Are Discretionary strategies any better than Mechanical systems ?

[Crap Buddist]

The rise o the machines.........

It's very easy just to say these things but it is very different putting it into practice. I've said it before and I will say it again, people overestimate the power of computers. Computers don't learn from experience and they do not have realizations.

Imagine the simple task of driving a car from Point A to Point B. Sounds easy enough to program a computer to do this, especially now with GPS guidance. But now think of all the things you would have to account for, things that you have gained through driving experience. Traffic jams, accidents and accident avoidance, unexpected road closures, kids running across the street, animals dashing across the street etc..etc..Think about it deeply and carefully. If you honestly have any sort of programming experience you would know that this would be a near impossible challenge.

I recall one time I was driving along a quiet suburban street when a tennis ball came out of knowhere and started bouncing across the street. The car infront of me didn't slow down, but I did. I knew that there will be a kid running after that ball who wont check to see if any cars are coming. I was right. The car infront had to skid to a halt and blew his horn

I know that driving a car is very different to trading but it highlights how easy it is for people to take what we do for granted. Especially when I read things like "that is too something that can also be accounted for in a program."

Remember computers don't see in the way humans do. 'Seeing' or pattern recognition would consume a phenomenal amount of processing power. Do you think it's just a simple matter of programming: If pattern = head and shoulders Then ?

mmm, I remember about <i dunno 10 years back, that they can have cars track the car in front, like a laser and the lead car tracks sensors in the road. So back then they could do it. Obviously the cost of putting sensors in the tarmac globally would be a bit but its about. Also these 747 jumbos flying themselves seems odd but computers have been doin it for a while although I think the psychological "letting go" of people worried about no pilot in the cockpit may have an impact.

also theres a lot of weaponary systems, automated all terrain vehicles that can scan or tell the difference between civi and potential enemy and only blow the hell out of one and not the other.....

hmm what I mean is,, The rise of the machines is happening or picking up pace a bit. A lot of the technology exists and is in use, perhaps its more a case of allowing humans to still have that personal interactive experience , that and money.....

Yes with driving there will be the odd freak event but that said plenty of knob heads plough into things because they are human and not machine like.....

Thank you riding in Jonny Cab.......

 

[new_trader]

I know very little about programming and how such a thing would work but 70 years ago a computer looked like this and now you can do everything on this......

I could tell you knew very little about programming which is NOT intended as an insult. Computers only understand numbers, two of them, 0 and 1. As processing power increases it means a computer can process more 0's and 1's at the same time. The human domain, which is analogue, must be converted to the computer domain, which is binary, or digital, before it can do anything with it. You don't need to know much about programming to imagine what it would take to convert everything you see around you into 0's and 1's in such a way that a computer can tell the difference between a TV and a refrigerator. High level languages mean you don't have to program in binary or machine code, but it doesn't make the task any easier. You can't simply scan in a picture of the two objects and say, "This is a fridge and this is a TV".

 

[blackcab]

I could tell you knew very little about programming which is NOT intended as an insult. Computers only understand numbers, two of them, 0 and 1. As processing power increases it means a computer can process more 0's and 1's at the same time. The human domain, which is analogue, must be converted to the computer domain, which is binary, or digital, before it can do anything with it. You don't need to know much about programming to imagine what it would take to convert everything you see around you into 0's and 1's in such a way that a computer can tell the difference between a TV and a refrigerator. High level languages mean you don't have to program in binary or machine code, but it doesn't make the task any easier. You can't simply scan in a picture of the two objects and say, "This is a fridge and this is a TV".

The same is true of the human brain, just with different terminology and physiology. Computers "only" understand 0 and 1 is true in the same sense that humans "only" understand low and high electrical impulses.

 

[tommog]

To be honest i think 90% of traders trade mechanically, they just dont realise it. I see mechanical trading as just a set of trading rules that you happen to have backtested. Discretionary trading is a set of rules you haven't back tested. Take an average discretionary trader. Do you buy breakout or sell resistance? This is a rule. Or do you sometimes sell resistance sometimes buy a break out, ok, what makes you decide which one you are going to do? Fundamental picture perhaps, volatility, time of day again all this can be quantified and can be made a trading rule.

Basically my point is no matter how discretionary you are it is your experience and judgement that have caused you to get into a position and that experience and judgement can be quantified in my opinion. All "hunches" and "in my experience" type situations are is your brain storing past data and patterns that you recognise and feel comfortbale with.

 

[JTrader]

To be honest i think 90% of traders trade mechanically, they just dont realise it. I see mechanical trading as just a set of trading rules

I agree.

 

[new_trader]

The same is true of the human brain, just with different terminology and physiology. Computers "only" understand 0 and 1 is true in the same sense that humans "only" understand low and high electrical impulses.

I don't know what this is meant to prove, either way it is a moot point. The question is whether you can program a computer to act in the way a discretionary trader does. This then just leads back to the original question: Are discretionary 'systems' better than mechanical systems? Do you program the computer to act in a mechanical way or in a discretionary way?

Knowing as much as I do about programming and with current available computing power I would say that programming a computer to behave in a discretionary manner would be a near impossible task. Computers act on instructions, unemotionally, unquestioningly and unerringly but they will never (at least not until there is some breakthrough in software algorithms and processing power) understand what they are doing.

 

[Tadragh1]

To be honest i think 90% of traders trade mechanically, they just dont realise it.

Exactly - isn't mechanical trading actually what most experienced traders tell newcomers to do? "Stick to the plan, be consistent in your strategy, don't let emotions play role in your judgment etc." I think there are very few strategies which could not be translated to an algorithm through mathematical description. Of course, some of such algorithms might be pretty complicated.

 

[Chorlton]

I don't know what this is meant to prove, either way it is a moot point. The question is whether you can program a computer to act in the way a discretionary trader does. This then just leads back to the original question: Are discretionary 'systems' better than mechanical systems? Do you program the computer to act in a mechanical way or in a discretionary way?

I think this depends purely on how difficult the discretionary rules are to code.

I know of 2 traders both of whom would admit to trading in a discretionary manner. I personally know them well enough to know how they trade. One uses a combination of Elliott Wave & Fibonnacci, the other trades breakouts and relies only on price & volume to confirm a trade.

Even with my limited programming experience I know I could reasonably easily code the 2nd strategy as my own strategies are based on the same kind of entry conditions and they are mechanical.

However, how one would even attempt to code a system based on the 1st strategies conditions is beyond me as each person would see a slightly different pattern while looking at the same chart!!!

However, I would argue that if a discretionary systems rules could be coded then I would suggest that for a large percentage of people it would be the preferred choice as it provides instant discipline, consistancy and removes the issue of emotion from trading.

All IMO,

Chorlton