Page 1 of 1

A cyborg Trader? (Half Man, Half Machine)

Posted: Sun Jul 13, 2003 3:13 pm
by shakyamuni
The trained mind is flexible and very good at adapting, changing and thinking in qualitative terms. However the mind has emotion because it relies on the amygdala as well as the neocortex to function in practical operation. Therefore, the mind is inherently imperfectly disciplined.

The computer, on the other hand, is not very flexible, nor is it proficient at making qualitative determinations. The computer has perfect discipline and thus can apply rules impartially.

Posted: Sun Jul 13, 2003 7:07 pm
by ForrestGump
G'day shakyamuni,

Very thought provoking. Touches on some of the issues that have been percolating through my mind. The distinction between the biological brain and electronic "brain" is an interesting approach to describing "the trading problem".

It is partly for the reasons that you have stated that I find myself (after wandering for several years in the wilderness) drawn to mechanical trading. Personally, mechanical trading fits my mindset, however I am not arrogant enough to rubbish other methods of trading. After all, we all know that no matter how much you "proove" a mechanical system and even if you provide it for free to others, most people will not succeed with it.

Anyway, for me, I see the use of the bio brain in developing the system - this is where some judgement can be applied and decisions about a path of investigation to follow can be made, sometimes even based on "a feeling" or "a hunch". The computer has a role here purely in computation (backtesting).

Actually running the previously tested system then becomes purely mechanical (!) and is well suited, once again, to being assisted by dumb computational power.

I believe in the above scenario, for me, for a simple reason - the human brain is too smart for it's own good. There are many ways in which this can have a negative impact on trading.

- Charts. This is my big bugbear. Now that we all have a PC (or two) on our desk and easy access to charting programs and data, everyone wanders around waving extremely detailed charts. The problem as I see it is that the human brain is exceptionally good at visual skills. The human brain is geared up to make sense of chaos (think about inkblot tests). Even is there is nothing there then the human brain will find what it is looking for. Patterns in charts are particularly obvious here. I am certain that many of us have looked at a chart and (sometimes subconscously) made a decision because we see a pattern forming that is a repeat of an earlier one. This behaviour has even been formalized in the form of support lines, resistance lines, fib levels etc etc...

- Reinforcement. This may well be the worse one of the lot. Human nature is to want to be right. This is the tendancy that we have to use whatever evidence we can to be able to say that we are right on a trade that we are in - even if it has gone against us. This is tied up with ego and is why so many people know about stops but let their trades blast straight through them (I am right, the rest of the market is wrong and it will turn). Removing this ego from the trading equation is easy to say but hard to do - it is. If the ego (the need to be right) is the driving force then the brain can pretty much take any evidence available and arrange it to justify a position.

- Overcomplexity. The human brain is an extremely powerfull device and likes to demonstrate that. This is done through developing the belief that more complex is better. however, just because the brain can come up with more and more complex ways of dealing with a situation does not mean that those methods are better. In my own experience I have, a number of times, followed a path of some form of TA only to have to hit myself on the head every so often to wake up to that fact that I am just adding more and more complexity to a system that either did not need it in the first place or was never going to really work, anyway. I have a friend who has not backed away from this and he has developed a system over the last couple of years. Everytime he shows me his system he has added another indicator in an attempt to either improve or correct the system. The system is now so complex that I certainly cant understand it (and I wonder if he can). I am sure that many people are surprised at exactly how simple the OTTRs were.

- Wrong Taget. This is an idea not fully formed in my head but one that I have started thinking about since becoming involved with this (excellent) forum. This comment applies to some of what I see on this forum but I am not trying to be rude - just giving my perspective. For me, the target is making money - not developing systems. I am a programmer by trade and like many here get a certain amount of enjoyment from designing and implementing computer systems. Fortunately, my work has always been in commercial environments so I have developed (mostly) the ability to ensure that the target (getting the job done) is not overshadowed by the enjoyment of meeting a challenge. (I have seen numerous computer systems that have never reached production primarily because they have been overscoped and not managed towards a practical target). I have seen sevearal refernces in the forum to "pulling the trigger". I just sometimes wonder if some potential traders get more involved in developing YABS (yet a better system) and never actually pull the trigger. Of course that may not necessarily be wrong. After all, we are on this earth for a limited time and how we "get our kicks" is different for each of us. Ie. If your taget is to get the sense of joy and happiness that you get from developing and refining a system then be sure that you know that.

I'll leave it there and hope that my thoughts are of some interest.


Peace
Fforrest




[/b]

Posted: Sun Jul 13, 2003 9:37 pm
by shakyamuni
Excellent comments!

Posted: Mon Jul 14, 2003 2:06 pm
by Forum Mgmnt
shakyamuni,

I don't make the same distinctions you do with regard to Money Management and the other aspects of trading.

As Forrest mentioned, there are many ways that the human brains' functioning does not suit traders well. There are many articles and books in behavioral finance that outline some of the common fallacies in human thinking that affect traders and markets.

I also believe that Forrest is correct that the human brain is good for building systems.

I have not found an instance where I couldn't program in a concept that my human brain thought was important. I have also found that most people who believe their discretion adds to their trading have never actually stopped and compared their actual trading results with what a purely mechanical approach might have done.

Irrespective of how desirable it may be for one to master their own psychology, it is for most people much easier said than done. So for that reason, purely mechanical systems are better for most people.

From my perspective, the greatest advantages of purely mechanical systems are that:
  • They can be tested over history.
  • A computer can do the work instead of me
I happen to believe that I can trade slightly better than a purely mechanical system. However, I hate watching the price screens, it was boredom with that aspect of trading that caused me to retire. Further, my boredom will cause my actual trading to suffer because I can't force myself to pay attention long enough and carefully enough.

I find trading research much more interesting, so I trade better in actual life by having a computer do most of the daily grunt work and by excercising my human brain in the system building part.

Posted: Mon Jul 14, 2003 6:58 pm
by Nathan
Interesting ideas. I have a few comments which I hope will not be perceived as splitting hairs. (But if you think I am let me know, haha)

shakyamuni says:

"One might also tentatively consider that we have two basic tools with which to develop our method; the mind and the computer."

I would say the situation is slightly different. The ultimate tool is man's mind. computers and other forms of technology are secondary tools that allow man to multiply his intelligence, ability, and computational power. That is why I do not think there is a conflict of "man vs. machine". Most people would not claim "a carpenter has these talents, and a nail gun has these talents." The computer/trading situation is more complex, but I think a fundamental principle may hold true for both situations.

If a trader is using a computerized system, the system is not "mindless". It is the systematic application of his intelligence/knowledge over a class of evens to which he/she believes their is an edge.

shakyamuni says:

"Since money management schemes are purely quantitative by nature, they can and should be developed prior to trading and left to the computer. Since guessing is an inherently qualitative process, it appears to me that it is better left to the mind."

I think this is surely a valid approach for many traders. However I look at it in a different way. A systematic money management scheme is usually devised as part of an overall system. As I c.f. has said in various places, the different parts of a system fit together into a unified whole, were changes in one element can have large effects on the entire system. (hopefully I summarized that accurately, if not let me know). I like to visualize it as something similar to a biological system, where the functioning of one set of subsystems is often dependant on the other sub-systems. Consequently if your entries and exits are truly discretionary, I wonder If It is possible to get the full benefits of a systematic betting scheme.

Most successful discretionary traders I have read (many) about or have met (only 1) talk about “loading up" or "pressing" their best trades. It appears that they take advantage of their unique qualitative focus by maximizing unique "windows of opportunity" that may not fit into a systematic framework, ie be repeating patterns that occur with enough frequency to historically test with any validity.

An example of this is the story Tudor Jones tells in market wizards, when he loads up on Bonds after the 1987 crash and made his biggest killing ever. He saw a unique window of opportunity and traded aggressively to maximize that opportunity. (recalling from memory, don't have the book in front of me) For this reason I am not sure that a true discretionary trader is best served by a systematic money management scheme. In fact, varying bet size may be an important part of many discretionary trader's success. The successful discretionary trader I met had his first "big break" when he loaded up on call options to take advantage of what he perceived to be a very unique opportunity. A better alternative may be to place absolute position limits that predefine maximum risk during "windows of opportunity."

For a systematic trader, I see the qualitative side being important when it comes to developing testable ideas or formulating a testable trading hypothesis. The quant side is important when validating/disproving ideas via testing, and when creating the systematic rules. That is the impression I have. I have limited experience so please take everything I say with a large grain of salt.
Nathanael

Posted: Mon Jul 14, 2003 9:49 pm
by shakyamuni
All great replies!

Posted: Thu Oct 07, 2004 9:39 pm
by MCT
Great thread :)

As part of the overall system are entries and exits justifiably quantifiable?

Can supposedly objective entry and exit methods take into account unquantifiable variables (complex interacting patterns)?

Does an attempt to quantify the taking on, and eventual exit, of risk treat people (people as a social system) as being passive, rather than having complex motivations?

Isn’t such quantification an attempt to harden, over simplify and maintain the elegance of trading system algorithms?

Do we not trade our system as part of a much larger complex system – the market?

Doesn’t information flow from this larger complex market system to our trading system?

Aren’t our entries and exits, so to speak, boundaries where these two systems merge?

Physicist and philosopher David Bohm, in his wonderful book, "Creativity," talks about what he calls artamovement (or the art of fitting in the flow). Are we not as much artistes as systematic mechanical traders? Isn’t fitting our system to this much larger and complext market-system as much art as what we call objective system analysis?

Isn’t our system context driven? Isn’t “meaningâ€

Posted: Fri Oct 06, 2006 3:42 am
by rabidric
MCT wrote: Would anyone trust a computer to land the airplane they're traveling in? There simply is no way to mechanicaly exactly measure the instantaneous rate of change. I find, takeoff and landing are an important element of risk management, and require human intervention. As complete mechanization is counter productive in flight, so is an attempt at complete mechanization of the trading process. I doubt if it is even possible.
hahaha computers land planes all the time. look into it. when it goes IFR these days practically all airliners land on automatic. when the us navy introduced it on their carrier planes years ago the %age of missed landing wires in bad weather dropped right off on those planes equipped.


here is something to think on- definitions of computer power: FLOPS and MIPS. google the definitions. humans can't even do 1 FLOPS computers can do millions. computers can do a lot of MIPS- your brain routinely does many orders more though.

by definition your brain has evolved into a design that is optomized for adaptive problem solving. ideally suited to creating new solutions. to trading problems.

computers are tailored to a narrow requirement: speedy calculation of solution to a defined problem.

dunno where i am going with this....


but in your pilot analogy where takeoff/landing is like trade execution, modern black box auto spreaders are very good at it.
i still like a pilot though too- the auto land computer is good for just that- it won't be able to do anything other than it is programmed for. so when an engine blows up and you are leaking hydrualic fluid and fuel at an unspecified rate, i would also prefer to put myself in the hands of a pilot.

likewise when i anticipate wierd new shit going on in markets behaviour, or my systems are spending to long in a drawdown for my liking, i will look into a different system variation to adapt.

preprogrammed methods can be better than a human at specified tasks with boundaries. the human mind, being an adaptive generalist, is better to have in the driving seat when boundaries are crossed.

Posted: Thu Nov 30, 2006 10:00 am
by painless
I think something is missing in this topic.

I believe the more understanding that you have of the problem the more intuitive CAN be your solutions. Lets say you knock yourself out testing and retesting and you are always thwarted by the same problem. So you seek an answer that is not readily available. That search creates tension in you and then one day you wake up and say "I know! If I add this to that then eureka". You have given the mind enough information to find what it seeks and this is the real process of discovery. At that moment you have the answer to a question that previously you could not define. Initially you could define the result that you wanted but not the question that would yield that answer. Invariably the answer that you come up with is quite simple.

The post about the friend who keeps adding more and more complexity is a case study in the opposite. Applying intellect to a problem that has not been accurately defined would be my take on that. True genius is posing the question in the first place not finding the answer. Einstein went where no one had been by asking the questions that had not been asked. My advice to your friend, for what it is worth, would be to define in the most concise way, the problem that he is trying to overcome.

I see no peace in seeking some ideal statistal result. The market changes all the time so we cannot seek consistancy where it does not exist. CAGR and many other stats imply a uniform return where none exists. If one is to seek a reasonable level of success then it must be in context of what is available to be had. Thus I tend to treat some of these statistics with a grain of salt. Perhaps my previous posts about the fund managers may seem flippant but it is not my intent to belittle their performance. I cannot judge their performance without knowing what they made out of what was available. When you can say that you did badly out of a really horrible year rather than being wiped out by it then I think you are a stunning success but what statistic tells me this?

Please stat lovers don't arc up ok. You are the guys who have the wherewithal to create your own measurements to suit your/our own requirements. I merely make the point that existing stat functions don't seem suited to what I seek, which is making the most of whatever get thrown at me.

There is beauty in simplicity and if one starts measuring their work in terms of return for effort than there is a point of dimishing returns. I get the feeling from reading these posts that there are many here who have done the hard yards, have made the choices and have that deeper understanding about what they have done. We all live under the veil of uncertainty but when you know in your gut that you have done all you can do and the rest is down to how you execute your plan/method then there is no more testing to be done. I would suggest that if you do not have that feeling then their your intuition is telling you that you are wrong even though your intellect is saying the opposite. You will fight and kill anyone who dares challenge you and you will attack all of those on the this forum who don't agree with your idea but come tomorrow that uneasy feeling will drive you to do some more testing. If you jump into the market that uneasy feeling will mess with your head and you will find it difficult to follow your trading system because you are not really content with it, even though you won't admit it to yourself.

My point? You don't need the super bionic brain because I think you already have it. Computers speed up the process of forming the questions by providing us with a quick way of absorbing vast quantities of information. The answer to the question is just logic.

Just my thoughts on yet another interesting topic.
Blair

Posted: Thu Nov 30, 2006 11:20 am
by sluggo
painless wrote:existing stat functions don't seem suited to what I seek, which is making the most of whatever get thrown at me.
You are in luck; a very nice man named Robert Pardo read your post and then time-travelled fourteen years back into the past, where he wrote a book that contains exactly the stat function you seek. A tip o' the hat to good old Robert. (hyperlink)
Robert Pardo wrote: ... this measure grounds system performance in the actual profit opportunities offered by the market.

Posted: Thu Nov 30, 2006 11:47 am
by painless
All my aspirations of redeeming myself for past posts are once again shot down in flames by the venerable Sluggo. I remain forever in Sluggo's debt as a most humbled, unworthy contributor. Don't anyone ever let me forget this or I might stray off the path again.

Damn! No matter what I think somebody has already done it. You forget Sluggo that I only recently crawled out from a rock so I am not wise in the ways of the world to say nothing of my ability to read past posts in spite of my numerous feeble attempts.

Now how do I calc this perfect thing in 5 easy steps? "every valley and every peak" mmmmm! that sounds messy! Is there some quick and dirty way I can whip that up?

the most unworthy grasshopper

Posted: Thu Nov 30, 2006 12:19 pm
by painless
Due to my impetuous nature I have decided to use a yearly open to high to low to close as the basis for the perfect value. Given the duration of my trades (longest about 9 months to date) that sounds plausible. What do you think Sluggo?

Posted: Thu Nov 30, 2006 1:49 pm
by sluggo
What do I think? I think Ted Annemann was correct when he said (link)
The reason why there are so many different ways of measuring Goodness of a trading system, is that so many traders want so many different things. Ask yourself why ... (sentences deleted for brevity) ... It's because no two people can agree on exactly what they want. So each trader figures out for himself exactly what he wants, then uses or invents a math formula to compute it.
In that spirit, if you have a way of calculating Perfect Profit which measures exactly what you want to measure, good on ya!! But don't expect other traders to enthusiastically adopt your procedure; after all, you didn't adopt theirs. :)

Posted: Thu Nov 30, 2006 2:09 pm
by nodoodahs
I've been struggling with a benchmark for testing trading contracts as well, and wanted to share some ideas.

We need to incorporate costs of trading in the benchmarks.

Over the trading time period, using the "catch every peak and valley," if done as a total distance traveled by summing the daily ranges, implies two commissions each day. Subtract that from the gains.

Another possible benchmark is the "buy and hold" or "short and hold" benchmark, i.e., assume you traded a contract at the start of testing, held and rolled over per your algorithms, and pyramided your gains as (if!) you gained equity. What would that equity curve look like? For some commodities, it might have been better to have been short since 1995 … use that as a benchmark.

Bottom line, the comparison I would like to use is the equity curve of a dividend-reinvested index fund, in terms of CAGR, drawdown, etc. The difficulty here is accounting for margin, since that benchmark implies no margin! So to test properly, one needs to assume no margin in testing the futures, or compare versus a more pedestrian investment with equivalent margin.

For example, div-reinvested SPY is 11.5% CAGR with 48% DD over a 3-year period. At 10:1 margin, I think it's 32% CAGR with 64% DD over a 3-year period.

Posted: Thu Nov 30, 2006 6:16 pm
by GENX
c.f. wrote"
I happen to believe that I can trade slightly better than a purely mechanical system. However, I hate watching the price screens, it was boredom with that aspect of trading that caused me to retire.
I couldn't agree more c.f.!

Except I don't know if I could do better. Perhaps trading off a floor somewhere...... :twisted: I have however, on occasion watched a couple of set up's manifest and I will deviate from a mechanical stance and do a discretionary swing trade or some such trade of the like.

I seem to recall Bill Eckhardt mentioned in New Market Wizards that he thought he could do better with his discretionary so he set up 2 accounts and traded one mechanically and the other with his discretion. What he found was that his mechanical actually outperformed his discretionary to his surprise. It would be interesting to see how I could do, no doubt some people have an inherent gift. But I wonder how much of it is training + experience and developing an innate feel for a probable outcome.

Posted: Fri Dec 01, 2006 10:35 am
by painless
"after all you didn't adopt theirs"

Ok Sluggo! I'll bite. I assume you mean cagr which is what I referred to. This is because unlike others I do not have 100000 starting capital so cagr is dreamtime stuff. I have to live in the real world where I am starting from next to nothing and thus testing must emulate my world.

But to make you happy I did a cagr on SPI (Aus ASX 200 index) which is what I have been trading. Assuming a starting capital of 10000 and margin/equity of 24% that = cagr of 171% over 5 years. Maybe I made a mistake but that is the formula I found at
http://www.investopedia.com/terms/c/cagr.asp

I have not deducted the 2/20 because from memory (since the iasg website appears to be updating now) that figure will leave most, if not all, entries over 5 years on the iasg website behind regardless of the 2/20 rule.

I have not previously made comparisons to anyone else other than to say that I am content with what I am doing and have done. I have not criticised others nor is it my intent to do so. The tips you have given me have been greatly appreciated. I do not expect anyone to adopt anything I suggest nor do I care. Free discussion should promote different points of view.

OpenOffice spreadheet file attached

Posted: Fri Dec 01, 2006 10:41 am
by painless
spreadsheet attached here

Posted: Fri Dec 01, 2006 8:26 pm
by painless
That should be 90% cagr but my point remains the same and I have not added contracts to the positions if I did then it would be higher still of course.

corrected spreadsheet attached

Posted: Sat Dec 02, 2006 6:46 pm
by BARLI
GENX wrote:I seem to recall Bill Eckhardt mentioned in New Market Wizards that he thought he could do better with his discretionary so he set up 2 accounts and traded one mechanically and the other with his discretion. What he found was that his mechanical actually outperformed his discretionary to his surprise. It would be interesting to see how I could do, no doubt some people have an inherent gift. But I wonder how much of it is training + experience and developing an innate feel for a probable outcome.
Looks like we're better off not arguing with our systems :wink: