Futures Trading Analysis

[marketsurfer]

@marketsurfer:

Do you understand what price action traders do? In all seriousness you would not keep repeating your gospel if you do understand. I have developed automated systems (which is still deployed at a CTA), and for my individual account, I trade manually — I see certain advantages trading manually which I have elaborated below. Having done both, I believe I might be in a position to say a thing or two about both. Please set aside your bias for a few minutes and read what I have to say.

First and foremost, all price action traders have a model of price action. They might not call it a “model”, but every one has one. They use this model to ‘read’ the market (‘read’ === ‘analyze’). The model is developed based on data from the market (you will agree with me that this method of developing a model is not inferior to a model developed based on fundamentals — if you do disagree with me here, then you also disagree with the idea of statistical inference, and we will have to stop this discussion here. Assuming we agree, I will continue). What a price action trader strives to achieve is an “ideal” model of price action. Unfortunately, markets never exhibit “ideal” price action behavior. So, it take a long time for a price action trader to develop this “ideal” model of price action.

So, Why is this “ideal” model of price action important? All price action trader understand that markets seldom produce “ideal” behavior. But, in order to evaluate the actions of the market, one needs a “fixed” point of reference which, in this case, is this “ideal” model of price action. Once a price action trader developed this “ideal” model, she will be in a position to determine the deviation of market behavior from her “ideal” market behavior and can trade accordingly. Please bear in mind that not all deviations from ideal behavior are tradable, and this is where “patterns” come to play. A “pattern” is nothing more than a “tradable” deviation of market behavior that the trader has identified using her “ideal” price action model.

A question arises: Does these “patterns” have any statistical merit. This question is not interesting as this can be easily backtested — back tests do not require automation; back tests can also be performed manually.

The more interesting question is why do these “patterns” have statistical merit? The answer, which was surprising to me, is that the “ideal” model of price action of successful price action traders are very similar — so, they are all looking at similar levels and wanting to do similar things around those levels.

Everything I have said above is also the process followed by a Systems Trader — have a model, identify statistically valid patterns, and trade it (although the word 'pattern' takes on a different meaning in this context).

So, what is the difference between the two? A Price Action trader believes that an inexact reproduction by the market of a [trader identified] "pattern" is source of information, and she will take this new information into account and adjust her trades accordingly. A Systems Trader, by design, decides not to use such new information, and so does not adjust her traders. This is the primary reason why you will see claims by Systems Traders that the “market conditions have changed, and so the system is no longer valid”; whereas, Price Action traders will most likely say “yes, markets have changed, yet so much remains the same”.

I, like many other Price Action traders, have come to believe that understanding Price Action provides a trader the skill set required to adapt to changing market conditions without having to change the ["ideal"] model of price action. A Systems Trader does not develop this skill set -- for a system trader, a change in market condition requires development of a new model. This, to me, is a big advantage that a Price Action trader has over a System Trader. The cost paid by a Price Action trader for securing such an advantage is forgoing full automation [although partial automation is still possible].

Now to the psychobabble part: Given that a Price Action trader is accounting for new market information (deviations to the identified pattern) as they are generated by the market, it take a lot of mental focus and conviction to execute the trade. This does not come naturally and hence the talk about needing the mental ability to execute trades.

So, instead of showing your ignorance, arrogance, and pretending to play the part of Savior, ask questions; you may learn a thing or two, which in-turn might help you in your System development.

All the best.

Regards,
Monoid.

Thank you for a well reasoned reply. Would you agree that price action trading is subjective rather than objective rule following? That there is an intuitive element that goes beyond the ability to properly test?

Would reasoned random entries (market has gone up 5 days in a row, today its time to short so shorts are entered at random times ) with strong money management yield the same or similar results as "pattern" or price action trading?

Thanks,
surf

 

[barcadia]

OK, the 2nd pricea action trade worked out.
One winner one loser

 

[monoid]

Thank you for a well reasoned reply. Would you agree that price action trading is subjective rather than objective rule following? That there is an intuitive element that goes beyond the ability to properly test?

Would reasoned random entries (market has gone up 5 days in a row, today its time to short so shorts are entered at random times ) with strong money management yield the same or similar results as "pattern" or price action trading?

Thanks,
surf

@marketsurfer:

Thank you for you kind words. I will try to answer your questions to the best of my abilities.

Answering your second question:

Reasoned or unreasoned, the expected value of random entries is (and will always be) zero. This is a basic tenant of probability theory and no money management (strong or not strong) is going to change it. So, for a system/method to be profitable, it has to have positive expectancy and hence cannot have random entries.

I like history for, I believe, we can learn a lot from it and, also it helps us not repeat the same mistakes people made before us. In that light I would encourage you to read a thread in ET (Van K. Tharp's Random Entry System), where @Mr Subliminal debated @dbphoenix on this very same topic (and Mr. Sub is correct in his assessment).

So, the answer to your second question is 'no'. Random entries (reasoned or unreasoned; with or without 'strong' money management) should not yield same or similar results as methods anchored in Price Action. The performance of a Price Action method should exhibit positive expectancy. As a corollary: If a Price Action method results in a performance similar to Random entries (i.e., expectancy is zero), then that Price Action method is ill-formed. And, this last point is where a lot of academic work fails to do justice (there is another political reason: As a PhD student or un-tenured prof., going against (or publishing work contradicting) the Chicago School (in Finance or financial economics) is committing career suicide. This is why you will see a lot of challenge to EMH coming from Physicists than economists; but we will not discuss that here!).

Now, to your first question:

You used four loaded words: 'subjective', 'objective', 'intuitive', and 'testable'. Given the baggage attached to these words in trading, I will try to be careful.

The 'objective element': The model of Price Action developed by a Price Action trader has to be objective [there cannot be any subjective element in that 'cos then the very purpose of having a model is lost]. The 'tradable' patterns (see definition of this term in my previous post) is also 'objective' 'cos these patterns are determined based on historical data. So, nothing different from a System Trader's process until here.

Given that a Price Action trader considers deviations from identified patterns as new information, and may or may not adjust her trades based on her interpretation of new information, it would seem like there is a 'subjective' element here. However, we should be a little bit more careful in drawing such conclusion and investigate the meaning of 'subjectivity' a little deeper as applied to this context. It would be 'subjective' (in the true sense of the word) if the trader uses that new information to adjust traders without reasoning about that new information within the framework of the model. However, successful price action traders don't do that. The new information generated by the market is analyzed within the framework of the trader's model of price action, and an inference is drawn. It is this inference that drives the adjustment to the trade. So, the process is still 'objective'.

The 'intuition' part of your question is very difficult to answer because intuition has an 'experience' component to it. While for a seasoned price action trader who has seen a lots of variations (just like a chess grand master) intuition will play a role, such intuition does not define or is a prerequisite for Price Action trading.

'testability': This, on the surface, seems to be the big point of contention between System Traders and Price Action Traders. I don't see it that way. Below is my reasoning.

A System Trader formulates the problem in such a way that it is conducive for programing (the methodology is based on a set of rules). However, a Price Action trader's formulation of the problem requires use of inference -- hence a need for inference engine. We all know there are no good inference engines available and it is still an active area of research in the 'strong' AI community. This makes it almost impossible for a Price Action trader's method to be 'coded up' for testing. However, this does not (and should not) prevent a Price Action trader from sharpening her inference skills by studying [historical] variation (just like chess players have done and still continue to do; Please don't bring up 'deep blue' example. It is not a inference machine but a brute force method of optimization) -- this is the only way available for Price Action trader to test and improve her skills at this present time.

Hope this answers your questions.

All the best.

Regards,
Monoid.

 

[marketsurfer]

@marketsurfer:

Thank you for you kind words. I will try to answer your questions to the best of my abilities.

Answering your second question:

Reasoned or unreasoned, the expected value of random entries is (and will always be) zero. This is a basic tenant of probability theory and no money management (strong or not strong) is going to change it. So, for a system/method to be profitable, it has to have positive expectancy and hence cannot have random entries.

I like history for, I believe, we can learn a lot from it and, also it helps us not repeat the same mistakes people made before us. In that light I would encourage you to read a thread in ET (Van K. Tharp's Random Entry System), where @Mr Subliminal debated @Dbphoenix on this very same topic (and Mr. Sub is correct in his assessment).

So, the answer to your second question is 'no'. Random entries (reasoned or unreasoned; with or without 'strong' money management) should not yield same or similar results as methods anchored in Price Action. The performance of a Price Action method should exhibit positive expectancy. As a corollary: If a Price Action method results in a performance similar to Random entries (i.e., expectancy is zero), then that Price Action method is ill-formed. And, this last point is where a lot of academic work fails to do justice (there is another political reason: As a PhD student or un-tenured prof., going against (or publishing work contradicting) the Chicago School (in Finance or financial economics) is committing career suicide. This is why you will see a lot of challenge to EMH coming from Physicists than economists; but we will not discuss that here!).

Now, to your first question:

You used four loaded words: 'subjective', 'objective', 'intuitive', and 'testable'. Given the baggage attached to these words in trading, I will try to be careful.

The 'objective element': The model of Price Action developed by a Price Action trader has to be objective [there cannot be any subjective element in that 'cos then the very purpose of having a model is lost]. The 'tradable' patterns (see definition of this term in my previous post) is also 'objective' 'cos these patterns are determined based on historical data. So, nothing different from a System Trader's process until here.

Given that a Price Action trader considers deviations from identified patterns as new information, and may or may not adjust her trades based on her interpretation of new information, it would seem like there is a 'subjective' element here. However, we should be a little bit more careful in drawing such conclusion and investigate the meaning of 'subjectivity' a little deeper as applied to this context. It would be 'subjective' (in the true sense of the word) if the trader uses that new information to adjust traders without reasoning about that new information within the framework of the model. However, successful price action traders don't do that. The new information generated by the market is analyzed within the framework of the trader's model of price action, and an inference is drawn. It is this inference that drives the adjustment to the trade. So, the process is still 'objective'.

The 'intuition' part of your question is very difficult to answer because intuition has an 'experience' component to it. While for a seasoned price action trader who has seen a lots of variations (just like a chess grand master) intuition will play a role, such intuition does not define or is a prerequisite for Price Action trading.

'testability': This, on the surface, seems to be the big point of contention between System Traders and Price Action Traders. I don't see it that way. Below is my reasoning.

A System Trader formulates the problem in such a way that it is conducive for programing (the methodology is based on a set of rules). However, a Price Action trader's formulation of the problem requires use of inference -- hence a need for inference engine. We all know there are no good inference engines available and it is still an active area of research in the 'strong' AI community. This makes it almost impossible for a Price Action trader's method to be 'coded up' for testing. However, this does not (and should not) prevent a Price Action trader from sharpening her inference skills by studying [historical] variation (just like chess players have done and still continue to do; Please don't bring up 'deep blue' example. It is not a inference machine but a brute force method of optimization) -- this is the only way available for Price Action trader to test and improve her skills at this present time.

Hope this answers your questions.

All the best.

Regards,
Monoid.

I completely disagree with 90% of your assumptions based on my knowledge and experience in the business. Just because you place a framework around a random system does not make the system any less random.


But thats what makes the market work.

Surf

 

[monoid]

I completely disagree with 90% of your assumptions based on my knowledge and experience in the business. Just because you place a framework around a random system does not make the system any less random.


But thats what makes the market work.

Surf

No harm done. Agree with you that "thats what makes market work". Now stop preaching!

Regards,
Monoid.

 

[dbphoenix]

:

I like history for, I believe, we can learn a lot from it and, also it helps us not repeat the same mistakes people made before us. In that light I would encourage you to read a thread in ET. (Van K. Tharp's Random Entry System), where @Mr Subliminal debated @dbphoenix on this very same topic (and Mr. Sub is correct in his assessment).

My God, that was thirteen years ago . Even so, I stand by what I wrote at the time. Computer simulations are not the same as RT trading given that behavior is an element in RT trading, unless one is automated. Therefore, Mr Subliminal's results are irrelevant. Magee understood this.

 

[k p]

Just because you place a framework around a random system does not make the system any less random.

Wow... this is a shocker. I don't know enough to claim superiority over saying this is blatantly wrong, but it is my belief that this is absolutely false. The minute you introduce a variable to give it some framework makes it completely non random.

If you bump into random people all day, but only stop to talk to pretty girls, hence that is your framework within the randomness of bumping into people, does it not stand to reason that your chances of marrying a pretty girl are that much higher since you only stop to talk to them?

 

[marketsurfer]

No harm done. Agree with you that "thats what makes market work". Now stop preaching!

Regards,
Moniod.

Thanks for being civil. Before we move on i need to define my term randomness. I mean randomness within the context of the market.

In addition, i find dramatic existential issues with the idea that patterns and price repeat in a non-random consistently exploitable ways. This does not mean that money cant be made trading however profits are despite the price action (prior to entry) belief rather than because of it.

The only price action that matters is the price action after you enter the trade. Remember, every trade you make, someone else thinks thr price action is saying the opposite-- when theres no forced transactions at the particular time.

Peace.

 

[k p]

How does your framework make the underlying system any less random? You only percieve it to be. There's no change in reality.

Well I'm at a loss to be able to argue this too well to be honest. First, the market isn't really random. Each tick might be given that there are all sorts of algos doing all sorts of things, but when you really look, important levels are defended over and over again, or they are rejected over and over again. So its not random to me.

But I think based on what I read monoid say, the fact that you're using trading systems, which I know nothing about, might make my arguments incomplete or even worthless as I'm not understanding systems too well.

My only point I wanted to make was that if you take something that looks random, but now start to add some structure to it, you might find you have something to go on.

Take speeding tickets. Perhaps people speeding in all sorts of cars is random, and who ever gets pulled over is just the luck of the draw. But if you take into account the color of the car, its been shown that red or yellow cars are pulled over more frequently. So if you want to get pulled over for speeding, buy a yellow or red car to increase your chances. Or, if you don't want to get pulled over, stick to black or gray which perhaps stands out less to the eye and blends in better with all the other stuff on the road. This is how I think of all of this.

 

[marketsurfer]

Monoid,

I also have issues with your belief thats its political for finance PhD to subscribe to random walk. The reason most do is thats what the evidence indicates. Should evidence be produced to show price action works as described on this site , it wouldnt be the provence of a few nameless folks on the internet.