height=600>
 

Attachments

This is one of those rare threads on ET that contributes real, unique value...

I dabbled in automating SCT myself for awhile but put the project aside. If nothing else it's interesting as an academic exercise, but ehorn's results are definitely intriguing.

Keep at it, this discussion is fascinating!
 
Status
Not open for further replies.
Back
Top